G. van Schoor

A comparative study on fault detection and correction techniques on active magnetic bearing systems

In this paper, the authors distinguish between three real-time fault detection, correction and identification schemes for vibration forces on the rotor of a rotational active magnetic bearing (AMB) system. Historical fault data obtained from a fully suspended 250 kW water cooling AMB pump was used in the design process of the three schemes. The real-time schemes perform three main tasks: 1) fault detection, 2) fault diagnosis and error correction and 3) fault identification. Displacement and current masking were performed during the fault detection stage and the vibratory amplitudes and frequencies were extracted by means of the Wigner-Ville distribution. Pattern recognition techniques, statistical diagnosis and fuzzy logic were used to calculate fault features during the fault diagnosis and error correction stages. During the fault identification stage, data fitting, fuzzy logic and ISO standards were used to calculate the type, parameters, vibratory level and zone of the vibration force. A comparison between the experimental results obtained from a double radial AMB test rack was performed to demonstrate the effectiveness of the proposed schemes in the real-time detection, correction and identification of vibration forces on the rotor of a rotational AMB system. The three real-time schemes were able to correct and minimize vibration forces to a stable working condition.

Real-time displacement analysis and correction system for vibration forces on rotor of rotational active magnetic bearing system

The real-time displacement analysis and correction of vibration forces on the rotor of a rotational active magnetic bearing (AMB) system were performed. The real-time system performs four main tasks: fault detection, fault diagnosis, error correction and fault identification. Historical fault data from a fully suspended 250 kW water cooling AMB pump was used in the design process of the fault detection system. A double radial AMB test rack was used to test the effectiveness of the proposed real-time system. The real-time system corrected, identified and minimized the vibration forces to a stable working condition.

A non-linear simulation model of an active magnetic bearings supported rotor system

Nowadays active magnetic bearings (AMBs) are utilized in a wide range of applications including high-speed flywheel energy storage devices. This paper aims at reporting on the modeling and simulation of a flywheel energy storage system supported by AMBs. Firstly the non-linear AMB model is discussed that comprises the power amplifiers, displacement sensors, anti-aliasing filters and controller. The results of the non-linear AMB model are compared to conventional linear AMB models. Secondly the rotor model is discussed. The rotor model was implemented by means of the finite element method and the simulated results compared to a commercial rotor-dynamic software package. The rotor model is then integrated with the AMB models and the system is simulated under step disturbance inputs at the bearing positions as well as a run-up test to determine the unbalance response of the rotor.

Linear model of a closed three-shaft Brayton cycle

A linear model of a closed three-shaft Brayton cycle is developed. The model is intended to give an understanding of the dominant dynamic behavior in a nuclear power system that utilizes a closed three-shaft Brayton cycle. The insights gained from the model can be used for system design as well as for the design of control algorithms [S.T. Glad et al. (2001)] [B.Hendrisson et al., (2001)]. Linear models of the turbines and compressors are developed, followed by models of the shafts and the volumes inside the circuit. Thereafter, the system equations are derived from the component models and put into state space format. Finally, a time domain comparison is made between the linear system model and the responses predicted by the thermal-fluid analysis software program, Flownex.

Sizing of renewable energy hydrogen systems

This paper explains the simulation and sizing optimization of renewable energy systems for hydrogen production. These systems are complex and challenging to size correctly due to the nonlinearities of the components and the stochastic nature of the sources. The same system will perform completely different at two separate sites. The simulation includes renewable energy sources, storage elements and loads. Matlab® and Simulink− are the tools utilized for the simulation and optimization. The simulation and optimization of the Renewable Energy (RenEn) system, to be constructed in Potchefstroom, South Africa, is described and some of the results obtained are discussed. The sizing of components is essential to ensure that the most efficient and cost-effective system is specified.

Data acquisition system for capturing dynamic transients in a control valve

This work presents the development of a data acquisition (DAQ) system to be used in conjunction with fuzzy modelling to derive a dynamic mathematical model of a control valve. Characterising the behaviour of a control valve or any other mechanical device, either requires expert knowledge of the device and its subcomponents or real-world data thereof. Capturing real-world data not necessarily requires expert device knowledge but requires some from of post-analysis with either neural networks or fuzzy logic systems to derive a mathematical model. In order to capture valve specific physical data, different tests need to be conducted with the intent of capturing transient specific data embedded within the temperatures, mass flow rate, pressures and corresponding valve opening. The dynamic mathematical model will be used in a simulator to predict the behaviour of the device within the system. An accurate DAQ system and modelling procedure needs to be in place to ensure the development of reliable mathematical models of such mechanical devices. Once all relevant data have been acquired it can be used to derive a mathematical model for the control valve which includes static, dynamic, choke and nonchoke characteristics.

Adaptive Neural network control of a helicopter system with optimal observer and actor-critic design

Abstract This paper proposes a methodology for developing an adaptive neural network controller for a simulated helicopter system. Since an indirect adaptive neural network framework is chosen, the controller comprises three interconnected neural networks called the observer, actor and critic. The actor and critic networks rely on the observer network responsible for state estimation. The main contribution of this paper is the development of an observer that has fast convergence capabilities in order to be used in a completely on-line stability control scheme. This improved convergence is obtained by uniquely modifying the observer network structure and update law. The observer parameters are also optimised by means of a genetic algorithm (GA) for improved performance. The developed observer is firstly evaluated on a first principle linear model and then on actual test flight data of an attack helicopter. The results indicate excellent performance in terms of the state estimation capability of the observer. Lyapunov’s direct method is used to derive update laws for both the critic and actor networks and the control parameters of these networks are also optimised by means of a multi-objective GA. Actual data from a wind-tunnel test set-up were used for controller evaluation purposes.

Genetic Algorithm based PID Tuning for Optimal Power Control of a Three-shaft Brayton Cycle based Power Conversion Unit

Abstract This paper considers the development of a PID control strategy to optimally control the power output of a High Temperature Gas-cooled Reactor (HTGR) power plant. A specific type of HTGR called the Pebble Bed Modular Reactor (PBMR) that utilises a closed recuperative Brayton cycle with helium as working fluid is considered. The power control of this kind of plant is significantly different from conventional steam cycle nuclear power plants. A distinguishing feature that complicates the control is the use of three separate shafts for different compressor/turbine or turbine/generator pairs. In addition the power output cannot be directly controlled by means of an upstream valve that regulates the flow through the power turbine, as is the case with conventional steam cycles. This paper addresses these challenges by means of a control strategy consisting of four PID control loops. The controller gains are optimised by means of a Genetic Algorithm (GA) that uses real-valued genes and the ITAE performance measure as a cost function. The control strategy is implemented and evaluated on a linear Simulink® model of the PBMR Power Conversion Unit (PCU). Results are presented illustrating the performance of the GA optimised PID control strategy.

Action Evolution for Intelligent Agents

This paper introduces an alternative method of reinforcement learning for intelligent agents. The aim with this method is to closer emulate the natural thought processes of problem solving. This paper only provides a conceptual description of this method. However, it puts forward realistic implementations with exciting new implications, using proven techniques. After a brief discussion of reinforcement learning, the newly suggested method is described. Reinforcement learning methods can be divided according to two main strategies. The first strategy uses evolutionary methods while the second makes use of incremental back-propagation methods. An overview is given of the methods used for implementing the popular actor-critic architecture in these strategies. This leads to a generic architecture for intelligent controllers. The alternative method of action reinforcement is then demonstrated at the hand of this architecture. This representation of the controller emphasizes some cognitive attributes and encourages the development of advanced intelligent controllers