Research On Detection And Quantitative Identification Of Broken Wires Of The Steel Wire Rope

Due to the advantages of high strength,light and easy bending,not easy to break suddenly,stable and reliable operation,steel wire ropes have important applications in mine hoisting,cable-stayed bridge,crane,lifting and other scenes.Steel wire ropes are usually used as the main load-bearing components and work in harsh environment,which will inevitably lead to wire breakage,wear,corrosion and other faults.These defects will affect the safety of production and even threaten the life of workers.As the main damage of the wire rope during service,broken wire is an important factor affecting the safe operation of the wire rope.Many countries and organizations regard the number of broken wires in a lay distance as the main index to evaluate whether the wire rope needs to be replaced.Therefore,it is of great theoretical significance and practical value to carry out quantitative identification research on broken wires of wire ropes for evaluating service state of wire ropes and ensuring production safety.In this paper,based on the summary and analysis of the current situation of domestic and foreign research,the research topic of detection and quantitative identification of broken wires of rope wires is proposed,and an in-depth and systematic theoretical,simulation and experimental research on quantitative identification of broken wires of wire ropes is conducted.(1)In order to explore the distribution characteristics of magnetic field inside the wire rope and the relationship between the magnetic field distribution and the structural parameters of exciter.The excitation effect of exciter with different magnet positions and excitation modes on the steel wire rope is studied by finite element simulation.It is proved that the multi loop circumferential uniform excitation structure with magnets at both ends is conducive to the saturation magnetization of wire rope,and the surface of the wire rope can form a uniform magnetization section which is suitable for detection.Based on the equivalent magnetic circuit calculation and simulation analysis,the optimal design of the exciter is carried out,which provides a theoretical basis and practical method for the design and manufacture of the exciter.(2)Aiming at the weak output signal of traditional Hall array sensor,a design idea of sensor based on the magnetic concentrating principle is proposed.The magnetic flux leakage(MFL)concentration effect of the magnetic concentrating sensor is studied and the influence of the angle between the damage and the sensor is studied by finite element method,which proves that the magnetic concentrating sensor can improve the intensity of the MFL detected by the Hall element and is not easily affected by the detection angle.A magnetic concentrating sensor with two magnetic collecting rings and two magnetic bridges for detecting broken wires of the steel wire rope is designed.The detection effects of the two sensors under different lift-off are compared and analyzed by simulation.The results show that the magnetic concentrating sensor can effectively improve the intensity of MFL detection signal.(3)In order to further verify the performance of the sensor and the simulation analysis results,a large number of broken wire detection experiments are carried out.The broken wires with different numbers,positions,diameters and fracture lengths were detected and analyzed.The damage signals collected by the two sensors are compared to verify the simulation results,which shows that the magnetic concentrating sensor can collect the MFL more comprehensively.For different types of broken wires,the signal intensity detected by the magnetic concentrating sensor is greater than that of the Hall array sensor,especially for internal broken wires and the broken wires with different fracture lengths,the signal detected by the magnetic concentrating sensor is better differentiated,which provides reliable and stable damage signals for the quantitative identification of broken wires.(4)Aiming at the limitation of manual feature extraction and selection,an adaptive feature extraction method based on convolution neural network(CNN)for broken wire signals of the steel wire rope is proposed.The continuous wavelet transform is introduced to transform the MFL signals of broken wires into time-frequency images.Fault features are automatically extracted from the time-frequency images by CNN,and gradually integrated and optimized into features suitable for classification.The t-SNE algorithm is used to visualize the features extracted from different broken wire signals by CNN,and compared with the artificial features.It is proved that the adaptive feature extraction method based on CNN has better damage identification effect than the traditional artificial feature extraction method.(5)To address the problem of low accuracy and generalization of currently available quantitative recognition models for broken wires,a quantitative identification method of broken wires of the steel wire rope based on CNN is proposed.The time-frequency images converted from MFL signals of broken wires is used as the input of the CNN.The CNNs with different structural parameters is tested,and the optimal CNN identification model is established.The experiments show that the model can accurately distinguish different kinds of broken wires,especially realizing the high-precision identification of internal broken wires of wire ropes with different diameters.At the same time,in order to solve the problem of quantitative identification of broken wires under the condition of small sample,a quantitative identification model of broken wires of the steel wire rope based on transfer learning is studied and established.The low-level parameters of the pre-training network are directly transferred,and the high-level parameters are adjusted and optimized according to the training set,which realize the migration application from natural images to the MFL time-frequency images of broken wires.The depth migration model is validated with small sample wire breaking data of different diameters and different types,which proves that the model can effectively solve the problem of quantitative identification of broken wires of wire ropes in the case of small sample.