Xiaodong Shi

A New Approach to Fault Diagnostics for Permanent Magnet Synchronous Machines Using Electromagnetic Signature Analysis

This paper proposes a novel approach to health monitoring and multifault detection in permanent magnet synchronous machines using direct flux measurement with search coils. Unlike other spectrum-based fault detection schemes, only the fundamental frequency component of the measured voltage is utilized for fault detection. Therefore, the performance of the proposed scheme is not affected by nonstationary speed or harmonics introduced by the power supply. In addition, location of interturn short circuits and direction of static eccentricity can be detected, which have never been done by any other scheme. In spite of the invasive nature of the technique, it is very suitable for mission-critical applications and emerging applications such as off-shore wind turbines, hybrid vehicle technology, and military applications, where early detection of faults is of paramount importance. 2-D simulations using finite element analysis have been presented to validate the proposed method under different operating conditions. Experimental introduction of stator interturn short circuit, demagnetization, and static eccentricity has been discussed, and the proposed scheme is experimentally implemented to examine its effectiveness.

Vibration monitoring of PM synchronous machine with partial demagnetization and inter-turn short circuit faults

Monitoring permanent magnet synchronous machine (PMSM) faults including partial demagnetization, inter-turn short circuit is necessary for safety and reliable operation. This paper studies vibration acceleration to detect faults using both mode shape and vibration frequency information. For this purpose, Finite element analysis is used to calculate the radial force distribution applied on the stator teeth and mode shape of the PMSM. 2-dimensional Fast Fourier Transformation (FFT) is employed to extract the significant harmonic orders, which dominate the vibration. Simulation and experimental results indicate that featured low mode order and corresponding frequency component would appear under fault conditions. Simulation and experimental results are used to analyze the effectiveness of the proposed approach.

A Novel Universal Sensor Concept for Survivable PMSM Drives

Permanent-magnet synchronous machines (PMSMs) are widely used as propulsion motors in hybrid electric, plug-in hybrid, and electric vehicles and as generators in renewable energy applications. Since position and current information is typically indispensable for PMSM control strategies such as vector control or direct torque control, failure of an encoder/resolver or current sensors could lead to a catastrophic failure, if no protection or backup plan is established. This paper lays groundwork for developing a novel backup universal sensor concept for PMSM with position and current estimation. Estimated position and current could be used for sensor fault detection or survivable drive, giving its drive system a “+1” fault tolerance. This technique uses search coils, which are implemented to directly monitor magnetic flux through stator teeth. Adaptive position estimator and a current estimator have been designed based on the search coil model, which overcomes the issue of resistance variation that exists in many estimators reported in literature. Simulation and experimental verification are presented to verify feasibility of the proposed strategy.

Survivable Operation of Induction Machine Drives With Smooth Transition Strategy for EV Applications

In electric vehicle (EV) applications, the failure of phase current sensors often leads to emergency shutdown of induction machine (IM) drive to avoid system malfunction and protect the driver and hardware. However, emergency stop is not necessarily the best way to react if current sensor fails. This paper proposes a survivable operation technique for three-phase IM drives in the event of current sensor failure. The basic objective of survivable operation for IM is to seamlessly switch the control method from vector control to a simplified digital control with smooth transition strategy in the event of current sensor failure. Using this technique, the IM system can be kept running even when there are current sensor faults. It can therefore add one extra redundancy in EVs and make them safer and more reliable. Simulation and experimental results, including current sensor failure detection, speed and torque response, voltage vector, and phase-current waveforms, are included to show the effectiveness of the proposed strategy.

Concept and implementation of a simplified speed control strategy for survivable induction motor drives

Induction machine is one of the most widely used machines in residential and industrial applications. Traditional drive methods for induction machines such as vector control usually require complex control routines, fast processing unit and multiple system status feedbacks. The complexity of these methods greatly reduces the reliability of the system since the failure of the sensor or even the drift of system parameters could potentially result in system malfunction. This creates the need for a simple, cost-effective and reliable control strategy as a backup to continue operation of the system in case of failure of current and voltage feedback sensors. In this paper, an effective, yet simple and low-cost state switching control technique is proposed and implemented for a three phase squirrel cage induction machine system. This state switching control operates at two specific duty cycles of PWM which produces phase voltages with different magnitudes across the phase windings. At the same time, the frequency of phase voltages is derived from the reference speed. By switching between these two states, precise speed regulation can be achieved. This new control method makes the controller extremely simple in design and implementation for induction machine. Simulation and experimental results are included in this paper to validate our claims.

