An AI intelligent analysis method and system for railway power supply equipment defects

Abstract

The present invention discloses an AI intelligent analysis method and system for defects in railway power supply equipment, which relates to the technical field of railway power supply. The system arranges detection points around the pantograph through a pantograph detection module, obtains vibration data, equipment defect data and pantograph picture data in real time, and generates a vibration data set, an equipment defect data set and a pantograph picture data set after preprocessing the data by a data acquisition module. The AI ​​image analysis module constructs an AI image recognition model to identify the pantograph picture data set and outputs a pantograph defect data set. The AI ​​defect algorithm module is used to construct an AI algorithm, obtain the vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx, and calculate the pantograph defect coefficient Sdgx. The intelligent evaluation module evaluates the data according to a preset threshold and generates relevant warning information. The system improves detection efficiency and accuracy, reduces the workload and errors of manual inspections, and ensures real-time monitoring of the pantograph status and early defect warning.

Description

Technical Field

[0001] The present invention relates to the technical field of railway power supply, and in particular to an AI intelligent analysis method and system for defects in railway power supply equipment. Background Art

[0002] Railway power supply equipment is a key facility to ensure the normal operation of railway power supply and electrified railways. The pantograph is a key device for electric locomotives or EMUs to obtain power from the contact network system. Installed on the top of the locomotive or carriage, it ensures the continuity and stability of power transmission through contact with the contact network. The design and performance of the pantograph directly affect the operating efficiency and safety of the train. During the operation of the EMU, the pantograph is in continuous contact with the overhead contact network, thereby providing a stable power supply for the EMU. In this process, the pantograph will be subject to varying degrees of wear and vibration, which may cause various defects.

[0003] At present, traditional detection methods mainly rely on manual inspection and regular maintenance, which have problems such as low efficiency, high cost and poor reliability. Manual inspection usually requires the train to be stopped and specialized technicians to conduct detailed inspections of the pantograph, which not only consumes a lot of manpower and material resources, but also affects the normal operation of the train. In addition, due to the low inspection frequency, regular maintenance may not be easy to detect early defects of the pantograph in time, resulting in the defects gradually worsening during the operation of the equipment. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides an AI intelligent analysis method and system for defects in railway power supply equipment, which solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: including a pantograph detection module, a data acquisition module, an AI image analysis module, an AI defect algorithm module and an intelligent evaluation module;

[0006] The pantograph detection module is used to arrange several detection points around the pantograph and install relevant detection equipment and visual detection equipment, so that the pantograph of the train can be detected in real time and vibration data, equipment defect data and pantograph image data can be obtained;

[0007] The data acquisition module is used to pre-process the collected vibration data, equipment defect data and pantograph data respectively, and then acquire the vibration data set, the equipment defect data set and the pantograph image data set through data acquisition;

[0008] The AI ​​image analysis module is used to build an AI image recognition model, transmit the captured pantograph image data set to the AI ​​image recognition model, quantitatively identify the defects of the pantograph connector and the slider, and output the pantograph defect data set;

[0009] The AI ​​defect algorithm module is used to construct a vibration algorithm based on the vibration feature data set to obtain the vibration coefficient Zdxs, and then construct a defect algorithm based on the equipment defect data set and the pantograph defect data set to obtain the comprehensive equipment defect coefficient Sbqx, and then associate it with the vibration coefficient Zdxs to obtain the pantograph defect coefficient Sdgx;

[0010] The intelligent evaluation module is used to perform preliminary comparative evaluation on the preset vibration abnormality threshold Z and the obtained vibration coefficient Zdxs, and trigger a secondary evaluation mechanism based on the preliminary comparative evaluation result to perform comprehensive evaluation on the preset equipment defect threshold Q and the obtained comprehensive equipment defect coefficient Sbqx, and generate relevant early warning information.

[0011] Preferably, the pantograph detection module includes a vibration data detection unit, an equipment defect detection unit and a pantograph image detection unit;

[0012] The vibration data detection unit is used to install a vibration integrated sensor group around the pantograph to perform real-time detection of the vibration generated by the pantograph during the operation of the EMU to obtain vibration data. The vibration integrated sensor group includes a force sensor, an accelerometer and a laser displacement sensor;

[0013] Force sensor: measures the actual force on the pantograph;

[0014] Accelerometer: used to measure the acceleration of the system, thereby indirectly obtaining the dynamic parameters of the pantograph;

[0015] Laser displacement sensor: used to measure the displacement of the pantograph;

[0016] The equipment defect detection unit is used to install an equipment defect integrated sensor group at the pantograph firmware connection to perform real-time detection of equipment defect data of the pantograph connection firmware and the slide, and the equipment defect integrated sensor group includes a pressure sensor and a linear displacement sensor;

[0017] Pressure sensor: used to measure the force applied to the slider or other parts of the pantograph;

[0018] Linear displacement sensor: used to measure the displacement or deformation of the slide;

[0019] The pantograph image detection unit is used to install visual sensors at the firmware connection and the slide, and to collect defect images of bolts, nuts and slide surfaces at the firmware connection in real time from eight different angles.

[0020] Preferably, the data acquisition module includes a vibration data acquisition unit, an equipment defect data acquisition unit and an image acquisition unit;

[0021] The vibration data acquisition unit is used to collect vibration data detected by the vibration integrated sensor group in real time through wireless transmission, and pre-process the collected vibration data and classify and summarize them to generate a vibration data set, wherein the pre-processing method includes denoising, missing value processing, outlier detection, outlier processing and normalization processing, and the vibration data set includes a pantograph displacement distance x and an external force F;

[0022] The equipment defect data acquisition unit is used to collect the equipment defect data detected by the equipment defect integrated sensor group in real time through wireless transmission, and to classify and summarize the collected equipment defect data after preprocessing to generate an equipment defect data set, wherein the preprocessing method includes denoising, missing value processing, outlier detection, outlier processing and normalization processing, and the equipment defect data set includes the external force HF of the sliding plate and the deformation HX of the sliding plate;

[0023] The image acquisition unit is used to classify the defect images of bolts, nuts and slide surfaces collected at the connection of the acquisition firmware, archive the photos of the bolts and nuts to generate connection image data, archive the photos of the slide separately to generate slide image data, and then summarize the connection image data and the slide image data to generate a pantograph image data set.

