Stability analysis of a mono inverter dual sensorless motors system

When a single brush-less dc motor is fed by an inverter with a sensor-less algorithm embedded in the switching controller, the system exhibits a linear and stable output in terms of the speed and torque. However, with two motors modulated by the same inverter, the system is unstable and rendered useless for a steady application, unless provided with some resistive damping on the supply lines. The project discusses and analysis the stability of such a system through simulations and hardware demonstrations and also will discuss a method to derive the values of these damping.

Adaptive control of sinusoidal brushless DC motor actuators

Electrical Power Assisted Steering system (EPAS) will likely be used on future automotive power steering systems. The sinusoidal brushless DC (BLDC) motor has been identified as one of the most suitable actuators for the EPAS application. Motor characteristic variations, which can be indicated by variations of the motor parameters such as the coil resistance and the torque constant, directly impart inaccuracies in the control scheme based on the nominal values of parameters and thus the whole system performance suffers. The motor controller must address the time-varying motor characteristics problem and maintain the performance in its long service life. In this dissertation, four adaptive control algorithms for brushless DC (BLDC) motors are explored. The first algorithm engages a simplified inverse dq-coordinate dynamics controller and solves for the parameter errors with the q-axis current (iq) feedback from several past sampling steps. The controller parameter values are updated by slow integration of the parameter errors. Improvement such as dynamic approximation, speed approximation and Gram-Schmidt orthonormalization are discussed for better estimation performance. The second algorithm is proposed to use both the d-axis current (id) and the q-axis current (iq) feedback for parameter estimation since id always accompanies iq. Stochastic conditions for unbiased estimation are shown through Monte Carlo simulations. Study of the first two adaptive algorithms indicates that the parameter estimation performance can be achieved by using more history data. The Extended Kalman Filter (EKF), a representative recursive estimation algorithm, is then investigated for the BLDC motor application. Simulation results validated the superior estimation performance with the EKF. However, the computation complexity and stability may be barriers for practical implementation of the EKF. The fourth algorithm is a model reference adaptive control (MRAC) that utilizes the desired motor characteristics as a reference model. Its stability is guaranteed by Lyapunov’s direct method. Simulation shows superior performance in terms of the convergence speed and current tracking. These algorithms are compared in closed loop simulation with an EPAS model and a motor speed control application. The MRAC is identified as the most promising candidate controller because of its combination of superior performance and low computational complexity. A BLDC motor controller developed with the dq-coordinate model cannot be implemented without several supplemental functions such as the coordinate transformation and a DC-to-AC current encoding scheme. A quasi-physical BLDC motor model is developed to study the practical implementation issues of the dq-coordinate control strategy, such as the initialization and rotor angle transducer resolution. This model can also be beneficial during first stage development in automotive BLDC motor applications.

Dysprosium transport in Nd-Fe-B pellets

The addition of heavy rare earth (RE) elements to Nd2Fe14B based magnets to form (Nd,Dy)2Fe14B is known to increase the coercivity and high temperature performance required for hybrid vehicle electric motors and other extreme temperature applications. Attempts to conserve heavy rare earth elements for high temperature (RE)2Fe14B based magnets have led to the development of a grain boundary diffusion process for bulk magnets. This process relies on transport of a heavy rare earth, such as Dy, into a bulk Nd2Fe14B magnet along pores, a low volume fraction of eutectic liquid along grain boundary grain triple junctions and grain boundaries. This enriches the grain surfaces in Dy through the thickness of the bulk magnet, leading to larger increases coercivity with a smaller Dy concentration than can be achieved with homogeneous alloys. Attempts to carry out the same process during sintering require significant control of Dy transport efficiency. The macroscopic transport of Dy in Nd2.7Fe14B1.4 based powder packs is studied using a 'layered' pellet, where Nd2.7Fe14B1.4powder is an interlayer and Dy source as a center layer. The sintering of this layered pellet provided evidence for very large effective diffusion lengths aided by Dy rich liquid flow through connected porosity. Approaches to controlling Dy transportation include decreasing the liquid phase transport capability of the powder pack by increasing the melting point of the Dy source and the decreasing amount of RE rich liquid in the powder packs. The solid-liquid reaction is studied in which melt spun Nd2.7Fe14B1.4 ribbons are PVD coated with Dy-Fe eutectic composition and then thermally treated. The resulting microstructure from the reaction between Dy-Fe eutectic coating and Nd2.7Fe14B1.4 ribbon is interpreted as support for a proposed dissolution/reprecipitation process between solid and liquid phases. The estimate the diffusion coefficient and the effective diffusion length of Dy sources in Nd2.7Fe14B1.4 layered pellets and melt spun ribbons were obtained from the calculation of Fick's second law combined with EDS results from the experiment. The results indicate that the effective diffusion coefficient of Dy in the layered pellets is higher than the diffusion in ribbons due to its higher porosity than ribbons.