No description available.

No description available.

This thesis presents an analysis of low speed sensorless control techniques for low voltage permanent magnet motor drives. The specific application of these drives is for power steering in small to medium sized cars. The focus of the project is to create a low speed sensorless drive with minimum current distortion, such that any additional torque ripple or audible noise is acceptable for the steering application. A comparison of two sensorless control methods is made, in terms of their benefits and drawbacks. The first method uses a measurement of the motor current derivative in response to the normally applied PWM switching vectors, to obtain a position estimate. The second method uses the high frequency current response to an injected high frequency voltage in the d-axis of the machine. Both methods track the variation of inductance in the machine due to local saturation as the machine rotates. The quality of the position estimate, current distortion and audible noise generated by the methods are investigated and compared. Experimental results show that the position quality of the current derivative method is better and is quieter than the d-axis method. However it produces a larger current distortion compared to the d-axis method. The d-axis method requires a large injection voltage (10% of rated) to be applied due to the poor quality of the estimated position signal below this voltage. For the current derivative method it was found experimentally that increas- ing the vector extension time (tmin) up to 1OJ.L3 gave an improved position es- timation performance. at the cost of increased current distortion and audible noise. For the d-axis method increasing the injection frequency from 1.5kHz to 2kHz causes a decrease in position estimation quality, an increase in audible noise but a decrease in current distortion.

This thesis investigates the principle of integrated vehicle dynamics control through proposing a new control configuration to coordinate active steering subsystems and dynamic stability control (DSC) subsystems. The active steering subsystems include Active Front Steering (AFS) and Active Rear Steering (ARS); the dynamic stability control subsystems include driveline based, brake based and driveline plus brake based DSC subsystems. A nonlinear vehicle handling model is developed for this study, incorporating the load transfer effects and nonlinear tyre characteristics. This model consists of 8 degrees of freedom that include longitudinal, lateral and yaw motions of the vehicle and body roll motion relative to the chassis about the roll axis as well as the rotational dynamics of four wheels. The lateral vehicle dynamics are analysed for the entire handling region and two distinct control objectives are defined, i.e. steerability and stability which correspond to yaw rate tracking and sideslip motion bounding, respectively. Active steering subsystem controllers and dynamic stability subsystem controller are designed by using the Sliding Mode Control (SMC) technique and phase-plane method, respectively. The former is used as the steerability controller to track the reference yaw rate and the latter serves as the stability controller to bound the sideslip motion of the vehicle. Both stand-alone controllers are evaluated over a range of different handling regimes. The stand-alone steerability controllers are found to be very effective in improving vehicle steering response up to the handling limit and the stand-alone stability controller is found to be capable of performing the task of maintaining vehicle stability at the operating points where the active steering subsystems cannot. Based on the two independently developed stand-alone controllers, a novel rule based integration scheme for AFS and driveline plus brake based DSC is proposed to optimise the overall vehicle performance by minimising interactions between the two subsystems and extending functionalities of individual subsystems. The proposed integrated control system is assessed by comparing it to corresponding combined control. Through the simulation work conducted under critical driving conditions, the proposed integrated control system is found to lead to a trade-off between stability and limit steerability, improved vehicle stability and reduced influence on the longitudinal vehicle dynamics.

This thesis has investigated the reduction of audible noise in low speed sensorless controlled drives for automotive electrical power steering (EPS) applications. The specific methods considered employ saliency tracking high frequency (hf) voltage injection in the machine's estimated d axis. In terms of the audible noise reduction, a novel random sinusoidal hf injection sensorless method has been proposed. The perceived audible noise due to the hf injection can be reduced by randomly distributing the injection frequencies around a centre frequency, such that it is perceived as a background hiss rather than the fixed tone heard with fixed hf injection methods. By analysing the A-weighting scales used to classify human perception of audible noise and frequency analysis of the recorded noise, an injection frequency of (lS00±328) Hz is found to have the lowest audible noise level compared to other random frequencies and other fixed frequencies methods. A 10 kHz square wave hf injection sensorless method has also been implemented. The frequency analysis of the recorded audible noise indicates that it also may be lower than for the fixed hf sinusoidal injection. In terms of control performance, sensorless torque control for these methods has been achieved from zero speed to ±240rpm with up to ±60A load (about 63% rated load). Similar position estimate quality has been demonstrated. Dynamic performance for a step change in torque current demand and for a speed reversal has been performed, and the random injection method with (1S00±328) Hz frequency has been found to be able to control a step change in torque demand current of 50A whilst for the 10kHz square wave injection method only a 40A step change can be achieved. On the other hand, the average position error after the speed transient has settled is less for the 10 kHz square ewave injection than for the random injection.