| Along with the development of science and technology, high-tech products toward miniaturization and integration in the direction of development in these products, 1 mm in diameter, following the application of more and more porous, its widely used in automotive,electronic,tiny mechanical instrument,Instrument,fiber and fluid control technology, and other areas. Micro-size in the various processing methods, the drilling will be better because of the processing quality and higher production efficiency, so dominant at home and abroad. But drilling there is a fatal weakness Micro-size tiny bit of low intensity, wear or fracture easily after the chip plug; and the great dispersion of its life, the same batch number of drilling bit before breaking up differences more than 10 times, difficult to estimate the life of the bit. Once broken drill bit, it will be very difficult removed from the work-piece, often leads to the work-piece scrapped. The existing line is automatically according to the actual processing time or the number of work-piece machining tool change the way a compulsory, even in the drill fell far short of the average life span before they had to be replaced, resulting in a great waste production. Therefore the establishment of micro drilling on-line monitoring system, real-time alerting tool change is of great significance.Drilling of the increase is the root cause of a broken drill bit, but in obtaining drilling (such as axial force or torque) signal, Dynamometer, often installed in the spindle or work platform, and implementation will be very complicated. The need for the existing equipment often the main axis of the work-piece clamping system or mechanical structure be major changes, the destruction of the processing system stiffness, to some extent, limit its application. In view of the present micro drilling mostly low voltage, high current of the spindle motor, the drilling of the changes will be a direct result of the spindle motor current changes, this paper to three-phase current of the spindle motor subject to monitoring by the spindle motor 3 Current changes in indirect bit wear reflect the changes in the state. By the spindle motor current monitoring tool condition of the existing equipment does not need to do more changes, only a slight change to the spindle motor power lines, and implementation will be relatively simple and convenient.This paper developed a porous drilling spindle motor current smart on-line monitoring system. The system consists of hardware and software system. LabVIEW software system is software for the programming environment, and the voltage signal of the collection, filtering, storage, display, control and microcontroller serial communication system.Drilling in the process of porous, it is necessary to carry out on-line monitoring, it is necessary to establish signal characteristics and wear bits of the state of relations between the two models. As drilling, the state bit of wear and tear there nonlinear, uncertainties, it is difficult to establish signal characteristics and bit wear the exact state of relations between the mathematical model. Neural network control as important in the field of intelligent control of intelligent control methods, the need to know the precise object model, through study of the object that is able to control. Therefore, this paper, the establishment of monitoring neural network models, using neural network access to state information bits wear, and the signal characteristics of the nonlinear mapping. BP neural network in the design of the learning process can be seen in existence less than the following: First, the study convergence rate is too slow and often requires hundreds or even thousands of times to study the convergence Second, no established network structure, often determined based on experience , Implied floors and implied the selection of nodes there is no theoretical guidance. Because of the existence of subjective factors, so the network can not achieve optimal selection and monitoring system to affect the accuracy. To address these issues, the paper proposed that neural networks and fuzzy control combine to establish five-story structure of the fuzzy neural network, the network structure based on input and output of the nodes identified, each floor has a specific meaning, structure clear and easy to understand people . Through experiments contrast, in accordance with fuzzy neural network set up to monitor the threshold, to better prevention of broken bits. But fuzzy neural network has its deficiencies, that is, when the fuzzy neural network system of importation of high-dimension, the fuzzy rules will be a geometric progression increase, leading to a huge network structure, the system's real-time worse. Therefore, the simplification of fuzzy neural network structure, improving network study and on the reaction speed is of great significance. Based on this, this paper, the rough set of neural networks and fuzzy combine, using rough set of samples extracted from the training, streamlining of rules set rules on the removal of redundant information. Construction of these rules fuzzy neural network so that the network structure more simple, easy to understand, reduce the network of training time and increase the efficiency of the network of learning.Finally the porous drilling spindle motor three-phase current on-line monitoring experiment, the monitoring process as: real-time acquisition will be the main motor of the corresponding three-phase current voltage signal input to structure a good rough sets fuzzy neural network, network with a given output Monitoring the threshold for comparison, if the output is less than the threshold will continue to drilling, if the threshold is greater than the output value alarm retreat knife. The results showed that select the appropriate monitoring threshold to avoid a broken drill bit, drill raise the utilization rate of a certain effect.In this paper, the main innovations are as follows:1. Neural networks and fuzzy control porous on online monitoring of microdrilling. Design of a five-story structure of the fuzzy neural network model, a bit fuzzy neural network signal characteristics of the wear and mapping, using fuzzy neural network implied tiny bit real-time access to state information wear. A monitoring system to solve mathematical model to establish the problem.2. Rough sets fuzzy neural network model used on online monitoring of microdrilling. Application of rough set method of extracting fuzzy rules, rules on the removal of redundant information, thus simplifying the network structure, reduce the network of training time.3. Multi-sensor data fusion used on online monitoring of microdrilling. Make up for a single sensor to obtain information, such as the accuracy and reliability of the shortage.
|
PDF Full Text Request