[0024] Preferably, the AI ​​image analysis module includes a picture processing unit and a model building unit;

[0025] The image processing unit is used to collect a large amount of image data of bolt, nut and slide cracks from the Internet, ensure that the data contains images of bolt, nut and slide cracks under different angles, lighting conditions and backgrounds, and mark the corresponding preload force, torque value and crack length value for each image, and pre-process the image data, including resizing, cropping and contrast enhancement operations;

[0026] The model building unit is used to select a convolutional neural network CNN as an AI image recognition model for an image recognition task, and then set a large number of marked image data of bolts, nuts and sliding cracks as a training set, mark the pantograph image data set collected in real time as a verification set, and use the training set to train the selected AI image recognition model. During the training process, the AI ​​image recognition model is optimized by adjusting hyperparameters and monitoring performance indicators. The verification set is used to evaluate the performance of the AI ​​image recognition model by means of accuracy, precision and recall rate. The trained model is then deployed in an actual environment for testing and application, and the pantograph image collected in real time is input into the AI ​​image recognition model, and a pantograph data set is output, wherein the pantograph data set includes bolt preload TJ, nut torque TN, and crack length TL.

[0027] Preferably, the AI ​​defect algorithm module includes a vibration algorithm unit, an equipment abnormality algorithm unit and a related algorithm unit;

[0028] The vibration algorithm unit is used to perform dimensionless processing on the collected vibration data set, and then perform summary calculation to obtain the vibration coefficient Zdxs. The vibration coefficient Zdxs is obtained by the following formula:

[0029]

[0030] In the formula, K represents the spring stiffness, w represents the natural frequency, d represents the derivative operator, which is used to represent the rate of change of the function over time, t represents time, c represents the damping coefficient, x(t) represents the displacement of the pantograph over time t, and F(t) represents the external force on the pantograph over time t. represents the second-order derivative of displacement x(t) with respect to time t, i.e. the acceleration of the pantograph, It represents the first derivative of displacement x(t) with respect to time t, i.e. the speed of the pantograph.

[0031] Preferably, the equipment abnormality algorithm unit includes a slide defect calculation unit and a comprehensive calculation unit;

[0032] The vane defect calculation unit is used to calculate and analyze the deformation defects and crack defects generated by the pantograph vane according to the collected equipment defect data set and pantograph data set, and obtain the deformation defect Xbxs and the crack defect coefficient Lwxs;

[0033] The deformation defect Xbxs and crack defect coefficient Lwxs are obtained by the following formula:

[0034]

[0035]

[0036] Wherein, γ represents surface free energy, E is Young's modulus, π represents pi, which is taken as 3.14 here, and σ represents stress. The surface free energy γ, Young's modulus E and σ stress are directly obtained from the literature on material properties.

[0037] Preferably, the comprehensive calculation unit is used to obtain the comprehensive equipment defect coefficient Sbqx by comprehensive calculation after dimensionless processing of the obtained deformation defect Xbxs and crack defect coefficient Lwxs of the sliding plate of the pantograph and the defects of the bolt preload force TJ and the nut torque TN at the fastener connection of the pantograph;

[0038] The comprehensive equipment defect coefficient Sbqx is obtained by the following formula:

[0039]

[0040] Where a1 is the proportionality coefficient, which represents the proportional relationship between the deformation defect Xbxs and the crack defect coefficient Lwxs.

[0041] Preferably, the correlation algorithm unit is used to correlate the obtained vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx to obtain the pantograph defect coefficient Sdgx;

[0042] The pantograph defect coefficient Sdgx is obtained by the following formula:

[0043]

[0044] Wherein, b1 and b2 represent the preset weight values ​​of the vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx respectively, and b1≠b2, 0<b1≤0.43, 0<b2≤0.57, and the specific value is adjusted and set by the user, and B represents the correction constant.

[0045] Preferably, the intelligent assessment module includes a vibration assessment unit and a defect assessment unit;

[0046] The vibration evaluation unit is used to perform a preliminary comparative analysis of the abnormal vibration of the pantograph during the driving of the EMU according to the obtained vibration coefficient Zdxs and the preset abnormal vibration threshold Z. The specific evaluation scheme is as follows;

[0047] When the vibration coefficient Zdxs ≥ the vibration abnormality threshold Z, it means that the pantograph vibration is abnormal, and further defect assessment is performed at this time;

[0048] When the vibration coefficient Zdxs is less than the abnormal vibration threshold Z, it means that the pantograph vibration is normal and no further analysis is required;

[0049] The defect assessment unit is used to perform a defect assessment on the pantograph firmware connection position and the slider according to the obtained pantograph defect coefficient Sdgx, and further analyze the cause of the vibration. The specific assessment scheme is as follows;

[0050] When the pantograph defect coefficient Sdgx>equipment defect threshold Q, it means that the connection firmware and the slider of the pantograph are defective and abnormal. At this time, the first warning information is generated to notify the relevant staff to stop the car for maintenance;

[0051] When the pantograph defect coefficient Sdgx = equipment defect threshold Q, it means that the pantograph connection firmware and sliding piece are loose and worn. At this time, the second warning information is generated to notify the relevant staff to carry out inspection and maintenance after the train journey is completed;

[0052] When the pantograph defect coefficient Sdgx is less than the equipment defect threshold Q, it indicates that there is no abnormality in the connection firmware and the slider of the pantograph. At this time, the third warning information is generated to notify the relevant staff to check the shock absorber system of the pantograph.

