Mina M. Rahimian

Implementation of a Fault-Diagnosis Algorithm for Induction Machines Based on Advanced Digital-Signal-Processing Techniques

In this paper, a complete cross-correlation-based fault-diagnostic method is proposed for real-time digital-signal-processor (DSP) applications that cover both the fault-monitoring and decision-making stages. In practice, a motor driven by an inverter or utility line is run at various operating points where the frequency, amplitude, and phase of the fault signatures vary unexpectedly. These changes are considered to be one of the common factors that yield erroneous fault tracking and unstable fault detection. In this paper, the proposed algorithms deal with the ambiguities of line-current noise or sensor-resolution errors and operating-point-dependent threshold issues. It is theoretically and experimentally verified that a motor fault can be continuously tracked when the sensor errors are within a limited range through the adaptively determined threshold definition of noise conditions. The offline experiments are performed via Matlab using actual line-current data obtained by a data-acquisition system. These results are verified on a DSP-based motor drive in real time where drive sensors and a digital signal processor are employed both for motor-control and fault-diagnostic purposes.

Performance-Oriented Electric Motors Diagnostics in Modern Energy Conversion Systems

This paper presents the analysis of a performance-oriented electric motors diagnostics in modern energy conversion system. With increased demand for electrical energy in world industries, the population of energy conversion devices such as generators/motors has greatly increased. As emerging and not being a mature enough technology in the application of renewable energy conversion or electric-drive transportation, the protection and diagnosis of electric motors have been extensively studied for safety and reliability. Meanwhile, motor phase currents commonly involve random noise components generated by harsh energy system environments, low- and high-order harmonics interferences caused by power inverters and fast switching devices, and various other design imperfections. Therefore, it is quite challenging to model the overall noise content and eliminate the disturbance while detecting motor fault signatures. Due to the inherent random variation of motor noise statistics, the noise model and elimination strategy should also be adaptively updated according to instantaneous noise conditions through which detection can be done with predefined performance expectation. Several successful solutions in the literature have managed to perform a diagnosis under certain noise conditions; however, a detailed performance and adaptability analysis covering arbitrary noise variation has not been satisfactorily addressed. This paper mainly deals with performance oriented threshold design strategies for fault signature detection utilizing the noise statistics of the motor phase current signal. The proposed solution is generalized to cover arbitrary Gaussian noise variations and derive the optimal form of the threshold that satisfies users prior detection quality expectations. The mathematical derivations are proved through statistical theory and the experimental verifications are performed by using a 3-hp motor setup.

Modeling of synchronous machines with damper windings for condition monitoring

In this paper, a method of modeling synchronous machines with damper windings based on the winding function approach (WFA) has been proposed. This method of machine analysis with the actual winding distribution does not require neglecting the harmonics of the winding distributions. The proposed model can be used to simulate the machine under faulty conditions such as broken damper bars and end ring, eccentricities and inter-turn faults. Given the geometry of the machine and its winding configuration, all the inductances are calculated as a function of the rotor position. The skewed rotor is included in the model and its smoothing effects on the waveforms are presented. Damper bar currents are studied under healthy and broken bar cases. The experimental results are presented and compared to the simulation results to verify the modeling. A special inside-out, round rotor synchronous machine is used for test and verification of the model. The field winding and the damper bars of this machine are mounted on the stator which makes it possible to monitor the bars and end ring currents directly and accurately without using flux probe.

Rail defect diagnosis using wavelet packet decomposition

One of the basic tasks in railway maintenance is inspection of the rail in order to detect defects. Rail defects have different properties and are divided into various categories with regard to the type and position of defects on the rail. This paper presents an approach for the detection of defects in rail based on wavelet transformation. Multi-resolution signal decomposition based on wavelet transform or wavelet packet provides a set of decomposed signals at distinct frequency bands, which contains independent dynamic information due to the orthogonality of wavelet functions. Wavelet transform and wavelet packet in tandem with various signal processing methods, such as autoregressive spectrum, energy monitoring, fractal dimension, etc., can produce desirable results for condition monitoring and fault diagnosis. Defect detection is based on decomposition of the signal acquired by means of magnetic coil and Hall sensors from the railroad rail, and then applying wavelet coefficients to the extracted signals. Comparing these extracted coefficients provides an indication of the healthy rail from the defective rail. Experimental results are presented for a healthy rail and some of the more common defects. Deviation of wavelet coefficients in the healthy rail case from the case with defects shows that it is possible to classify healthy rails from defective ones.

