Railway signal cable welding method, system and equipment based on deep learning

Abstract

The invention relates to the field of image processing, and discloses a deep learning-based railway signal cable welding method, system and equipment, and the welding method comprises the following steps: S1, carrying out synchronous welding operation on a signal conductor and a shielding layer; s2, the image is detected through a YOLOv5 algorithm, the electrical performance data is detected through an MLP algorithm, whether each detection result meets the welding requirement or not is judged, if yes, the step S3 is executed, and if not, the step S1 is executed; s3, integrating output results in the S2 through a weighted voting mechanism, performing evaluation, judging whether an evaluation score is lower than a preset score or not, if so, repeating the operation in the S1, and if not, executing the S4; and S4, a multi-layer composite protective coating from the bottom layer to the outer layer is constructed in the welding completion area. According to the invention, a complete high-quality production process is formed, dynamic optimization of the welding process is realized, uniformity and consistency of welding spot protection are ensured, comprehensive welding quality tracing is provided, and efficient connection of the signal conductor and the shielding layer in the same process is realized.

Description

Technical Field

[0001] The present application relates to the field of image processing, and specifically to a welding method, system and equipment for railway signal cables based on deep learning. Background Art

[0002] As an important part of the railway transportation system, railway signal cables undertake the key tasks of transmitting signals, data and electricity. Their performance is directly related to the safety and reliability of the railway signal system. In modern railway systems, especially in the fields of high-speed railways and intelligent railways, the requirements for signal cables are getting higher and higher. They must have excellent conductivity, anti-interference ability, mechanical strength and durability to cope with the test of long-term high loads, high vibrations and harsh environments. The traditional railway signal cable welding process faces technical bottlenecks. The problems of unstable electrical performance, insufficient mechanical strength and poor corrosion resistance in the welding area have long plagued the development of the industry.

[0003] At the same time, with the advancement of intelligent railway signal systems, higher requirements are placed on intelligent monitoring and quality inspection of welding processes. Traditional manual inspection methods are not only inefficient, but also have the problems of strong subjectivity and high misjudgment rate. The uncontrollability of welding quality may lead to hidden faults during cable operation, threatening the reliability and stability of railway signal systems. In addition, in the use environment of railway signal cables, the solder joints are exposed to high humidity and high salt fog, which are prone to corrosion, thus affecting the reliability of long-term operation. Summary of the invention

[0004] The technical problem to be solved by this application is to overcome the shortcomings of the existing technology and provide a welding method, system and equipment for railway signal cables based on deep learning, forming a complete high-quality production process. Deep learning algorithms can be used to detect welding quality, achieve efficient connection of signal conductors and shielding layers in the same process, and use multi-layer coatings to significantly improve the protective performance of solder joints.

[0005] To achieve the above-mentioned object, the first aspect of the present application provides a welding method for railway signal cables based on deep learning, the welding method comprising the following steps: Step S1, performing a synchronous welding operation on the signal conductor and the shielding layer, fixing the cable by a multi-axis positioning fixture, using a laser system with integrated temperature feedback to perform a welding operation, and using at least two groups of high-power laser heads to act on the welding areas of the signal conductor and the shielding layer respectively, so as to achieve synchronous welding of the signal conductor and the shielding layer; Pre-place a silver-based solder and apply a fluorine-containing soldering flux at the welding interface between the signal conductor and the shielding layer, and implement gradient heating through multi-stage temperature control, including a preheating stage, a main heating stage, a constant temperature maintenance stage, and a cooling stage; The laser power of the laser system is adjusted in real time by a closed-loop temperature control system to keep the temperature of each welding area within a preset threshold range; Step S2, welding area quality detection, using the YOLOv5 algorithm to detect the collected welding area image, using the MLP algorithm to detect the collected welding area electrical performance data, and judging whether each detection result meets the welding requirements. If so, continue to step S3, if not, melt the welding area to remove the solder, clean the signal conductor and the shielding layer, and then execute step S1; Step S3, after synthesizing the output results of each algorithm in step S2 through a weighted voting mechanism, the synthesized output results are evaluated to determine whether the evaluation score is lower than a preset score. If so, the soldering area is melted to remove the solder, the signal conductor and the shielding layer are cleaned, and step S1 is repeated. If not, step S4 is executed; Step S4, constructing a multi-layer composite protective coating from the bottom layer to the outer layer in the welding completion area.

[0006] Optionally, the laser system in step S1 includes: A first laser head acting on the signal conductor welding area has an output power range of 20-40W and a working distance of 3-8mm; The second laser head acting on the welding area of ​​the shielding layer has an output power range of 15-25W and a working distance of 3-8mm; The first laser head and the second laser head are both integrated with an infrared temperature measurement module with a temperature detection accuracy within ±2°C.