[0053] An AI intelligent analysis method for railway power supply equipment defects includes the following steps:

[0054] S1. First, deploy a vibration integrated sensor group, an equipment defect integrated sensor group and a visual inspection device around the pantograph to detect and collect the vibration data, equipment defect data and image data of the pantograph in real time, and then obtain the vibration data set, equipment defect data set and pantograph image data set after preprocessing;

[0055] S2. Then, an AI image recognition model is constructed by using a convolutional neural network (CNN). A large number of defective images of pantograph firmware and sliders are collected from the Internet, and each image is marked. Then, the defective image data of the pantograph firmware and sliders are quantitatively identified through the AI ​​image recognition model, and a pantograph defect data set is output.

[0056] S3. Based on the collected vibration data set, dimensionless processing is performed to calculate the vibration coefficient Zdxs. Then, based on the equipment defect data set and the pantograph defect data set, the vane deformation defect Xbxs and the crack defect coefficient Lwxs are calculated. Further comprehensive calculation is performed to obtain the comprehensive equipment defect coefficient Sbqx.

[0057] S4. Combine the vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx to perform correlation calculation to obtain the final pantograph defect coefficient Sdgx;

[0058] S5. Finally, the preset vibration abnormality threshold Z is compared and evaluated with the obtained vibration coefficient Zdxs, and the secondary evaluation mechanism is triggered based on the preliminary comparative evaluation results to conduct a comprehensive evaluation of the preset equipment defect threshold Q and the obtained comprehensive equipment defect coefficient Sbqx, and generate relevant early warning information.

[0059] Beneficial Effects

[0060] The present invention provides an AI intelligent analysis method and system for railway power supply equipment defects. It has the following beneficial effects:

[0061] (1) The system realizes the automated real-time monitoring and analysis of railway power supply equipment, especially pantographs, by introducing an AI intelligent analysis system. The pantograph detection module arranges several detection points around the pantograph, and installs a vibration integrated sensor group, an equipment defect integrated sensor group and a visual detection device to obtain vibration data, equipment defect data and pantograph image data in real time. After preprocessing the collected data, the data acquisition module generates a vibration data set, an equipment defect data set and a pantograph image data set respectively. The AI ​​image analysis module constructs an AI image recognition model by constructing a convolutional neural network CNN, identifies the pantograph image data set, quantitatively identifies the defects of the pantograph connector and the slider, and outputs the pantograph defect data set. The AI ​​defect algorithm module constructs a vibration algorithm based on the vibration data set to obtain the vibration coefficient Zdxs; constructs a defect algorithm based on the equipment defect data set and the pantograph defect data set to obtain the comprehensive equipment defect coefficient Sbqx, and calculates and associates it with the vibration coefficient Zdxs to obtain the pantograph defect coefficient Sdgx. The intelligent assessment module performs a preliminary assessment by presetting the vibration abnormality threshold Z and the vibration coefficient Zdxs, triggers the secondary assessment mechanism, performs a comprehensive assessment by presetting the equipment defect threshold Q and the comprehensive equipment defect coefficient Sbqx, and generates relevant early warning information. This system significantly improves the detection efficiency and accuracy, reduces the workload and subjective errors of manual inspections, ensures real-time monitoring of the pantograph status and early defect warnings, and prevents serious equipment failures and safety hazards.

[0062] (2) The system monitors the operating status of the pantograph in real time through a variety of sensors, and uses AI algorithms to analyze the acquired data to promptly detect early defects of the pantograph. The intelligent evaluation module can generate early warning information based on the evaluation results of the vibration coefficient and defect coefficient, helping maintenance personnel to take corresponding measures in a timely manner to prevent the defects from further deteriorating, avoid serious equipment failures and safety hazards, and improve the reliability and safety of the system.

[0063] (3) The system's detection process mainly relies on automated equipment and AI algorithms, which reduces reliance on manpower, reduces the frequency and cost of manual inspections and regular maintenance, and can accurately identify and locate pantograph defects, thereby optimizing maintenance strategies, reducing unnecessary downtime and maintenance times, and further reducing maintenance costs. At the same time, the system's automation and real-time nature reduce safety risks in the manual detection process and improve overall operational safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a schematic diagram of the process of an AI intelligent analysis system for railway power supply equipment defects of the present invention;

[0065] Figure 2 This is a schematic diagram of the steps of an AI intelligent analysis method for defects in railway power supply equipment according to the present invention; DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] Example 1

[0068] See also Figure 1 , the present invention provides an AI intelligent analysis system for railway power supply equipment defects. To achieve the above purpose, the present invention is implemented through the following technical solutions: including a pantograph detection module, a data acquisition module, an AI image analysis module, an AI defect algorithm module and an intelligent evaluation module;

[0069] The pantograph detection module is used to arrange several detection points around the pantograph and install relevant detection equipment and visual detection equipment. The pantograph of the train is inspected in real time and vibration data, equipment defect data and pantograph image data are obtained;

[0070] The data acquisition module is used to pre-process the collected vibration data, equipment defect data and pantograph data respectively, and then acquire the vibration data set, equipment defect data set and pantograph image data set during data acquisition;

[0071] The AI ​​image analysis module is used to build an AI image recognition model, transfer the captured pantograph image data set to the AI ​​image recognition model, quantitatively identify the defects of the pantograph connector and the slider, and output the pantograph defect data set;

[0072] The AI ​​defect algorithm module is used to construct a vibration algorithm based on the vibration feature data set to obtain the vibration coefficient Zdxs, and then construct a defect algorithm based on the equipment defect data set and the pantograph defect data set to obtain the comprehensive equipment defect coefficient Sbqx, and then associate it with the vibration coefficient Zdxs to obtain the pantograph defect coefficient Sdgx;

[0073] The intelligent evaluation module is used to perform preliminary comparative evaluation on the preset vibration abnormality threshold Z and the obtained vibration coefficient Zdxs, and trigger the secondary evaluation mechanism based on the preliminary comparative evaluation results to perform comprehensive evaluation on the preset equipment defect threshold Q and the obtained comprehensive equipment defect coefficient Sbqx, and generate relevant early warning information.

