Font Size: a A A

Research On The High Speed And High Accuracy Fast Tool Servo Used In The Noncircular Turning Process

Posted on:2007-11-21Degree:DoctorType:Dissertation
Country:ChinaCandidate:H F WangFull Text:PDF
GTID:1101360185497277Subject:Mechanical Manufacturing and Automation
Abstract/Summary:PDF Full Text Request
The noncircular parts have been widely used in industry. As their shapes are be-coming more and more complicated, the requirements for machining accuracy are im-proved too, and the corresponding accurate high-efficient process technology evolvesas the research focus of the field gradually. Piston is the heart of an engine. To meetthe challenge of constant improvement in the engine performance, pistons are oftendesigned to be of middle convex and variable ellipse at present. The key point ofnumerically controlled turning process lies in two sides: one is the fast tool servo(FTS), which is of high frequency, long travel and high precision; the other is thecorresponding servo control algorithm of high performance. The subject investigatedin this paper is the PZT stack actuator (PSA) based FTS mechanism. Using thismechanism, several control problems are explored to make the noncircular turningprocess possible. Some experiments were carried out to validate the proposed controlalgorithm.According to the machining requirements of middle convex and variable ellipsepistons used in sedan cars, the main design parameters of FTS are determined, suchas position accuracy, frequency response, travel range, stiffness, driving force andseveral dynamic performance. Since the stroke of PSA is not long enough, the ?ex-ible hinge amplifier is popular in the PSA based FTS mechanism. For the controllerdesign, a more accurate model is needed. So, the swept sine method is used to ob-tain the frequency response of this FTS, and an autoregressive with exogenous input(ARX) model is utilized to identify the model parameters of FTS. The test results offrequency response and step response indicate that the used FTS has broader bandwidth and better follow-up performance. In the case of open-loop control, high-frequency and heavy-load applications, thenonlinear hysteresis of PSA is so serious that it can be up to 43% of the full stroke.Therefore, it is necessary to model the hysteresis to compensate feedforwardly. Toimprove the control precision of PSA, a dynamic generalized hybrid model is putforth, which is based on the Preisach principle and the support vector machine (SVM)algorithm. The linear spline kernel based SVM hybrid model can substitute the look-up table and complete the bilinear interpolation used in the classical Preisach model,but also can describe the nonlinear hysteresis characteristic of PSA when load andfrequency change in a large range. As a result, both the generalization ability and theprediction precision are improved obviously. Comparison test results show that theproposed SVM hybrid model is valid.In the turning process of middle convex and variable ellipse pistons, the referencetarget is known, and the feed movement of the lathe tool is of periodic and repetitive.Based on these characteristics, a modified plug-in repetitive controller is put forward.In accordance to the stability analysis, the design method of controller parameter isrecommended. Although the tracking error is enlarged when the load is varied, ahigher tracking control accuracy is obtained.Since the longitudinal feed motion to machine the middle convex shaped con-ductor, and the high-frequency reciprocating motion to shape the oval piston skirt canbe separated in the realization of numerically controlled turning process, it is possibleto build up a noncircular cutting pilot system on a centre lathe. In the end, a group ofaluminum alloy bar stocks are machined with different ovalities and distinct speeds,to validate the proposed control methods used for the PSA based FTS mechanismprimarily.
Keywords/Search Tags:fast tool servo, middle convex and variable ellipse piston, PZT stack actuator, ARX model, support vector machine, dynamic generalized hybrid model, repetitive control
PDF Full Text Request
Related items