Hydraulic systems are widely used in production systems and processes due to their high energy density, robustness, and reliability. Often, the dynamic and nonlinear characteristics of hydraulic press drive components complicate the understanding of their components on system and components. This doctoral dissertation presents a methodology for developing a virtual model of the hydraulic press drive through modeling and simulation, with a focus on regression modeling. This enables accurate predictions of hydraulic press drive behavior under varying intensities of the bending process and changes in the friction characteristics of the components. Qualitative and quantitative comparative analyses of the hydraulic press drive models are based on experimental results ("what-if" scenarios) in real environment. While simulation models are less accurate but satisfactory, regression models show greater reliability and predictive capability for the hydraulic press drive behavior. Intelligent algorithms for selfrecognition and selfadaptation of the control signal, part of the adaptive control mechanism, are integrated into the expert system. The use of response surface methodology and polynomial regression modeling proves most effective, reducing the response error of the hydraulic press drive by up to 94%. These methods are validated in both simulation and real environments of the hydraulic press drive.
|