[0074] In this embodiment, the system arranges detection points around the pantograph through the pantograph detection module, collects vibration data, equipment defect data and pantograph image data in real time, and transmits them to the data acquisition module. The data acquisition module preprocesses these data to generate vibration data sets, equipment defect data sets and pantograph image data sets, providing basic data support for subsequent analysis. Next, the AI ​​image analysis module constructs an AI image recognition model, quantitatively identifies the pantograph image data set, accurately detects the defects of the pantograph connector and the slider, and generates a detailed pantograph defect data set. At the same time, the AI ​​defect algorithm module constructs a vibration and defect algorithm by analyzing the vibration data set and the equipment defect data set, and calculates the comprehensive equipment defect coefficient Sbqx and the pantograph defect coefficient Sdgx. These calculation results are associated with the vibration feature data, providing the system with multi-dimensional defect identification and analysis capabilities. Finally, the intelligent evaluation module performs comprehensive evaluation and comparative analysis on the acquired data through the preset vibration abnormality threshold Z and equipment defect threshold Q. According to the preliminary evaluation results, the system can trigger a secondary evaluation mechanism to further improve the accuracy and reliability of the evaluation. Compared with traditional detection methods, the system realizes real-time, accurate and intelligent defect detection and evaluation, which not only improves detection efficiency and accuracy, but also greatly reduces labor costs and misjudgment rate, thereby ensuring the safety and reliability of train operation.

[0075] Example 2

[0076] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the pantograph detection module includes a vibration data detection unit, an ,equipment defect detection unit and a pantograph image detection unit;

[0077] The vibration data detection unit is used to install a vibration integrated sensor group around the pantograph to perform real-time detection of the vibration generated by the pantograph during the operation of the EMU to obtain vibration data. The vibration integrated sensor group includes a force sensor, an accelerometer and a laser displacement sensor;

[0078] Force sensor: measures the actual force on the pantograph;

[0079] Accelerometer: used to measure the acceleration of the system, thereby indirectly obtaining the dynamic parameters of the pantograph;

[0080] Laser displacement sensor: used to measure the displacement of the pantograph;

[0081] The equipment defect detection unit is used to install an equipment defect integrated sensor group at the firmware connection of the pantograph to perform real-time detection of equipment defect data of the pantograph connection firmware and the slider. The equipment defect integrated sensor group includes a pressure sensor and a linear displacement sensor.

[0082] Pressure sensor: used to measure the force applied to the slider or other parts of the pantograph;

[0083] Linear displacement sensor: used to measure the displacement or deformation of the slide;

[0084] The pantograph image detection unit is used to install visual sensors at the firmware connection and the slide to collect defect images of bolts, nuts and slide surfaces at the firmware connection in real time from eight different angles.

[0085] The data acquisition module includes a vibration data acquisition unit, an equipment defect data acquisition unit and an image acquisition unit;

[0086] The vibration data acquisition unit is used to collect vibration data detected by the vibration integrated sensor group in real time through wireless transmission, and to classify and summarize the collected vibration data after preprocessing to generate a vibration data set. The preprocessing method includes denoising, missing value processing, outlier detection, outlier processing and normalization processing. The vibration data set includes the pantograph displacement distance x and the external force F;

[0087] The equipment defect data acquisition unit is used to collect the equipment defect data detected by the equipment defect integrated sensor group in real time through wireless transmission, and to classify and summarize the collected equipment defect data after preprocessing to generate an equipment defect data set. The preprocessing method includes denoising, missing value processing, outlier detection, outlier processing and normalization processing. The equipment defect data set includes the external force HF of the slider and the deformation variable HX of the slider;

[0088] The image acquisition unit is used to classify the defect images of bolts, nuts and slide surfaces collected at the connection of the acquisition firmware, archive the photos of the bolts and nuts to generate connection image data, archive the photos of the slide separately to generate slide image data, and then summarize the connection image data and slide image data to generate a pantograph image data set.

[0089] In this embodiment, the system comprehensively and accurately monitors the status of the pantograph in real time by comprehensively utilizing the vibration data detection unit, the equipment defect detection unit and the pantograph image detection unit. The vibration data detection unit obtains the vibration data of the pantograph in real time through the vibration integrated sensor group installed around the pantograph, and accurately measures its actual force, dynamic parameters and displacement. The equipment defect detection unit detects the equipment defects of the pantograph's firmware connection and the slider in real time through the pressure sensor and the linear displacement sensor, and accurately measures the force applied to the slider and its displacement or deformation. The pantograph image detection unit uses the visual sensor installed at the key position to collect the defect images of the firmware connection and the slider surface from multiple angles in real time to ensure the comprehensiveness and accuracy of the detection. The data acquisition module further enhances the real-time data processing capability of the system. The vibration data acquisition unit collects and preprocesses the vibration data in real time through wireless transmission to generate a vibration data set after denoising, normalization and other processing. The equipment defect data acquisition unit also collects the equipment defect data in real time through wireless transmission, and generates the equipment defect data set after corresponding preprocessing, ensuring the integrity and accuracy of the data. The image acquisition unit classifies and archives the defect images of bolts, nuts and slide surfaces, generates detailed connection image data and slide image data, and summarizes them into a pantograph image data set, providing a high-quality data foundation for subsequent AI image analysis.