[0007] Optionally, the multi-stage temperature control in step S1 includes: A preheating stage, in which the temperature of the signal conductor interface is raised to 180-220° C. and the temperature of the shielding layer interface is raised to 140-160° C. by the laser system; A main heating stage, in which the temperature of the signal conductor interface is raised to 630-720° C. and the temperature of the shielding layer interface is raised to 590-660° C. by the laser system; A constant temperature maintenance stage, wherein the laser system is used to maintain the temperature of each soldering point within a stable fluctuation range of ±10°C; In the cooling stage, a gas convection cooling method is adopted in the cooling stage, and the solder joints are evenly cooled by a high-pressure fan, and the cooling rate is not less than 30°C / s.

[0008] Optionally, the closed-loop temperature control system in step S1 has a temperature sampling frequency of no less than 50 Hz for each welding area, and the closed-loop temperature control system adjusts the laser power in real time, including establishing a three-dimensional mathematical model of the temperature field in the welding area, and dynamically adjusting the laser power of the first laser head and the second laser head according to the real-time temperature detection value, with an adjustment range of ±20% of the set power. If the temperature deviation exceeds the set threshold, an audible and visual alarm is triggered and the welding operation is suspended.

[0009] Optionally, in step S3, the output results of the two are combined through a weighted voting mechanism, including combining the output results of the YOLOv5 algorithm with the MLP electrical performance output results. A weighted voting mechanism is used, with the image detection weight accounting for 70% and the electrical performance analysis weight accounting for 30%. The output includes image detection results, electrical performance detection results and comprehensive evaluation results.

[0010] Optionally, the multi-layer composite protective coating in step S4 includes at least: The bottom layer is an epoxy coating bonded to the welding area, with a thickness of 10-30 microns; The middle layer is a polyurethane coating covering the bottom layer, with a thickness of 20-60 microns; The outer layer is a graphene-based layer forming the outer surface, with a thickness of 5-15 microns; The chemical bonds between the coating layers are formed through a step curing process.

[0011] Optionally, a multilayer composite protective coating is constructed in step S4, including cleaning the surface of the solder joint, evenly covering the soldering area with epoxy resin by spraying, coating a polyurethane coating on the epoxy resin coating by a dip coating process, evenly spraying nanographene on the polyurethane coating, and finally performing a curing treatment.

[0012] To achieve the above-mentioned purpose, the second aspect of the present application provides a welding system for railway signal cables based on deep learning, the welding system comprising: A welding module, wherein the welding module is used to select a silver-based solder as a welding material, and simultaneously weld a signal conductor and a shielding layer; fix the signal conductor at a welding position, and the shielding layer is evenly distributed on the periphery of the signal conductor and keeps a preset distance from the signal conductor, and the modified silver-based solder is arranged at the welding area, and covers the contact surface between the signal conductor and the shielding layer at the same time, and the flux is evenly covered on the welding area; high-precision local heating is performed on the welding area of ​​the signal conductor and the shielding layer, and the silver-based solder penetrates into the shielding layer through capillary action, and a uniform welding layer is formed at the welding area of ​​the signal conductor and the shielding layer at the same time; the welding area is rapidly cooled, and the cooling rate of the welding area of ​​the signal conductor and the shielding layer is kept consistent during the cooling process; A detection module, wherein the detection module detects the collected welding area image through the YOLOv5 algorithm, detects the collected welding area electrical performance data through the MLP algorithm, and determines whether each detection result meets the welding requirements. If so, the output results of each algorithm in step S2 are integrated through a weighted voting mechanism. If not, the welding module is made to melt the welding area to remove the solder, clean the signal conductor and the shielding layer, and then repeat the welding module operation; the integrated output result is evaluated to determine whether the evaluation score is lower than the preset score. If so, the welding area is melted to remove the solder, clean the signal conductor and the shielding layer, and then repeat the welding module operation. If not, the spraying module is made to work; A spraying module is used to spray a multi-layer composite protective coating from a bottom layer to an outer layer in a welding completion area.

[0013] To achieve the above-mentioned purpose, the third aspect of the present application provides a railway signal cable welding device based on deep learning, which is used to implement the above-mentioned method, including: A multi-axis positioning fixture device for positioning the cable with a positioning accuracy of ±0.1 mm; At least two sets of independently temperature-controlled laser heads for simultaneous welding of signal conductors and shielding layers, with a wavelength range of 900-1100 nanometers; A closed-loop temperature control system integrated with a high-speed data acquisition card is used to adjust the laser power of the laser system in real time to keep the temperature of each welding area within a preset threshold range, with a sampling frequency of 50-200 Hz; A detection module that integrates the YOLOv5 algorithm and the MLP algorithm to detect whether the welding area is obviously damaged or cracked; The coating spraying device comprises an electrostatic spraying unit and a gradient curing oven, which is used for spraying a multi-layer composite protective coating from a bottom layer to an outer layer on a welding completion area.

