Paul Haase Sørensen

Implementation of neural network based non-linear predictive control

Abstract This paper describes a control method for non-linear systems based on generalized predictive control. Generalized predictive control (GPC) was developed to control linear systems, including open-loop unstable and non-minimum phase systems, but has also been proposed to be extended for the control of non-linear systems. GPC is model based and in this paper we propose the use of a neural network for the modeling of the system. Based on the neural network model, a controller with extended control horizon is developed and the implementation issues are discussed, with particular emphasis on an efficient quasi-Newton algorithm. The performance is demonstrated on a pneumatic servo system.

On condition monitoring of exhaust valves in marine diesel engines

The feasibility of noninvasive characterisation of exhaust valve conditions in large marine diesel engines were experimentally investigated on a four cylinder 500 mm bore 2-stroke marine diesel engine at MAN B&W Diesels Research Center in Copenhagen, Denmark. The experiments comprised three different valve conditions, where two were concerned with artificially induced valve burn-through situations. The primary potential monitoring measurements were vibration and structure-borne stress waves, also known as Acoustic Emission (AE). Results have shown that the AE signals have a major advantage over other involved sensors, indicating sensitivity to both mechanical and fluid-mechanical combustion related activity. Recorded data has been preprocessed and features extracted using principal component analysis (PCA). From a number of applied heuristics and statistics, a search for the optimal sub-space of principal components to use, have been carried out. The chosen feature-space has been used for classification of involved exhaust valve conditions by applying both regularized feedforward neural classifiers and linear discriminators. The complexity of the neural networks have further been optimized by optimal brain damage pruning, leading to increased generalization.

Linear Control System Design

In this chapter a review of the design of multivariable feedback controllers for linear systems will be considered. This review treats mainly deterministic control objects with deterministic disturbances. After giving an overview of the type of linear systems to be treated, this chapter will handle the basic control system design method known as pole or eigenvalue placement. First systems where measurements of all the states are available will be treated. For cases when such complete state measurements are not available the concept of deterministic observers to estimate the states which are not measured directly will be introduced. It will also be shown that it is often possible to design reduced order observers where only the unmeasured states are estimated.

Optimal Observers: Kalman Filters

This chapter has the purpose of reviewing the most important design aspects of Kalman filters as well as some of their most important properties. Heuristic derivations are given of the Kalman filter `equations for both continuous time and discrete time dynamic systems. It is shown that the state mean values propagate according to the same observer equations as given in Chap. 4. Moreover it is shown that the state noise propagates according to the time dependent Lyapunov equation derived in Chap. 6. When measurements are made on the system this equation has to be modified with a term which expresses the decrease of uncertainty which the measurements make possible. The combination of these two results yields the main stochastic design equation for Kalman filters: the Riccati equation. Solving this equation immediately gives the optimal observer gain for a Kalman filter. Combining a Kalman filter with optimal or LQR feedback results in a very robust controller design: the LQG or Linear Quadratic Gaussian regulator.

Experimental Determination of the Static Characteristics of Hydraulic Motors and Pumps

Abstract In this paper an experimental system for identifying reliable and accurate steady-state characteristics of hydrostatic machines for hydraulic drives is presented. The core of the experimental system consists of two unique components: A servo controlled dynamometer for 4-quadrant operational modes and a servo controlled volume rate flowmeter. Both components have been developed at the Technical University of Denmark. First, a short review of the state-of-the-art and a discussion of analytical and experimental modelling of the steady-state characteristics for hydraulic motors and pumps are given. Then the concept, control and implementation of the developed dynamometer system are described. Further, the flowmeter principle is explained and its specifications are stated. Finally, a typical result obtained by the experimental system is presented.

Noise in Dynamic Systems

The purpose of this chapter is to present a brief review of the salient points of the theory of stochastic processes which are relevant to the study of stochastic optimal control and observer systems. Starting with a brief review of the main properties of random variables, this chapter goes forward to a detailed description of the main random process which is used as a model for noise in technological systems: Gaussian white noise. Both time domain and frequency domain descriptions of this important noise model are given. The main result of these considerations is the time dependent Lyapunov equation which is a compact way to express how white noise propagates through linear dynamic systems. Both continuous time and discrete time versions of this equation are given.

State Space Modelling of Physical Systems

Modelling of state space models based on relevant physical laws is introduced. Linearization of nonlinear models is discussed and the connection between the transfer function model and the state space model is derived. Discrete time models are also introduced.

Analysis of State Space Models

In this chapter an overview of the properties of the state space models will be given. A basis for the investigation of these properties is the solution of the state equation given appropriate boundary conditions. The important notions of stability, controllability and observability will be introduced and the similarity transformation discussed. This makes possible the construction of state space models with a number of useful properties.

Active Suspension for a Field Sprayer Boom

Abstract The possibilities of implementing an active boom suspension is investigated. Compared with a traditional passive suspension, it could improve the spraying result, and make the operators task much easier. In a simulation is compared the possible performance of both an active and a passive system. The simulated passive suspension is an advanced construction, that combines a traditional trapezoid with a spring pendulum system. The active suspension is combining a hydraulic actuator and a spring. According to the simulations, the performance is improved significantly. An attempt has been made to implement the active suspension on a full-size field sprayerboom provided with the hydraulics of a running tractor. A closed loop controlsystem with two ultrasonic sensors, a proportional valve and a hydraulic actuator has been constructed. In the article is described how the system has been build, and what changes should be made for future experiments.

CAD AND IMPLEMENTATION OF DIGITAL CONTROLLERS FOR THE FAST TUD-HYDRAULIC TEST ROBOT MANIPULATORS

A very fast digital control system implemented with a AT&T signalprocessor DSP32 which has been developed to the fast TUD-hydraulic test robot manipulator installed in the hydraulic laboratory at the Control Engineering Institute, the Technical University of Denmark (TUD) is presented. Results and experiences with CAD/CAE-based tools for design and implementation of digital controller algorithms for hydraulic robot manipulators and other multiaxes hydraulic machines are presented and discussed. Furthermore, a new design concept applying the AT&T-signal processor DSP32C is discussed. Finally, the paper describes and discusses the simulation results with an adaptive perturbation control scheme and the obtained experimental results with a digital PID controller.