[0090] Example 3

[0091] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: The AI ​​image analysis module includes an image processing unit and a model building unit;

[0092] The image processing unit is used to collect a large amount of image data of bolt, nut and slide cracks from the Internet, ensure that the data contains images of bolt, nut and slide cracks under different angles, lighting conditions and backgrounds, and mark the corresponding preload force, torque value and crack length value for each image, and pre-process the image data, including resizing, cropping and contrast enhancement operations;

[0093] The model building unit is used to select the convolutional neural network CNN as the AI ​​image recognition model for the image recognition task, and then set the labeled image data of a large number of bolts, nuts and sliding cracks as the training set, mark the pantograph image data set collected in real time as the verification set, and use the training set to train the selected AI image recognition model. During the training process, the AI ​​image recognition model is optimized by adjusting hyperparameters and monitoring performance indicators. The verification set is used to evaluate the performance of the AI ​​image recognition model by means of accuracy, precision and recall rate. The trained model is then deployed in the actual environment for testing and application, and the pantograph image collected in real time is input into the AI ​​image recognition model to output the pantograph data set, which includes bolt preload TJ, nut torque TN, and crack length TL.

[0094] In this embodiment, the system greatly improves the intelligence and automation level of the pantograph detection system by introducing an AI image analysis module. The image processing unit collects a large number of bolt, nut and slide crack images under multiple angles, multiple lighting conditions and multiple backgrounds from the Internet, and marks the relevant bolt preload, nut torque value and slide crack length value, ensuring the diversity and richness of the training data. Preprocessing these images further improves the quality and applicability of the data. The model building unit selects the convolutional neural network CNN as the AI ​​image recognition model for the image recognition task, and trains the CNN model by setting a large amount of labeled image data as a training set, and optimizes the model performance by adjusting hyperparameters and monitoring performance indicators. The real-time collected pantograph image data set is used as a validation set, and the efficiency and accuracy of the model are ensured by evaluating the accuracy, precision and recall rate. The optimized and verified model is deployed in the actual environment, processes and recognizes the pantograph image in real time, and outputs the pantograph data set including bolt preload, nut torque and crack length to ensure accurate monitoring of the pantograph status.

[0095] Example 4

[0096] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: The AI ​​defect algorithm module includes a vibration algorithm unit, an equipment anomaly algorithm unit and an associated algorithm unit;

[0097] The vibration algorithm unit is used to perform dimensionless processing on the collected vibration data set, and then perform summary calculation to obtain the vibration coefficient Zdxs. The vibration coefficient Zdxs is obtained by the following formula;

[0098]

[0099] In the formula, K represents the spring stiffness, w represents the natural frequency, d represents the derivative operator, which is used to represent the rate of change of the function over time, t represents time, c represents the damping coefficient, x(t) represents the displacement of the pantograph over time t, and F(t) represents the external force on the pantograph over time t. represents the second-order derivative of displacement x(t) with respect to time t, i.e. the acceleration of the pantograph, It represents the first derivative of displacement x(t) with respect to time t, i.e. the speed of the pantograph.

[0100] In this embodiment, the system realizes accurate analysis and evaluation of the pantograph status by introducing the vibration algorithm unit, the equipment abnormality algorithm unit and the associated algorithm unit through the AI ​​defect algorithm module. The vibration algorithm unit first performs dimensionless processing on the collected vibration data set to eliminate the influence between different dimensions, and then obtains the vibration coefficient Zdxs through summary calculation. The equipment abnormality algorithm unit uses the equipment defect data set to analyze the stress condition and deformation degree of the equipment, establish an equipment abnormality algorithm model, and calculate the comprehensive equipment defect coefficient Sbqx. The associated algorithm unit associates the vibration coefficient Zdxs with the comprehensive equipment defect coefficient Sbqx to generate the pantograph defect coefficient Sdgx. Through multi-dimensional data cross-validation, accurate identification and quantitative evaluation of pantograph defects can be achieved.

[0101] Example 5

[0102] This embodiment is explained in Example 3, please refer to Figure 1 ,Specifically: the equipment abnormality algorithm unit includes a slide defect calculation unit and a ,comprehensive calculation unit;

[0103] The vane defect calculation unit is used to calculate and analyze the deformation defects and crack defects generated by the pantograph vane based on the collected equipment defect data set and pantograph data set, and obtain the deformation defect Xbxs and crack defect coefficient Lwxs;

[0104] The deformation defect Xbxs and crack defect coefficient Lwxs are obtained by the following formula;

[0105]

[0106]

[0107] Wherein, γ represents surface free energy, E is Young's modulus, π represents pi, which is taken as 3.14 here, and σ represents stress. The surface free energy γ, Young's modulus E and σ stress are directly obtained from the literature on material properties.

[0108] The comprehensive calculation unit is used to obtain the comprehensive equipment defect coefficient Sbqx by comprehensive calculation after dimensionless processing of the obtained deformation defect Xbxs and crack defect coefficient Lwxs of the sliding plate of the pantograph and the defects of the bolt preload force TJ and the nut torque TN at the fastener connection of the pantograph;

[0109] The comprehensive equipment defect coefficient Sbqx is obtained by the following formula;

[0110]

[0111] Where a1 is the proportionality coefficient, which represents the proportional relationship between the deformation defect Xbxs and the crack defect coefficient Lwxs.

