Nonlinear Model and Control of Electro Hydraulic Servo-Systems

In many applications, the use of hydraulic drives is still preferable to other driving powers. For instance, in shaking table systems for simulating earthquake signal, hydraulic actuators are still widely applied, because the technology of electrical actuators does not(yet) provide the superior performance of hydraulic actuators in generating high power to weight ratio. However, with increasing demands on the performance of complex motion systems, the limits of performance of hydraulic servo-systems, due to the nonlinear and dynamic characteristics of these systems, come into the picture. Most nonlinearities of these systems arise from compressibility of the hydraulic fluid, the complex flow properties of the servo-valve, valve overlap and friction in the hydraulic cylinder. Aside from the nonlinear nature of the hydraulic dynamics, there are many considerable model uncertainties, such as internal and external leakages and external disturbances which cannot be modeled exactly. Therefore, in order to design a high performance controller for simulating the earthquake signal which is the goal of a servo-hydraulic shaking table, a suitable dynamical model of the system needs to be formulated. In this thesis beside a specific application of hydraulic actuator, a shaking table servo system, an integrated approach to the modeling and control of a hydraulic servo system is presented. The application of servo-hydraulic shaking table shows that the dynamics and nonlinearities of the servo-valve and the compressibility of hydraulic oil which basically constitute the limits of the performance of the controlled servo-system. Especially, in case of acceleration model and control of a hydraulic servo-system. In the shaking table application, for modeling and identification of the system, in contrary of the previous works which in them position and pressure sensors are used, only the position and acceleration sensor are available. In order to obtain structural insight in the way that the performance is limited by the properties of (the subsystems of) the hydraulic servo-system, the modeling of this system has been treated thoroughly in this thesis. At the one hand, this has opened the way to model-based control design, so that unavoidable limits of performance can be narrowly approached. At the other hand, the obtained insight appears to be useful in the system design stage, such that potential control problems may be avoided by proper system design. Because of the twofold purpose of the modeling, with control design requiring quantitatively accurate models and simulating the behavior of the system precisely which requiring qualitative insight in the system behavior, the so-called grey-box modeling approach has been applied. This approach comprises physical modeling including model analysis by means of simulation, and subsequent identification and validation of the obtained physical models, using experimental data. In the physical modeling stage, a consistent integration of the nonlinear dynamic modeling of the different subsystems of the hydraulic servo-system, namely servo-valve and actuator are presented. In this model, the most nonlinearities of the system which arise from compressibility of the hydraulic fluid, the complex flow properties of the servo-valve, valve overlap and friction in the hydraulic cylinder are simulated. Four different kinds of friction model are considered and the accuracy of these models for simulating the behavior of the system are compared experimentally. However, due to the limitation of these models in high velocity and bandwidth, it has been shown that for simulating the behavior of the system, nonlinear modeling of the friction is not enough. Then, by gathering some position and acceleration information of the real system, the sensitivity of the model to different nonlinearity of the system are investigated. This led to the insight, that only some of the modeled nonlinear effects are really relevant, such as the nonlinear flow characteristic of the servo-valve spool due to non-ideal port geometries and the compressibility of hydraulic oil, and the position dependence of the actuator dynamics. Based on the experimental results, two new nonlinear dynamic models for simulating the behavior of the servo-hydraulic shaking table are proposed. First, with defining 6 main parameters of the model and identifying them for different sinusoidal inputs, a neural network model is proposed. Second, a new empirical nonlinear model for effective bulk modulus of hydraulic oil is proposed which increase the accuracy of the model to predict the behavior of the position and acceleration output of the system. In this approach, the link between the physical and the system theoretic interpretation of the properties of the hydraulic servo system are strongly emphasized. This makes, that the presented models are not only useful for shaking table design, but also for the design of the hydraulic servo-system. Based on the task specification of the shaking table which is tracking the position and velocity reference signals with considering uncertain load conditions, different kinds of robust controller design are presented. For control designing purpose, the full order dynamic model of the system is simplified in a new approach and then due to the availability of only the position sensor on the experimental setup, a robust sliding mode observer is designed which can estimate the velocity and acceleration states of the system from the position sensor. Finally, experiments with a hydraulic actuator in a single degree-of-freedom setup have shown the validity of the approach for control design. The experimental results of the closed loop system controlling by three different controllers (feed forward PI controller, sliding mode and super twisting controller) with considering different load condition and reference signals are presented. An analysis of different control strategies for this setup led to the conclusion, that super twisting sliding mode controller shows a smaller error and smother response for position and velocity trajectory tracking with respect to sliding mode and feedforward PI controller.