[0014] After adopting the above technical solution, the present application has the following beneficial effects compared with the prior art: In this application, from welding to coating construction and then to welding quality detection through deep learning algorithms, this application has formed a complete high-quality production process. Through intelligent welding system and closed-loop temperature feedback control, dynamic optimization of the welding process is achieved; through automated multi-layer coating construction technology, the uniformity and consistency of solder joint protection are ensured, and by using deep learning algorithms to detect welding quality, comprehensive welding quality traceability and improvement suggestions are provided; by synchronously welding signal conductors and shielding layers, efficient connection of signal conductors and shielding layers in the same process is achieved. Compared with traditional step-by-step welding processes, synchronous welding can not only simplify the process and improve production efficiency, but also reduce the heat-affected zone and effectively protect the material properties of the solder joints.

[0015] In this application, the YOLOv5 algorithm and the MLP algorithm are combined to achieve real-time detection of solder joint appearance defects such as cracks and pores and accurate analysis of electrical performance data such as resistance and voltage values; during the production process, the welding quality can be evaluated in real time, potential problems can be discovered and repaired in time, and the reliability and consistency of the solder joint quality can be ensured, completely solving the shortcomings of low efficiency and insufficient accuracy of manual inspection in traditional methods.

[0016] In this application, the inner epoxy resin coating provides adhesion and basic anti-corrosion protection, the middle polyurethane coating enhances flexibility and wear resistance, and the outer nanographene coating improves conductivity and long-term corrosion resistance. The hierarchical design takes into account the conductivity, mechanical protection and long-term corrosion resistance of the solder joints, which is particularly suitable for the use of railway signal cables in complex environments such as high humidity and high corrosion; compared with traditional single-layer coating, the multi-layer coating in this application significantly improves the protective performance of the solder joints.

[0017] The specific implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are part of this application and are used to provide a further understanding of this application. The illustrative embodiments of this application and their descriptions are used to explain this application, but do not constitute an improper limitation on this application. Obviously, the drawings described below are only some embodiments. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] In the drawings of the specification: Figure 1 It is a schematic diagram of the overall welding process in this specific implementation mode; Figure 2 is a schematic diagram of welding a signal conductor and a shielding layer by a laser head in this specific implementation mode; Figure 3 is a schematic flow chart of multi-stage temperature control in this specific implementation mode; Figure 4 is a flow chart of step 4 in this specific implementation mode; Figure 5 is a structural schematic diagram of a railway signal cable welding system in this specific implementation mode; Figure 6 It is a structural schematic diagram of the railway signal cable welding equipment in this specific implementation manner. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. The following embodiments are used to illustrate the present application but are not used to limit the scope of the present application.

[0021] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0022] To achieve this, see Figures 1 to 4 , the present application provides a welding method for railway signal cables based on deep learning, the welding method comprising the following steps: Step S1, performing synchronous welding operation on the signal conductor c and the shielding layer d, fixing the cable by a multi-axis positioning fixture, using a laser system with integrated temperature feedback to perform welding operation, using at least two groups of high-power laser heads to act on the welding areas of the signal conductor c and the shielding layer d respectively, so as to achieve synchronous welding of the signal conductor c and the shielding layer d; Pre-place silver-based solder and apply fluorine-containing flux at the welding interface of the signal conductor c and the shielding layer d, and realize gradient heating through multi-stage temperature control, including preheating stage, main heating stage, constant temperature maintenance stage and cooling stage; The laser power of the laser system is adjusted in real time through a closed-loop temperature control system to keep the temperature of each welding area within the preset threshold range; Step S2, welding area quality detection, using the YOLOv5 algorithm to detect the collected welding area image, and using the MLP algorithm to detect the collected welding area electrical performance data, to determine whether each detection result meets the welding requirements. If so, continue to step S3, if not, melt the welding area to remove the solder, clean the signal conductor and the shielding layer, and then execute step S1; Step S3, after synthesizing the output results of each algorithm in step S2 through a weighted voting mechanism, the synthesized output results are evaluated to determine whether the evaluation score is lower than a preset score. If so, the soldering area is melted to remove the solder, the signal conductor and the shielding layer are cleaned, and step S1 is repeated. If not, step S4 is executed; Step S4, constructing a multi-layer composite protective coating from the bottom layer to the outer layer in the welding completion area.

[0023] See also Figure 1 and Figure 2In practical applications, the laser system in step S1 includes at least a first laser head a and a second laser head b, wherein the first laser head a acting on the welding area of ​​the signal conductor c has an output power range of 20-40W and a working distance of 3-8mm; the second laser head b acting on the welding area of ​​the shielding layer d has an output power range of 15-25W and a working distance of 3-8mm; the first laser head a and the second laser head b are both integrated with infrared temperature measurement modules with a temperature detection accuracy within ±2°C.