[0112] In this embodiment, the system accurately calculates and comprehensively evaluates the equipment defects of the pantograph through the equipment abnormality algorithm unit, the slider defect calculation unit and the comprehensive calculation unit. The slider defect calculation unit analyzes the deformation defects and crack defects of the pantograph slider based on the equipment defect data set and the pantograph data set, and calculates the deformation defect coefficient Xbxs and the crack defect coefficient Lwxs respectively. The deformation defect coefficient Xbxs is calculated by formula, using material property data such as surface free energy γ, Young's modulus E and stress σ to ensure the scientificity and accuracy of the calculation. Similarly, the crack defect coefficient Lwxs is calculated by related formulas, which fully considers the physical and mechanical characteristics of the material. The comprehensive calculation unit performs dimensionless processing on the deformation defect coefficient Xbxs and crack defect coefficient Lwxs of the slider with the bolt preload TJ and nut torque TN at the connection of the pantograph firmware, and comprehensively calculates the comprehensive equipment defect coefficient Sbqx. The comprehensive calculation formula takes into account the proportional relationship between the deformation defect and the crack defect, and ensures the accuracy and representativeness of the comprehensive equipment defect coefficient through the proportional coefficient a1.

[0113] Example 6

[0114] This embodiment is explained in Example 5. Please refer to Figure 1 Specifically: the correlation algorithm unit is used to correlate the obtained vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx to obtain the pantograph defect coefficient Sdgx;

[0115] The pantograph defect coefficient Sdgx is obtained by the following formula;

[0116]

[0117] Wherein, b1 and b2 represent the preset weight values ​​of the vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx respectively, and b1≠b2, 0<b1≤0.43, 0<b2≤0.57, and the specific value is adjusted and set by the user, and B represents the correction constant.

[0118] In this embodiment, the system comprehensively considers vibration data and equipment defect data, and the algorithm realizes multi-dimensional data fusion, ensuring the comprehensiveness and accuracy of defect assessment. Secondly, the flexible adjustment of weight values ​​makes the system more adaptable, and users can optimize the accuracy and reliability of defect assessment according to actual conditions. Finally, the precise calculation of the pantograph defect coefficient Sdgx provides key data support for the intelligent evaluation module, which helps to achieve more accurate fault warning and maintenance decisions. Compared with traditional detection methods, the associated algorithm unit significantly improves the accuracy and reliability of pantograph defect detection through scientific calculation and comprehensive evaluation, reduces the misjudgment rate and missed judgment rate, thereby effectively ensuring the safety and stability of train operation and improving the overall maintenance level of the rail transit system.

[0119] Example 7

[0120] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the intelligent evaluation module includes a vibration evaluation unit and a defect evaluation unit;

[0121] The vibration evaluation unit is used to perform a preliminary comparative analysis of the abnormal vibration of the pantograph during the operation of the EMU based on the obtained vibration coefficient Zdxs and the preset abnormal vibration threshold Z. The specific evaluation scheme is as follows;

[0122] When the vibration coefficient Zdxs ≥ the vibration abnormality threshold Z, it means that the pantograph vibration is abnormal, and further defect assessment is performed at this time;

[0123] When the vibration coefficient Zdxs is less than the abnormal vibration threshold Z, it means that the pantograph vibration is normal and no further analysis is required;

[0124] The defect assessment unit is used to perform a defect assessment on the pantograph firmware connection position and the slider according to the obtained pantograph defect coefficient Sdgx and the preset equipment defect threshold Q, and further analyze the cause of the vibration. The specific assessment scheme is as follows;

[0125] When the pantograph defect coefficient Sdgx>equipment defect threshold Q, it means that the connection firmware and the slider of the pantograph are defective and abnormal. At this time, the first warning information is generated to notify the relevant staff to stop the car for maintenance;

[0126] When the pantograph defect coefficient Sdgx = equipment defect threshold Q, it means that the pantograph connection firmware and sliding piece are loose and worn. At this time, the second warning information is generated to notify the relevant staff to carry out inspection and maintenance after the train journey is completed;

[0127] When the pantograph defect coefficient Sdgx is less than the equipment defect threshold Q, it indicates that there is no abnormality in the connection firmware and the slider of the pantograph. At this time, the third warning information is generated to notify the relevant staff to check the shock absorber system of the pantograph.

[0128] In this embodiment, the system realizes intelligent evaluation of the pantograph state through the vibration evaluation unit and the defect evaluation unit through the intelligent evaluation module. First, the vibration abnormality threshold Z is preset through the vibration evaluation unit, and a preliminary comparative evaluation is performed with the obtained vibration coefficient Zdxs to analyze whether the vibration of the pantograph in different states is abnormal. At the same time, when the pantograph vibration is detected to be abnormal, the defect evaluation unit is quickly started to compare and analyze whether there are abnormal defects between the pantograph slider and the connecting firmware according to the pantograph defect coefficient Sdgx, and the corresponding warning level is output according to the evaluation results through relevant staff. This method can quickly judge the vibration state, identify abnormalities in time; accurately locate and diagnose defects, provide targeted warning information; generate warning information of different levels, and optimize the allocation of maintenance resources. Compared with traditional methods, the intelligent evaluation module improves the accuracy and reliability of pantograph monitoring, reduces the risk of misjudgment and missed judgment, ensures the safety of EMU operation, and improves the timeliness and pertinence of maintenance work.

