Design And Application On Control System For Technology Of Ultrasonic Vibration Aided Electrical Discharge Grinding Process

Ultrasonic vibration aided grinding and electrical-discharge machining technology is a new developed combined techonology. It integrates three machining methods, including ultrasonic vibration machining, electric-discharge machining and grinding. By adjusting parameters, it can increase machining efficiency and improve. surface quality. Grinding is not only an effective method for difficult-to-machine materials but also the main method for achieving good surface roughness. Its material removal rate could be increased when admitting electrical-discharge machining. Grinding force and temperature may be reduced when admitting ultrasonic vibration in the process, and loading of grinding wheel takes place lessly, and it insures more stable performance of electrical discharge machining. This combined machining techonology can be widely used for processing conducting materials. It is not influenced by the properties of workpiece materials, such as hardness, brittleness, toughness. This dissertation studies this ternary compound research and its intelligent control, including making better hardware configuration, developing control software system and setting up the model of prediction appreciate machining parameters for better surface integrity. The research contents are as follows.Based on abundant literature and experimentation, ultrasonic vibration aided grinding and electrical-discharge machining mechanism and the interaction between them were investigated. Based on the analysis of variance (ANOVA) and response surface method, the influence and tendency about main technology parameters and their mutual interactions to surface roughness were analyzed systemically. Rearch results show: voltage reference is the most influenced parameter, and impulse width,impulse interval, ultrasonic amplitude next. Their mutual interactions influence distinctly on surface roughness. The analysis result of variance (ANOVA) is matched with the result of response surface method. This research provided reliable basis for predicting machining parameters intelligently, laid the foundation of the whole control design solutions.The control system adopted multi-CPU control technology including industrial PC, PMAC and Single-Chip. It structured modularized and open numerical control hardware system and used lots of intelligent technologies such as case-based reasoning, artificial neural network, fuzzy control, data base and expert system to make these three machining methods more coordinate gradually and gain more highly intelligent control system. After that, it gave an improvement to the hardware configuration, customizing one special tool and special fixture, which makes the structure of this open NC system better.The control software of ultrasonic vibration aided grinding and electrical discharge machining was developed by using Visual C++. It has the friendly human-computer interface and modularized design, which are easier for user and maintenance. In this software, the communication between PC and PMAC is implemented by PComm32 drive, which carries out the monitoring of process state and adjustment of some parameters. Serial control achieves the communication between PC and Single-Chip, which is used for the acquisition of grinding-force and short-circuit ratio signals to carry out the process control. This control interface contains parameters-input, parameters-output, condition monitoring, processing program, data bases and selective preference of parameters, machining procedure control, etc.Based on abundant experiments, lots of good sample dataset were obtained, and the database was built by SQL Server. As a result, it provided the data source for the prediction model about machining parameters based on case based reasoning and artificial neural network technology. According to the similarity theory, it tested matching grinding parameters with material physical properties and machining precision, etc. With these results and the known surface roughness, it can be predicted through BP neural network model which is structured by Visual C++, SQL Server and MATLAB. This model was transferred data to train to predict the results. It turned out that prediction is in the error 10% permitted range. Among them, ADO control achieved the communication between PC and Database. Likewise, engine technology achieved the interface between PC and MATLAB. ODBC drive achieved the extended interface between MATLAB and database.