Quantitative Identification Of Broken Wire Damage Of Wire Rope Based On Magnetic Leakage Principle

As a ferromagnetic component,steel wire rope has the characteristics of strong bearing capacity,light weight,and local small defects will not cause sudden breaks.It is widely used in ports,coal mines,lifting and other fields.Its working environment is harsh,and it will inevitably occur after a long time of use Defects such as wear,fatigue,and rust eventually lead to broken wires or sudden breaks,which will pose a major threat to human safety.This thesis aims to detect the broken wire defects in the steel wire rope,and uses embedded ARM technology,pattern recognition and neural network to detect and quantitatively identify the broken steel wire wire.The main research and engineering practices have been completed:1.The form of wire rope breakage damage is introduced,the formation mechanism of wire rope defect leakage magnetic field is analyzed,and a simulation model of wire rope leakage signal based on magnetic charge theory is constructed,which provides a theoretical basis for subsequent analysis and identification of wire breakage damage signal.2.The wire rope magnetization device is designed,and the detection principle of the Hall sensor and its arrangement position in the wire rope magnetization device are introduced.The software and hardware platform of the steel wire rope detection system based on STM32 was built,and the performance of each circuit module of the system was tested and analyzed.The system adopts equal space sampling to avoid the influence of the detection speed,and at the same time record the location of the wire rope defect,the designed detection system works stably and reliably.3.The frequency spectrum characteristics of the wire rope defect signal are studied,and two methods for suppressing the strand signal are proposed.Aiming at the characteristics that the amplitude of the stranded wave signal is large but the frequency band is narrow,a notch filter is designed to eliminate the stranded wave signal;the wavelet analysis method is studied,the Mallat algorithm is used to decompose the wire rope defect signal,and the threshold is set to reconstruct.In the end,the real noise signal is highlighted while filtering out the noise.After filtering the noise,the time-frequency features such as amplitude and wave shape of the broken wire signal are extracted,and the characteristic values are normalized,which lays the foundation for the construction of a quantitative identification model of wire rope broken wire damage.4.The classification principles of BP neural network and support vector machine(SVM)are analyzed.Two quantitative models of steel wire broken wire damage identification based on BP neural network and support vector machine are constructed respectively.The two steel wire broken wire damage identification models are compared and analyzed the performance.Aiming at the problem that the initial weights and thresholds of the BP network and the penalty coefficient and kernel function parameters in the support vector machine classification model are random,an improved adaptive particle swarm optimization algorithm is proposed to optimize the model parameters.The results show that the improved particle swarm optimization algorithm has fast optimization speed and high accuracy.Both models constructed can identify the number of broken wire ropes with high accuracy,and the SVM model has better performance in classifying small samples of broken wire damage.