[0129] Example 8

[0130] See also Figure 1 and Figure 2 , an AI intelligent analysis method for railway power supply equipment defects, comprising the following steps:

[0131] S1. First, deploy a vibration integrated sensor group, an equipment defect integrated sensor group and a visual inspection device around the pantograph to detect and collect the vibration data, equipment defect data and image data of the pantograph in real time, and then obtain the vibration data set, equipment defect data set and pantograph image data set after preprocessing;

[0132] S2. Then, an AI image recognition model is constructed by using a convolutional neural network (CNN). A large number of defective images of pantograph firmware and sliders are collected from the Internet, and each image is marked. Then, the defective image data of the pantograph firmware and sliders are quantitatively identified through the AI ​​image recognition model, and a pantograph defect data set is output.

[0133] S3. Based on the collected vibration data set, dimensionless processing is performed to calculate the vibration coefficient Zdxs. Then, based on the equipment defect data set and the pantograph defect data set, the vane deformation defect Xbxs and the crack defect coefficient Lwxs are calculated. Further comprehensive calculation is performed to obtain the comprehensive equipment defect coefficient Sbqx.

[0134] S4. Combine the vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx to perform correlation calculation to obtain the final pantograph defect coefficient Sdgx;

[0135] S5. Finally, the preset vibration abnormality threshold Z is compared and evaluated with the obtained vibration coefficient Zdxs, and the secondary evaluation mechanism is triggered based on the preliminary comparative evaluation results to conduct a comprehensive evaluation of the preset equipment defect threshold Q and the obtained comprehensive equipment defect coefficient Sbqx, and generate relevant early warning information.

[0136] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A railway power supply equipment defect AI intelligent analysis system, characterized by: It includes pantograph detection module, data acquisition module, AI image analysis module, AI defect algorithm module and intelligent evaluation module; The pantograph detection module is used to arrange several detection points around the pantograph, and install a vibration integrated sensor group, an equipment defect integrated sensor group and a visual sensor to perform real-time detection of the pantograph of the train and obtain vibration data, equipment defect data and pantograph image data; The data acquisition module is used to pre-process the detected vibration data, equipment defect data and pantograph image data respectively to obtain a vibration data set, an equipment defect data set and a pantograph image data set; the vibration data set includes the pantograph displacement distance x and the external force F; the equipment defect data set includes the external force HF of the slider and the slider deformation HX; the AI ​​image analysis module is used to build an AI image recognition model, transmit the captured pantograph image data set to the AI ​​image recognition model, quantitatively identify the defects of the pantograph connector and the slider, and output the pantograph defect data set, and the pantograph defect data set includes the bolt preload force TJ, the nut torque TN and the crack length TL; The AI ​​defect algorithm module is used to construct a vibration algorithm based on the vibration data set to obtain the vibration coefficient Zdxs, and then construct a defect algorithm based on the equipment defect data set and the pantograph defect data set to obtain the comprehensive equipment defect coefficient Sbqx, and then associate it with the vibration coefficient Zdxs to obtain the pantograph defect coefficient Sdgx; The intelligent evaluation module is used to perform preliminary comparative evaluation on the preset vibration abnormality threshold Z and the obtained vibration coefficient Zdxs, and trigger the secondary evaluation mechanism according to the preliminary comparative evaluation result, perform comprehensive evaluation on the preset equipment defect threshold Q and the obtained pantograph defect coefficient Sdgx, and generate relevant early warning information; Wherein, the AI ​​defect algorithm module includes a vibration algorithm unit, an equipment abnormality algorithm unit and an associated algorithm unit; The vibration algorithm unit is used to perform dimensionless processing on the collected vibration data set, and then perform summary calculation to obtain the vibration coefficient Zdxs. The vibration coefficient Zdxs is obtained by the following formula: ； In the formula, K represents the spring stiffness, w represents the natural frequency, d represents the derivative operator, which is used to represent the rate of change of the function over time, t represents time, and c represents the damping coefficient; x(t) represents the displacement of the pantograph over time t, and F(t) represents the external force on the pantograph over time t; The equipment abnormality algorithm unit includes a slide defect calculation unit and a comprehensive calculation unit; The vane defect calculation unit is used to calculate and analyze the deformation defects and crack defects generated by the pantograph vane according to the collected equipment defect data set and pantograph defect data set, and obtain the deformation defect Xbxs and the crack defect coefficient Lwxs; The deformation defect Xbxs and crack defect coefficient Lwxs are obtained by the following formula: ； ； In the formula, represents the surface free energy, E is Young's modulus, represents pi, represents stress, surface free energy , Young's modulus E and Stress is obtained directly from literature data on material properties; The comprehensive calculation unit is used to obtain the comprehensive equipment defect coefficient Sbqx by comprehensive calculation after dimensionless processing of the obtained deformation defect Xbxs and crack defect coefficient Lwxs of the sliding plate of the pantograph and the defects of the bolt preload force TJ and the nut torque TN at the connection of the pantograph firmware; The comprehensive equipment defect coefficient Sbqx is obtained by the following formula: ； In the formula, a1 is the proportionality coefficient; The correlation algorithm unit is used to correlate the obtained vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx to obtain the pantograph defect coefficient Sdgx; The pantograph defect coefficient Sdgx is obtained by the following formula: ； Wherein, b1 and b2 represent the preset weight values ​​of the vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx respectively, and their specific values ​​are adjusted and set by the user, and B represents the correction constant.

2. According to claim 1, a railway power supply equipment defect AI intelligent analysis system is characterized by: The pantograph detection module includes a vibration data detection unit, an equipment defect detection unit and a pantograph image detection unit; The vibration data detection unit is used to install a vibration integrated sensor group around the pantograph to perform real-time detection of the vibration generated by the pantograph during the operation of the EMU to obtain vibration data. The vibration integrated sensor group includes a force sensor, an accelerometer and a laser displacement sensor; The equipment defect detection unit is used to install an equipment defect integrated sensor group at the pantograph firmware connection to perform real-time detection of equipment defect data of the pantograph connection firmware and the slide, and the equipment defect integrated sensor group includes a pressure sensor and a linear displacement sensor; The pantograph image detection unit is used to install visual sensors at the firmware connection and the slide, and to collect defect images of bolts, nuts and slide surfaces at the firmware connection in real time from eight different angles.

