Research On The Breakage Mechanism Of Micro-drills And Drilling Forces Online Monitoring

With the development of high-tech products toward miniaturization, integration and precision, the number of micro-holes has greatly increased. It widely used in aviation, spaceflight, automotive, electronic, chemical industry, medics, fiber and fluid control technology, and other areas. At present, among various methods used for producing micro-holes such as high-speed drilling, laser beam machining, electron beam machining, electric discharge machining, etc., Micro-drilling(drill diameter is smaller than 1mm) is widely applicable for obvious reasons such as cost, efficiency, reliability. But micro-drilling has a fatal weakness decided by micro-drill structure characteristics, that of low strength and rigidity, wear or break easily after the chip plug. Because of the great dispersion of their life of the micro-drills, the number of drilling holes before breaking up are difficult to be estimated. Once drills broken, they will be very difficult to remove from the work-piece, often lead to the work-piece scrapped. So, to micro-drilling, it is one of puzzles to guard against breakage in effect. Micro-drilling meets the serious challenge. Therefore researching on the breakage mechanism of micro-drills, and designing the micro-drilling on-line monitoring system, real-time alerting and changing tools are of great significance.In the interest of avoiding breakage and rising the lives of micro-drills, many specialists and scholars at home and abroad have much research. Some new technics have been figured out, example for high speed drilling, multi-stations drilling, step feed drilling, vis-a-vis drilling, vibration drilling ,and so on. However, there have little research on the breakage mechanism of micro-drills.By analytic demonstrations at three aspects of the situation including observing the microscopic breaks, computing the limit stresses and stress cycle times, the conclusion was maken that the breakage mechanism of micro-drills should be subject to static strength failure. Using a great deal of experimental data into the micro-drills'stress model based on the third strength theory, the paper computed their contribution rates to limit stress while micro-drills close to breakage about drilling thrust, torque and lateral force, and drew a conclusion that increasing torque is the crucial factor influencing the failure of micro-drills, second is the lateral force, the last is thrust.Thrust and torque are essential, inevitable to sustainable drill processing, but the lateral force mainly from original entering drilling offset is a adverse factor to location accuracy of entering drilling, bending deformation. The effects of the structure errors, run-out error of micro-drills and the gradient error of the workpiece surface among many factors on the location error of entering drilling measuring the lateral force were mainly analyzed theoretically and verified experimentally. Their contribution rates to the location error of entering drilling or the bending stress of micro-drills were computed and analyzed comparatively, the paper thinks that the effect of run-out error of micro-drills is most remarkable, next is the gradient error of the workpiece surface, the third is the structure errors of micro-drills. Furthermore, the dynamic modes of micro-drills during the three segments:before penetrating, entrance and drilling were calculated by finite element simulating, and the bending stress, the critical load, the stucture stability effectted by the bending vibration modal parameters, the spindle-speed and the original entering drilling offset were anlyzed.On-line monitoring micro-drilling may real-time monitor wear state of micro-drills, in order to effectively prevent their breakage. But it is difficult to establish the accurate mathematical model between signal characteristics and wear state of micro-drills,because of the nonlinearities and uncertainties of wear state of micro-drills. Neural network control, as an important intelligent control methods, is able to control the object by network studying and training, even though the accurate mathematical model of objects haven't been known. In this paper, their maximum amplitudes of drilling thrust and torque as static characteristic parameters were figured out by time-domain analysis of signals, its second, third and sixth layer power maximum amplitudes of drilling torque as dynamic characteristic parameters were figured out by time-frequence-domain wavelet decomposing analysis of signals, the input eigenvectors of neural network were consisted of the static and dynamic characteristic parameters. The wavelet neural network model built by combined loosely with wavelet transform and neural network to on-line monitoring micro-drilling, has fully used wavelet's good local property in time-frenquence-domain and character extracting function. The input signal numbers were reduced, trainning time was shortened, system identifying ability was improved greatly by using this model.Some disadvantages of the wavelet neural network caused that the network performance and the monitoring accuracy could not achieve optimal, such as selecting the numbers of implicit layer or implicit layer node is no theoretical guidance, only according to experience, secondly the model don't have fixed network structure,moreover, the study convergence rate is too slow. To overcome these issues, the paper proposed to build the wavelet fuzzy neural network monitoring model by combining fuzzy control with wavelet neural network. The network structure based on input nodes and output nodes is immobile, definite and easily understandable, its every layer has a specific meaning. It has showed in compared experiments that the wavelet fuzzy neural network model has better performance than the wavelet neural network model in monitoring micro-drilling and micro-drills.This paper developed a on-line monitoring micro-drilling software system to monitor drilling force signals based on virtual instrument technology. The system may realize these functions, including signals collecting, data processing, signals real-time displaying, historical data reappearing, network decision-making and spindle feed motion servo controlling. Finally the on-line monitoring experiments have been carried out by using the network model. Its monitoring process is that the real-time data will be input into the trainned network model, next, by constrast the network output to the given monitoring threshold, micro-drills will continue to drill while the output is less than the threshold, or else micro-drills will been alertted to draw back while the output is greater than or equal the threshold. The results showed that selecting appropriate monitoring threshold could effectively avoid micro-drills breakage. So monitoring micro-drilling using the wavelet fuzzy neural network is feasible and avaiable.In this paper, the main innovations are as follows:1. Researched on the breakage mechanism of micro-drills. By observing the microscopic breaks, computing the limit stresses and stress cycle times, the conclusion was maken that the breakage mechanism of micro-drills should be subject to strength failure, while its strength reachs the limit of micro-drills' material. Computed their contribution rates to limit stress while micro-drills breakage about drilling thrust, torque and lateral force. It is a conclusion that the torque increasing was the crucial factor that influences the failure of micro-drills, second was the lateral force, the last was thrust.2.Static and dynamic characteristics parmeters were redrawed to represent the wear state of micro-drills by the methods including wave analysis in time-domain, specta analysis in frequency-domain and wavelet analysis in time-frequency-domain. These characteristics parmeters may reflect the paratical state of micro-drills fully and closely, and may remedy defects which a single characteristics parameter can't accomplish the pattern recognition exactly.3.Applied wavelet transform, fuzzy control and neural network into on-line monitoring micro-drilling. A wavelet fuzzy neural network model was designed to model between signal characteristics and wear state of micro-drills, and to obtain implicit information on wear state of micro-drills. The method solved the difficult problem of building the monitoring model.