[0024] Specifically, two first laser heads a and two second laser heads b are provided, and the vertical distance between each laser head and the welding surface is controlled at 3-5mm to ensure that the spot diameter is ≤0.5mm; the first laser head a is responsible for the signal conductor c welding point, with an output power of 30W and a wavelength of 1 micron; the second laser head b is responsible for the shielding layer d welding area, with an output power of 20W and a wavelength of 1 micron. Each laser head is integrated with an independent infrared temperature sensor with an accuracy of ±1°C to monitor the soldering point temperature in real time. The sampling frequency of the infrared temperature sensor is 100Hz.

[0025] See also Figure 1 , Figure 2 and Figure 3 In one feasible implementation, the multi-stage temperature control in step S1 includes: a preheating stage, a main heating stage, a constant temperature maintenance stage, and a cooling stage. In the preheating stage, the temperature of the signal conductor c interface is raised to 180-220°C and the temperature of the shielding layer d interface is raised to 140-160°C by the laser system; in the main heating stage, the temperature of the signal conductor c interface is raised to 630-720°C and the temperature of the shielding layer d interface is raised to 590-660°C by the laser system; in the constant temperature maintenance stage, the temperature of each solder joint is maintained within ±10°C by the laser system; in the cooling stage, the gas convection cooling method is adopted, and the solder joint is evenly cooled by a high-pressure fan, and the cooling rate is not less than 30°C / s.

[0026] Please continue to see Figure 3 In the preheating stage, within 0 to 5 seconds, the first laser head a is started and the signal conductor c welding surface is preheated to 200°C with an output power of 10W. At the same time, the second laser head b is started and the shielding layer d welding surface is preheated to 150°C with an output power of 5W. During this process, the flux is activated to remove surface oxides.

[0027] Specifically, 5 to 25 seconds into the main heating stage, the power of the first laser head a will be increased to 30W to heat the welding surface of the signal conductor c so that its target temperature reaches 650 to 700°C, and the output power of the second laser head b will be increased to 20W to heat the welding surface of the shielding layer d so that its target temperature reaches 600 to 650°C. At this time, the solder will melt synchronously at the contact surface of the signal conductor c and the shielding layer d and penetrate into the tinned copper wire braided layer through capillary action. At this stage, the temperature control system of the multi-point laser heater will monitor the temperature changes of the solder joints in real time through thermal sensors. If the solder joint temperature exceeds the threshold of plus or minus 10°C, the power of each corresponding laser head will be automatically adjusted by plus or minus 5W.

[0028] Specifically, within 25 to 30 seconds of the constant temperature maintenance stage, the power of all laser heads is reduced to 50% of the initial value to maintain the solder joint temperature stable within the range of plus or minus 5°C to ensure sufficient diffusion of the solder; finally, in the cooling stage, that is, within 30 to 35 seconds, all laser heads are turned off and a high-pressure fan with a wind speed of 5 meters per second is started to evenly cool the solder joint and the cooling rate is 50°C per second. At the same time, the fixture is kept fixed during the cooling period to prevent the solder joint from shifting.

[0029] See also Figure 1 and Figure 2 In another feasible embodiment, the closed-loop temperature control system in step S1 has a temperature sampling frequency of no less than 50 Hz for each welding area, and the closed-loop temperature control system adjusts the laser power in real time, including establishing a three-dimensional mathematical model of the temperature field in the welding area, and dynamically adjusting the laser power of the first laser head a and the second laser head b according to the real-time temperature detection value, with an adjustment range of ±20% of the set power. If the temperature deviation exceeds the set threshold, an audible and visual alarm is triggered and the welding operation is suspended.

[0030] Specifically, the closed-loop temperature control system includes an infrared temperature sensor, a central controller and a power regulation module. The central controller receives data from the infrared temperature sensor and executes a PID control algorithm to analyze the deviation between the temperature and the target value. The power regulation module dynamically adjusts the output power of the corresponding laser head according to the analysis results.

[0031] Specifically, the closed-loop temperature control system in step S1 adopts high-precision temperature sampling technology, and the temperature sampling frequency of each welding area is 100Hz to ensure the real-time and accuracy of the temperature data. The closed-loop temperature control system realizes continuous monitoring of the temperature field in the welding area by integrating thermal imaging sensors and high-speed data acquisition modules. In terms of temperature regulation, the closed-loop temperature control system adopts an intelligent adjustment strategy based on the fuzzy PID control algorithm. By establishing a three-dimensional mathematical model of the temperature field in the welding area, the temperature distribution gradient is accurately calculated, and the laser power output of the first laser head a and the second laser head b is dynamically adjusted according to the real-time temperature detection value. The adjustment amplitude is accurately controlled within the range of ±20% of the set power.

[0032] Specifically, the closed-loop temperature control system first constructs a three-dimensional heat conduction model of the welding area through the finite element analysis method, and establishes a temperature field prediction model by combining the thermophysical parameters of the material and the laser energy distribution characteristics. In the actual operation process, the system compares and analyzes the real-time collected temperature data with the model prediction value. When it is detected that the temperature deviation exceeds the preset threshold, the closed-loop temperature control system will immediately activate the multi-level response mechanism, first fine-tuning the output power of each laser head. If the temperature deviation continues to expand, the sound and light alarm device will be triggered, and a pause command will be sent at the same time to immediately stop the welding operation.