3. According to claim 2, a railway power supply equipment defect AI intelligent analysis system is characterized by: The data acquisition module includes a vibration data acquisition unit, an equipment defect data acquisition unit and an image acquisition unit; The vibration data acquisition unit is used to collect vibration data detected by the vibration integrated sensor group in real time through wireless transmission, and to classify and summarize the collected vibration data after preprocessing to generate a vibration data set; The equipment defect data acquisition unit is used to collect equipment defect data detected by the equipment defect integrated sensor group in real time through wireless transmission, and to pre-process the collected equipment defect data and classify and summarize them to generate an equipment defect data set; The image acquisition unit is used to classify the acquired defect images of bolts, nuts and slide surfaces at the connection of the acquisition firmware, archive the photos of the bolts and nuts to generate connection image data, archive the photos of the slide separately to generate slide image data, and then aggregate the connection image data and the slide image data to generate a pantograph image data set; The preprocessing methods include denoising, missing value processing, outlier detection, outlier processing and normalization processing.

4. According to claim 1, a railway power supply equipment defect AI intelligent analysis system is characterized by: The AI ​​image analysis module includes a picture processing unit and a model building unit; The image processing unit is used to collect a large amount of image data of bolt, nut and slide cracks from the Internet, ensure that the data contains images of bolt, nut and slide cracks under different angles, lighting conditions and backgrounds, and mark the corresponding preload force, torque value and crack length value for each image, and then preprocess the image data, including resizing, cropping and contrast enhancement operations; The model building unit is used to select a convolutional neural network (CNN) as an AI image recognition model for an image recognition task, and then set a large number of marked image data of bolts, nuts and sliding cracks as a training set, mark the pantograph image data set collected in real time as a verification set, and use the training set to train the selected AI image recognition model. During the training process, the AI ​​image recognition model is optimized by adjusting hyperparameters and monitoring performance indicators, and the verification set is used to evaluate the performance of the AI ​​image recognition model by means of accuracy, precision and recall rate. The trained model is then deployed in an actual environment for testing and application, and the pantograph image collected in real time is input into the AI ​​image recognition model, and a pantograph defect data set is output.

5. According to claim 1, a railway power supply equipment defect AI intelligent analysis system is characterized by: The intelligent assessment module includes a vibration assessment unit and a defect assessment unit; The vibration evaluation unit is used to perform a preliminary comparative analysis of the abnormal vibration of the pantograph during the driving of the EMU according to the obtained vibration coefficient Zdxs and the preset abnormal vibration threshold Z. The specific evaluation scheme is as follows; When the vibration coefficient Zdxs ≥ the vibration abnormality threshold Z, it means that the pantograph vibration is abnormal, and further defect assessment is performed at this time; When the vibration coefficient Zdxs is less than the abnormal vibration threshold Z, it means that the pantograph vibration is normal and no further analysis is required; The defect assessment unit is used to perform a defect assessment on the pantograph firmware connection position and the slider according to the obtained pantograph defect coefficient Sdgx, and further analyze the cause of the vibration. The specific assessment scheme is as follows; When the pantograph defect coefficient Sdgx>equipment defect threshold Q, it means that the connection firmware and the slider of the pantograph are defective and abnormal. At this time, the first warning information is generated to notify the relevant staff to stop the car for maintenance; When the pantograph defect coefficient Sdgx = equipment defect threshold Q, it means that the pantograph connection firmware and sliding piece are loose and worn. At this time, the second warning information is generated to notify the relevant staff to carry out inspection and maintenance after the train journey ends; When the pantograph defect coefficient Sdgx is less than the equipment defect threshold Q, it indicates that there is no abnormality in the connection firmware and the slider of the pantograph. At this time, the third warning information is generated to notify the relevant staff to check the shock absorber system of the pantograph.

6. A railway power supply equipment defect AI intelligent analysis method, comprising a railway power supply equipment defect AI intelligent analysis system according to any one of claims 1 to 5, characterized in that: The following steps are involved: S1. First, deploy a vibration integrated sensor group, an equipment defect integrated sensor group and a visual inspection device around the pantograph to detect and collect the vibration data, equipment defect data and image data of the pantograph in real time, and then obtain the vibration data set, equipment defect data set and pantograph image data set after preprocessing; S2. Then, an AI image recognition model is constructed by using a convolutional neural network (CNN). A large number of defective images of pantograph firmware and sliders are collected from the Internet, and each image is marked. Then, the defective image data of pantograph firmware and sliders are quantitatively identified through the AI ​​image recognition model, and a pantograph defect data set is output. S3. Based on the collected vibration data set, dimensionless processing is performed to calculate the vibration coefficient Zdxs. Then, based on the equipment defect data set and the pantograph defect data set, the vane deformation defect Xbxs and the crack defect coefficient Lwxs are calculated. Further comprehensive calculation is performed to obtain the comprehensive equipment defect coefficient Sbqx. S4. Combine the vibration coefficient Zdxs and the comprehensive equipment defect coefficient Sbqx to perform correlation calculation to obtain the final pantograph defect coefficient Sdgx; S5. Finally, a preliminary comparative evaluation is performed on the preset vibration abnormality threshold Z and the obtained vibration coefficient Zdxs, and a secondary evaluation mechanism is triggered based on the preliminary comparative evaluation result to conduct a comprehensive evaluation on the preset equipment defect threshold Q and the obtained pantograph defect coefficient Sdgx, and generate relevant early warning information.

Images