Gerulf K.m. Pedersen

Multi-objective PID-controller tuning for a magnetic levitation system using NSGA-II

This paper investigates the issue of PID-controller parameter tuning for a magnetic levitation system using the non-dominated sorting genetic algorithm (NSGA-II). The magnetic levitation system is inherently unstable and the PID-controller parameters are hard to find using conventional methods. Based on four different performance measures, derived from the step response of the levitation system, the algorithm is used to find a set of non-dominated parameters for a PID-controller that can stabilize the system and minimize the performance measures.

The SURE_REACH Model for Motor Learning and Control of a Redundant Arm: From Modeling Human Behavior to Applications in Robotics

The recently introduced neural network SURE_REACH (sensorimotor unsupervised redundancy resolving control architecture) models motor cortical learning and control of human reaching movements. The model learns redundant, internal body models that are highly suitable to flexibly invoke effective motor commands. The encoded redundancy is used to adapt behavior flexible to situational constraints without the need for further learning. These adaptations to specific tasks or situations are realized by a neurally generated movement plan that adheres to various end-state or trajectory-related constraints. The movement plan can be implemented by proprioceptive or visual closed-loop control. This chapter briefly reviews the literature on computational models of motor learning and control and gives a description of SURE_REACH and its neural network implementation. Furthermore, we relate the model to human motor learning and performance and discuss its neural foundations. Finally, we apply the model to the control of a dynamic robot platform. In sum, SURE_REACH grounds highly flexible task-dependent behavior on a neural network framework for unsupervised learning. It accounts for the neural processes that underlie fundamental aspects of human behavior and is well applicable to the control of robots.

Efficiency optimization of a multi-pump booster system

This paper discusses a way to optimize the speed settings for a multi-pump system such that it can operate with highest efficiency. A short description of an actual multi-pump system is given and the mathematical formulas for single pump and multi-pump systems are presented. Then a set of objectives are formulated which, when used in a MATLAB toolbox implementation of NSGA-II, describe the most efficient distribution of speeds amongst the pumps in the system. The system is tested for a number of pressure references with good results.

Learning sensorimotor control structures with XCSF: redundancy exploitation and dynamic control

XCS has been shown to be an effective genetics-based classification, datamining, and reinforcement learning tool. The systems learns suitable, compact, maximally general problem solutions online. In the robotics and cognitive systems domains, however, applications of XCSF are very sparse and mostly restricted to small, symbolic problems. Recently, a sensorimotor XCSF system was applied to cognitive arm control. In this paper, we show how this XCSF-based armcontrol mechanisms can be extended (1) to efficiently exploit redundant behavioral alternatives and (2) to guide the control of dynamic arm plants. The XCSF system encodes redundant alternatives in its inverse control representations and resolves the encoded redundancies dependent on current constraints--such as arm posture preferences - on the fly. An adaptive PD controller translates the XCSF-based direction and distance commands into actual motor commands for dynamic arm control. We apply the complete system to the control of a simulated, physical arm with three degrees of freedom in a two-dimensional environment and to a simulation of the industrial KR16 Kuka arm with ODE-based physics engine.

Modeling and Control of Indoor Climate Using a Heat Pump Based Floor Heating System

Modeling and control of an indoor climate using a floor heating system based on a heat pump is discussed. The thermodynamic models of the heat pump, room, floor and circulated water are theoretically developed and then identified through experimental data. Several control strategies, namely relay control, P control, PID control and PID control with pre-filtering of reference input, are developed and simulated based on the developed system model. The simulation results turn out that the PID control with pre-filtering strategy generated the best system performance in terms of the smallest overshoot, fast response, highest COP in the transient period, and the smallest deviation and the least sensitivity to disturbance in the steady period. Meanwhile, this strategy also uses least power compared with other strategies. Instead of using the standard relay control, this study reveals a huge potential to use advanced control techniques to optimize operation of this kind of thermal system from thermal comfort and energy consumption perspectives.

Automatic tuning of PID controller for a 1-D levitation system using a genetic algorithm - a real case study

The automatic PID control design for a onedimensional magnetic levitation system is investigated. The PID controller is automatically tuned using the non-dominated sorting genetic algorithm (NSGA-II) based on a nonlinear system model. The developed controller is digitally implemented and tested. The preliminary simulation and test results show a bright potential to use artificial intelligence methods for supporting the control design for complicated nonlinear and open-loop unstable systems.

A Study of Rolling-Element Bearing Fault Diagnosis Using Motor's Vibration and Current Signatures

Abstract This paper investigates the fault detection and diagnosis for a class of rolling-element bearings using signal-based methods based on the motors vibration and phase current measurements, respectively. The envelope detection method is employed to preprocess the measured vibration data before the FFT algorithm is used for vibration analysis. The average of a set of Short-Time FFT (STFFT) is used for the current spectrum analysis. A set of fault scenarios, including single and multiple point- defects as well as generalized roughness conditions, are designed and tested under different operational conditions, including different motor speeds, different load conditions and samples from different operating time intervals. The experimental results show the powerful capability of vibration analysis in the bearing point-defect fault diagnosis under stationary operation. The current analysis showed a subtle capability in diagnosis of point-defect faults depending on the type of fault, severity of the fault and operational condition. The generalized roughness fault can not be detected by the proposed frequency methods. The temporal features of the considered faults and their impact on the diagnosis analysis are also investigated.

Controlling a four degree of freedom arm in 3D using the XCSF learning classifier system

This paper shows for the first time that a Learning Classifier System, namely XCSF, can learn to control a realistic arm model with four degrees of freedom in a three-dimensional workspace. XCSF learns a locally linear approximation of the Jacobian of the arm kinematics, that is, it learns linear predictions of hand location changes given joint angle changes, where the predictions are conditioned on current joint angles. To control the arm, the linear mappings are inverted--deriving appropriate motor commands given desired hand movement directions. Due to the locally linear model, the inversely desired joint angle changes can be easily derived, while effectively resolving kinematic redundancies on the fly. Adaptive PD controllers are used to finally translate the desired joint angle changes into appropriate motor commands. This paper shows that XCSF scales to three dimensional workspaces. It reliably learns to control a four degree of freedom arm in a three dimensional work space accurately and effectively while flexibly incorporating additional task constraints.

Dynamic Uniform Scaling for Multiobjective Genetic Algorithms

Before Multiobjective Evolutionary Algorithms (MOEAs) can be used as a widespread tool for solving arbitrary real world problems there are some salient issues which require further investigation. One of these issues is how a uniform distribution of solutions along the Pareto non-dominated front can be obtained for badly scaled objective functions. This is especially a problem if the bounds for the objective functions are unknown, which may result in the non-dominated solutions found by the MOEA to be biased towards one objective, thus resulting in a less diverse set of tradeoffs. In this paper, the issue of obtaining a diverse set of solutions for badly scaled objective functions will be investigated and the proposed solutions will be implemented using the NSGA-II algorithm.

Estimation of pump-curves using genetic algorithms

This paper presents a variety of different ways of estimating the general parameters for pump-curves. First a formulation is made that converts the problem into estimating four parameters in this general formulation. Then three different methods that utilize the general formulation are presented. The estimated parameters for each method are used in generating the desired pump-curves, and the quality of the estimates are determined. The paper finishes with a recommendation that, for estimation of the generalized parameters, the method utilizing speed differences is preferable over the other methods discussed in the paper.