[0033] In an achievable implementation, the material of the shielding layer d is a tinned copper wire braided layer, the shielding layer d is tensioned, and the silver-based solder is filled into the contact gap of the tinned copper wire braided layer. The shielding layer d uses an optimized tinned copper wire braided structure to ensure that the electrical performance and mechanical strength of the connection between the shielding layer d and the conductor solder joint meet high standards.

[0034] It should be noted that copper itself has excellent conductivity, and tinning can further reduce contact resistance, improve welding performance, and has excellent welding compatibility with silver-based solder. The copper wire braided structure has high flexibility and tensile strength, and can remain stable in a long-term vibration environment. The tinning layer effectively protects the surface of the copper wire, prevents oxidation and corrosion, and improves the durability of the shielding layer d. Tinned copper wire has a moderate cost and is suitable for large-scale industrial applications.

[0035] See also Figure 1 and Figure 2 In a feasible implementation, in step S2, the collected welding area image is detected by the YOLOv5 algorithm, and the collected welding area electrical performance data is detected by the MLP algorithm to determine whether each detection result meets the welding requirements. If so, continue to step S3. If not, melt the welding area to remove the solder, clean the signal conductor and the shielding layer, and then execute step S1.

[0036] In practical applications, the collected welding area images are detected by the YOLOv5 algorithm to determine the small external defects in the welding area, including preprocessing the input welding area images to adapt to the input format of the YOLOv5 algorithm; the features of the input image are extracted through the YOLOv5 backbone network and a feature pyramid is generated. The backbone network adopts a lightweight convolutional neural network architecture; the multi-level features extracted by FPN and PANet are fused to output the bounding box, confidence and category probability of each target; the final output results include the location (coordinates of cracks or pores) and category (cracks, pores, etc.) of the weld defect, as well as the confidence score of each prediction.

[0037] Specifically, the YOLOv5 algorithm receives the solder joint image as input, and after passing through the backbone network, feature fusion module and detection head, it outputs the bounding box, confidence and category probability of each target. In order to reduce redundant predictions, YOLOv5 uses the non-maximum suppression algorithm (NMS) to remove overlapping bounding boxes and retain the results with the highest confidence. The final output of YOLOv5 includes the location (coordinates of cracks or pores) and category (cracks or pores) of the solder joint defect and the confidence score of each prediction. Based on the detection results generated by YOLOv5, it is judged whether the welding area meets the welding requirements.

[0038] It should be noted that the welding requirements herein are not specifically limited, and technicians in this field set the allowable standards for external defects and internal defects required for welding according to actual needs.

[0039] In one feasible implementation, the collected electrical performance data of the welding area is tested by the MLP algorithm, and the internal defects such as smaller bubbles and cracks in the welding area can be detected, and the resistance and voltage values ​​of the welding area are analyzed by the MLP algorithm. The MLP algorithm is a fully connected neural network. In real-time detection, the electrical performance data of the welding point, such as the resistance and voltage values, are input into the trained MLP algorithm, and the MLP algorithm generates a predicted value such as a regression result of resistance or voltage through forward propagation. The generated prediction result can be directly compared with the standard range to determine whether the electrical performance of the welding point meets the welding requirements.

[0040] In another feasible implementation, in step S2, the collected welding area image is detected by the YOLOv5 algorithm, and the collected welding area electrical performance data is detected by the MLP algorithm. If each detection result meets the welding requirements, the output results of the two are combined through a weighted voting mechanism to determine whether the evaluation score is lower than the preset score. If so, the welding area is melted to remove the solder, and the signal conductor c and the shielding layer d are cleaned, and then step S2 is executed; if not, step S4 is executed.

[0041] In practical applications, the output results of the two are integrated through a weighted voting mechanism, including combining the output results of the YOLOv5 algorithm with the electrical performance output results of the MLP algorithm. A weighted voting mechanism is adopted, with the image detection weight accounting for 70% and the electrical performance analysis weight accounting for 30%. The output includes image detection results, electrical performance detection results and comprehensive evaluation results, and the location and size of cracks and pores are marked on the solder joint image.

[0042] In one achievable implementation, the YOLOv5 algorithm assigns a score to each detection result based on the detected target category and confidence score. When a "welding defect" is detected, the score is low; when a "good solder joint" is detected, the score is high. The confidence score is directly used as the weight of the score. The higher the confidence, the more reliable the score. MLP is used to predict the electrical performance data of the welding area, and the predicted value output by it is used as the evaluation score.

[0043] Specifically, if the score is lower than the preset score, the welding area is judged to be unqualified, the welding area is melted to remove the solder, the signal conductor c and the shielding layer d are cleaned, and the process returns to step S1 to re-collect data and perform testing. If the score is higher than the preset score, the welding area is judged to be qualified, and step S4 is executed.