Incipient bearing fault diagnostics for inverter fed induction motor drive using ANFIS

Incipient fault detection of electrical machine is a challenging task and requires intelligent diagnostic approach. Huge research effort is put to automate the fault diagnostic schemes. Among several causes of electrical machine failure the most frequent occurring fault is the mechanical bearing failure. Thus this paper present on-line diagnostic technique for incipient bearing failure in an inverter fed three-phase induction motor drive system. The adaptive neuro-fuzzy inference system is utilized for the diagnostic purpose. The proposed technique is verified using simulation approach. The simulation is done using Matlab/Simulink and the complete model is presented in the paper.

Optimal design of PM assisted synchronous reluctance generators using lumped parameter model and Differential Evolution Strategy

This paper presents the design of high performance permanent magnet-assisted synchronous reluctance generators (PMa-SynRG) for the 3kW tactical quiet generator set. By introducing a proper quantity of permanent magnets into the synchronous reluctance generator rotor core an extended constant power-speed range at high efficiency and high power factor can be achieved. Different stator winding configurations i.e. distributed winding and concentrated winding of PMa-SynRG are compared using an analytical model based on lumped parameter model (LPM). For comparison, initially the distributed winding machine is optimized using differential evolution strategy (DES) and then the rotor structure of concentrated winding machine is optimized using the same stator. Finally, output performances are compared using finite element analysis. This design process is developed for optimized design of PMa-SynRG with minimum magnet volume, cogging torque and maximum efficiency and power factor.

Fault Diagnosis Implementation of Induction Machines based on Advanced Digital Signal Processing Techniques

In this paper, a comprehensive cross correlation-based fault diagnostic method is proposed for real time DSP implementation. It covers both fault monitoring and decision making stages. In practice, a motor driven by an adjustable speed drive is run at various operating points where the frequency, amplitude and phase of the fault signatures varies with time. These dynamic changes are considered as one of the common factor that yields erroneous fault tracking and unstable fault detection. In this paper, the proposed algorithms deals with the operating point dependent ambiguities and threshold issues. It is theoretically and experimentally verified that the motor fault can continuously be tracked when the operating point changes within a limited range.

Optimal design and comparison of stator winding configurations in permanent magnet assisted synchronous reluctance generator

This paper presents design of high performance permanent magnet-assisted synchronous reluctance generators (PMa-SynRG) for 3kW tactical quiet generator set. Adding the proper quantity of permanent magnets into the synchronous reluctance generator rotor core can offer large constant power-speed range, high efficiency and high power factor. Different stator winding configurations such as distributed windings and concentrated windings are compared using lumped parameter model (LPM). For this comparison, the distributed winding machine is optimized by differential evolution strategy (DES) and the rotor structure of the concentrated winding machine is optimized based on the same stator. Finally, output performances are compared using finite element analysis. The design is optimized for achieving the minimum magnet volume, minimum cogging torque, maximum efficiency, and maximum power factor.

A generalized condition monitoring method for multi-phase induction motors

As high power applications in industry commonly employ multi-phase motors and generators, a generic solution is needed for fault diagnosis of various multi-phase systems. Furthermore, the fault criteria set for the severity of the detected fault signature has still been a challenge due to the dynamic behavior of the fault-related signatures in the motor line current. Therefore a generalized solution is needed for multi-phase systems including the adaptive threshold that validates the fault detection algorithm through the entire operation range. In this paper, it is focused on a mathematical analysis of conventional DSP algorithms, frequency tone detectors, including Parks vector approaches. Based on the analysis, it is theoretically extended and simulated for a contribution of a future online fault diagnosis of multiphase machine.

A robust sensorless fault diagnosis algorithm for low cost motor drives

This paper presents a sensorless fault diagnosis technique for low cost AC drives. Currently, in order to achieve a reliable fault diagnosis, a high resolution speed sensor is employed to measure the frequencies of fault signature which depends on motor shaft speed. There is an increased tendency toward sensorless control of AC motor drives because of mounting problems, associated cost, etc. Therefore, the speed sensors become less common feedback tools in high performance motor control applications. The fault diagnosis becomes quite a challenging task in the absence of information from a speed sensor as highly precise motor speed estimation is needed. In this paper, a simple and efficient algorithm for fault diagnosis is proposed utilizing frequency tracking method which does not require speed sensor feedback. It is explicitly verified that the performance of the proposed algorithm is almost comparable to the cases where an accurate speed sensor is used. The algorithm is derived mathematically and its efficacy is proved experimentally using a 3-hp motor generator setup.