论文部分内容阅读
本文用崭新的时间序列分析方法——动态数据系统(DDS)建模方法对电火花成型加工机床的EDM过程进行了系统辨识,先用Marple算法建立了自回归统计模型且进行了系统参数识别(估计),发现在不同工况如由正常火花放电向电弧放电转化时,模型阶数和再生的二次信息(系统阻尼率,标准方差等)有显著差异。又通过现场工艺试验确认了由正常火花放电百分率为代表的综合工艺指标的主要影响因素。 在上述两项工作的基础上,对EDM过程进行了计算机在线预报自适应控制研究。在IBM-PC/XT机上开发了8088汇编语言与8087协处理器的功能。采用时序模型AR(n),采取Durbin-leyenson算法(模型阶数由五阶降为二阶)仅由有限个输出观测值进行任意步预报递推,得到正常火花放电率的预报值,与其实测值比较,两者的趋势非常接近或吻合。对EDM过程采用适应控制的效果是:综合工艺指标可提高20％～30％。
In this paper, a new time series analysis method - Dynamic Data System (DDS) modeling method is used to systematically identify the EDM process of EDM machines. The autoregressive statistical model is first established by Marple algorithm and the system parameters are identified It is found that there is a significant difference between the model order and the secondary information of regeneration (system damping ratio, standard deviation, etc.) under different conditions such as normal spark discharge to arc discharge. And through the field test confirmed the percentage of normal spark discharge represented by the comprehensive process indicators of the main factors. On the basis of the above two work, the EDM process is studied on computer online forecasting adaptive control. Developed 8088 assembly language and 8087 coprocessor capabilities on IBM-PC / XT machines. Using the time series model AR (n), Durbin-leyenson algorithm (the order of the model is reduced from fifth order to second order) is only recursive by arbitrary step prediction with a limited number of output observations to obtain the predicted value of normal spark discharge rate, Value comparison, the trend of the two is very close or consistent. The effect of adapting control to the EDM process is that the comprehensive process index can be increased by 20% ~ 30%.