[0044] In another feasible implementation, before returning to execute step S1, the process further includes using a hot air gun to melt the solder joint, using a solder sucker to remove the solder, and then using isopropyl alcohol to clean the signal conductor c and the shielding layer d.

[0045] In a feasible embodiment, the welding area of ​​the signal conductor c forms a continuous metallurgically bonded silver-based solder layer, the penetration depth reaches more than 50% of the woven structure of the shielding layer d, the interface of the shielding layer d presents a gradient transition alloying area, the microstructure grain size is less than 10 microns, and the dielectric loss tangent value of the protective coating system on the outer surface at a frequency of 1 MHz is less than 0.01.

[0046] In practical applications, the silver-based solder layer contains 50-70% Ag, 10-20% Cu, 5-15% Sn and 0.5-2% rare earth elements by mass percentage. The rare earth elements include yttrium, lanthanum and cerium. The porosity of the solder layer is less than 5%, the Vickers hardness value is in the range of 80-120HV, and the wetting angle with the copper conductor is less than 15°.

[0047] It should be noted that silver-based solder not only has the highest conductivity, which can significantly reduce the resistance loss during signal transmission and ensure the signal stability of railway signal cables, but also has a moderate melting point (about 650-800°C), which is convenient for precise temperature control welding. The combination of silver and copper gives the solder a higher fatigue resistance and adapts to the complex vibration and mechanical shock environment of the railway. Adding zinc or tin can further improve the corrosion resistance and extend the life of the solder joint. The addition of rare earth elements can improve the wettability of the interface between the solder and the base material and enhance the interface bonding strength. The addition of manganese can improve the oxidation resistance and toughness of the solder, especially for high vibration environments.

[0048] In practical applications, the multilayer composite protective coating in step S4 includes at least the following from the inside out: a bottom layer is an epoxy resin coating bonded to the welding area, with a thickness of 10-30 microns; a middle layer is a polyurethane coating covering the bottom layer, with a thickness of 20-60 microns; an outer layer is a graphene-based layer forming the outer surface, with a thickness of 5-15 microns; and chemical bonds are formed between each coating through a step curing process.

[0049] It should be noted that nanographene has extremely high conductivity, which can ensure that the surface of the solder joint has good conductivity, does not affect the quality of signal transmission, and provides excellent corrosion resistance to prevent oxidation of the solder joint surface. The polyurethane coating provides flexible protection to meet the performance requirements of railway signal cables under vibration and deformation, and enhances the wear resistance of the outer graphene coating to extend the service life of the coating. The epoxy resin coating provides the interface bonding force between the solder joint and the coating, which can prevent the coating from peeling off and provide a solid foundation protection for the middle layer and the outer layer.

[0050] See also Figure 1 , Figure 2 and Figure 4 In a feasible implementation, in order to meet the high performance requirements of the railway signal cable welding points, it is necessary to spray a multi-layer composite protective coating from the bottom layer to the outer layer in the welding completion area, including: cleaning the welding point surface, and evenly covering the welding area with epoxy resin by spraying, and the thickness of the epoxy resin coating is controlled to be 10-30 microns; using a dip coating process, a polyurethane coating is coated on the epoxy resin coating, and the thickness of the polyurethane coating is controlled to be 20-60 microns; nano-graphene is evenly sprayed on the polyurethane coating, and the thickness of the nano-graphene coating is controlled to be 5-15 microns, and finally a curing treatment is performed.

[0051] Specifically, before applying the coating, ensure that the surface of the solder joint is clean and remove impurities that affect the adhesion of the coating. Use a non-corrosive cleaning fluid such as isopropyl alcohol or a water-based cleaning agent to clean the soldering area to remove any residual flux that may remain during the soldering process. Use a hot air gun or a drying oven to dry the cleaned soldering area to ensure that there is no moisture or solvent residue on the surface. Spray the epoxy resin to evenly cover the soldering area. The coating thickness is controlled to be 10-30 microns to ensure that the solder joint is completely covered and tightly bonded to the parent material. Use a dip coating process to apply a polyurethane coating on the epoxy resin coating, and the coating thickness is controlled to be 20-60 microns to ensure uniform coverage without bubbles or gaps. Use high-precision spraying equipment such as an electrostatic sprayer to evenly spray nanographene on the polyurethane coating. The coating thickness is controlled to be 5-15 microns to ensure that the graphene is evenly distributed. Finally, a curing treatment is performed, and the graphene coating is baked in a gradient curing oven at 80 degrees Celsius to 100 degrees Celsius for 1 hour to ensure that the graphene coating is firmly attached.