Design and implementation of a novel power assisted drivetrain for a wheelchair

Power assisted wheelchairs are one of the most advanced wheelchairs in the current commercial market. Although such wheelchairs are well suited for a physically disabled person who needs physical activity, they tend to be expensive and inefficiently designed. This paper introduces the overall development of a novel power assisted wheelchair. Research on the conventional three types of wheelchair is presented, as well as the comparisons among three kinds of wheelchair and the recent design research. The proposed power assisted wheelchair is meant to match and exceed the conventional wheelchairs performance but with a more intelligent and lower cost design. Instead of using a torque sensor or pressure transducers for measuring and amplifying human force, the proposed drive train controller uses infrared sensors to trigger two Brushless DC hub motors. Using this information in addition to the information from a motion sensor that detects the road conditions, appropriate torque command is generated. The whole drive train systems power is supplied by a Li-battery package. One wheelchair prototype has been built, and conceptual designs are modeled and simulated in MATLAB/Simulink.

Noise Source Localization on a Small Wind Turbine Using a Compact Microphone Array with Advanced Beamforming Algorithms: Part II — A Study of Mechanical Noise from Nacelle Using a Wind Turbine Drive Train Simulator

The main goal of this study was to separate and evaluate the noise contribution of each sub-component of a small scale wind turbine drive train that contributes to the overall mechanical noise inside the nacelle of a small (8 kW) wind turbine. The experiments were performed in a drive train simulator that housed the drive train of a Viryd 8000 wind turbine. Different wind speeds were simulated on the test stand through a drive system consisting of a motor, gearbox and flywheel. In order to evaluate the overall sound generation single microphone acoustic measurements were made. The frequency of sound generation was observed to be the same as that of vibrations of the drive train measured using two accelerometers. Sound radiated by each component of the drive train was evaluated through acoustic beamforming using a compact microphone array in conjunction with advanced beamforming algorithms. A strong relationship between the drive train power generation, and sound emitted by various mechanical components was observed. A unique contribution of this study is our ability to localize, separate and quantify noise from sub-components in a complex functioning wind turbine drive train.

Digital Control of Induction Machines as a Backup Control Strategy for Fault Tolerant Operation of Traction Motors

Induction machine (IM)-based electric vehicle applications have gained significant popularity over recent years. However, failure of IM drivetrain components can lead to system malfunction and result in vehicle shutdown. Controller redundancy method is often used to tolerate microcontroller failure. However, this method results in high cost and complicates system design in terms of controller synchronization. This paper proposes a simplified digital control method for IM with low hardware requirements. Potentially, digital control method can work using simple auxiliary circuits without a microcontroller and can be used as a backup control strategy to maintain continuous operation of the IM in the event of microcontroller failure. This paper focuses on introducing the principle and implementation of digital control for IM and evaluating its performance. This digital control method can control three-phase squirrel cage IM system and enables motoring and generating operation. This method is extremely simple in design and implementation. Simulation and experimental results are included in this paper to validate our claims.

A simplified state switching control strategy for survivable variable-speed induction generators

Induction machines are one of the most widely used electric machines in wind power applications. However, the control of an induction generator is complicated if advanced control strategy is employed in order to maximize the performance. Also, the complexity of advanced control strategies and their high demand on hardware and sensors can lower the overall reliability of the generator system and potentially result in system malfunction, leading to unscheduled maintenance and power yield reduction. In this paper, an effective, yet simple and low-cost state switching control technique is proposed and implemented as a backup control strategy in case of control or sensor failure. This state switching control system is only allowed to operate at two different duty cycles of PWM which produces phase voltages with different magnitudes. At the same time, the frequency of the phase voltages is derived from the reference speed. By switching between these two states, precise speed regulation can be achieved. Owing to its simplicity, the control strategy can be implemented on a low cost FPGA or embedded into existing controllers for use as a backup strategy. In addition since it does not require the use of current sensors, it is very suitable for survivable operation. Simulation and experimental results are included to validate our claims.

A back EMF-based rotor position prediction in Permanent Magnet machines for survivable wind generator systems

In wind power generators, position information is often required for tracking maximum power point as well as implementing control strategies for the Permanent Magnet (PM) generator. For such a system, failure of position sensor could potentially lead to major fault or require the system to be shut down. This could cause significant economic losses and also require unscheduled maintenance. This paper proposes a new sensorless position estimation approach to track position of the PM machine using the back EMF if the position sensor fails. This technique can also be used to provide initial rotor position for the proper implementation of a fall-back strategy. This technique can also be extended to detect position information at low speeds. Simulation and experimental results have been presented showing the effectiveness the proposed scheme and validate the claims presented.