[0052] See also Figure 1 , Figure 2and Figure 5 Based on the same inventive concept, the present application also provides a welding system for railway signal cables based on deep learning, the welding system comprising: The welding module is used to select a silver-based solder as a welding material, and simultaneously weld the signal conductor c and the shielding layer d; fix the signal conductor c at a welding position, and the shielding layer d is evenly distributed on the periphery of the signal conductor c and keeps a preset distance from the signal conductor c, and the modified silver-based solder is arranged at the welding area, and the contact surface of the signal conductor c and the shielding layer d is covered at the same time, and the flux is evenly covered on the welding area; the welding area of ​​the signal conductor c and the shielding layer d is locally heated with high precision, and the silver-based solder penetrates into the shielding layer d through capillary action, and a uniform welding layer is formed at the welding area of ​​the signal conductor c and the shielding layer d at the same time; the welding area is rapidly cooled, and the cooling rate of the welding area of ​​the signal conductor c and the shielding layer d is kept consistent during the cooling process; The detection module detects the collected welding area image through the YOLOv5 algorithm, and detects the collected welding area electrical performance data through the MLP algorithm to determine whether each detection result meets the welding requirements. If so, the output results of each algorithm in step S2 are integrated through a weighted voting mechanism. If not, the welding module is made to melt the welding area to remove the solder, clean the signal conductor and the shielding layer, and then repeat the welding module operation; the integrated output result is evaluated to determine whether the evaluation score is lower than the preset score. If so, the welding area is melted to remove the solder, clean the signal conductor c and the shielding layer d, and then repeat the welding module operation. If not, the spraying module is made to work; The spraying module is used to spray a multi-layer composite protective coating from the bottom layer to the outer layer in the welding completion area.

[0053] See also Figure 1 , Figure 2 and Figure 6 Based on the same inventive concept, the present application also provides a welding device for railway signal cables based on deep learning, which is characterized in that it is used to implement the above method, including: Multi-axis positioning fixture device, used to position the cable, with a positioning accuracy of ±0.1 mm; At least two sets of independently temperature-controlled laser heads are used to achieve synchronous welding of the signal conductor c and the shielding layer d, with a wavelength range of 900-1100 nanometers; A closed-loop temperature control system with an integrated high-speed data acquisition card is used to adjust the laser power of the laser system in real time to keep the temperature of each welding area within the preset threshold range, with a sampling frequency of 50-200 Hz; A detection module that integrates the YOLOv5 algorithm and the MLP algorithm to detect whether the welding area is obviously damaged or cracked; The coating spraying device comprises an electrostatic sprayer and a gradient curing oven, which are used for spraying a multi-layer composite protective coating from a bottom layer to an outer layer on a welding completion area.

[0054] The above are only preferred embodiments of the present application, and are not intended to limit the present application in any form. Although the present application has been disclosed as above with preferred embodiments, it is not intended to limit the present application. Any technician familiar with the present application can make some changes or modifications to equivalent embodiments of equivalent changes using the above-mentioned technical contents without departing from the scope of the technical solution of the present application. The implementation schemes in the above embodiments can also be further combined or replaced. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present application without departing from the content of the technical solution of the present application are still within the scope of the solution of the present application.

Claims

1. A welding method for railway signal cables based on deep learning, characterized in that: The welding method comprises the following steps: Step S1, performing a synchronous welding operation on the signal conductor and the shielding layer, fixing the cable by a multi-axis positioning fixture, using a laser system with integrated temperature feedback to perform a welding operation, and using at least two groups of high-power laser heads to act on the welding areas of the signal conductor and the shielding layer respectively, so as to achieve synchronous welding of the signal conductor and the shielding layer; Pre-place a silver-based solder and apply a fluorine-containing soldering flux at the welding interface between the signal conductor and the shielding layer, and implement gradient heating through multi-stage temperature control, including a preheating stage, a main heating stage, a constant temperature maintenance stage, and a cooling stage; The laser power of the laser system is adjusted in real time by a closed-loop temperature control system to keep the temperature of each welding area within a preset threshold range; Step S2, welding area quality detection, using the YOLOv5 algorithm to detect the collected welding area image, using the MLP algorithm to detect the collected welding area electrical performance data, to determine whether each detection result meets the welding requirements, if so, continue to step S3, if not, then melt the welding area to remove the solder, clean the signal conductor and the shielding layer, and then execute step S1; Step S3, after synthesizing the output results of each algorithm in step S2 through a weighted voting mechanism, the synthesized output results are evaluated to determine whether the evaluation score is lower than a preset score. If so, the soldering area is melted to remove the solder, the signal conductor and the shielding layer are cleaned, and step S1 is repeated. If not, step S4 is executed; Step S4, constructing a multi-layer composite protective coating from the bottom layer to the outer layer in the welding completion area.

2. The method according to claim 1, characterized in that The laser system in step S1 includes: A first laser head acting on the signal conductor welding area has an output power range of 20-40W and a working distance of 3-8mm; The second laser head acting on the welding area of ​​the shielding layer has an output power range of 15-25W and a working distance of 3-8mm; The first laser head and the second laser head are both integrated with an infrared temperature measurement module with a temperature detection accuracy within ±2°C.

3. The method according to claim 2, characterized in that The multi-stage temperature control in step S1 includes: A preheating stage, in which the temperature of the signal conductor interface is raised to 180-220° C. and the temperature of the shielding layer interface is raised to 140-160° C. by the laser system; A main heating stage, in which the temperature of the signal conductor interface is raised to 630-720° C. and the temperature of the shielding layer interface is raised to 590-660° C. by the laser system; A constant temperature maintenance stage, wherein the laser system is used to maintain the temperature of each soldering point within a stable fluctuation range of ±10°C; In the cooling stage, a gas convection cooling method is adopted in the cooling stage, and the solder joints are evenly cooled by a high-pressure fan, and the cooling rate is not less than 30°C / s.

4. The method according to claim 2, characterized in that: The closed-loop temperature control system in step S1 has a temperature sampling frequency of no less than 50 Hz for each welding area. The closed-loop temperature control system adjusts the laser power in real time, including establishing a three-dimensional mathematical model of the temperature field in the welding area, and dynamically adjusting the laser power of the first laser head and the second laser head according to the real-time temperature detection value. The adjustment range is ±20% of the set power. If the temperature deviation exceeds the set threshold, an audible and visual alarm is triggered and the welding operation is suspended.

5. The method according to claim 1, characterized in that In step S3, the output results of the two are combined through a weighted voting mechanism, including combining the output results of the YOLOv5 algorithm with the output results of the MLP electrical performance. A weighted voting mechanism is adopted, with the image detection weight accounting for 70% and the electrical performance analysis weight accounting for 30%. The output includes image detection results, electrical performance detection results and comprehensive evaluation results.

6. The method according to claim 1, characterized in that The multi-layer composite protective coating in step S4 comprises at least: The bottom layer is an epoxy coating bonded to the welding area, with a thickness of 10-30 microns; The middle layer is a polyurethane coating covering the bottom layer, with a thickness of 20-60 microns; The outer layer is a graphene-based layer forming the outer surface, with a thickness of 5-15 microns; The chemical bonds between the coating layers are formed through a step curing process.

7. The method according to claim 6, characterized in that In step S4, a multilayer composite protective coating is constructed, including cleaning the surface of the solder joint, uniformly covering the soldering area with epoxy resin by spraying, coating the epoxy resin coating with a polyurethane coating by a dip coating process, uniformly spraying nanographene on the polyurethane coating, and finally performing a curing treatment.

8. A welding system for railway signal cables based on deep learning, characterized in that: The welding system comprises: A welding module, wherein the welding module is used to select a silver-based solder as a welding material, and simultaneously weld a signal conductor and a shielding layer; fix the signal conductor at a welding position, and the shielding layer is evenly distributed on the periphery of the signal conductor and keeps a preset distance from the signal conductor, and the modified silver-based solder is arranged at the welding area, and covers the contact surface between the signal conductor and the shielding layer at the same time, and the flux is evenly covered on the welding area; high-precision local heating is performed on the welding area of ​​the signal conductor and the shielding layer, and the silver-based solder penetrates into the shielding layer through capillary action, and a uniform welding layer is formed at the welding area of ​​the signal conductor and the shielding layer at the same time; the welding area is rapidly cooled, and the cooling rate of the welding area of ​​the signal conductor and the shielding layer is kept consistent during the cooling process; A detection module, wherein the detection module detects the collected welding area image through the YOLOv5 algorithm, and detects the collected welding area electrical performance data through the MLP algorithm to determine whether each detection result meets the welding requirements. If so, the output results of each algorithm in step S2 are integrated through a weighted voting mechanism. If not, the welding module is made to melt the welding area to remove the solder, clean the signal conductor and the shielding layer, and then repeat the welding module operation; after the output results of the two are integrated through a weighted voting mechanism, the integrated output result is evaluated to determine whether the evaluation score is lower than a preset score. If so, the welding area is melted to remove the solder, clean the signal conductor and the shielding layer, and then repeat the welding module operation. If not, the spraying module is made to work; A spraying module is used to spray a multi-layer composite protective coating from a bottom layer to an outer layer in a welding completion area.

9. A welding device for railway signal cables based on deep learning, characterized in that: The method for implementing any one of claims 1 to 7 comprises: A multi-axis positioning fixture device, used to position the cable, with a positioning accuracy of ±0.1mm; At least two sets of independently temperature-controlled laser heads for simultaneous welding of signal conductors and shielding layers, with a wavelength range of 900-1100 nanometers; A closed-loop temperature control system integrated with a high-speed data acquisition card is used to adjust the laser power of the laser system in real time to keep the temperature of each welding area within a preset threshold range, with a sampling frequency of 50-200Hz; A detection module that integrates the YOLOv5 algorithm and the MLP algorithm to detect whether the welding area is obviously damaged or cracked; The coating spraying device comprises an electrostatic spraying unit and a gradient curing oven, which is used for spraying a multi-layer composite protective coating from a bottom layer to an outer layer on a welding completion area.

Images