Power drives are used for induction motor control, uninterruptible power supplies, and in electrical vehicles. The increasing penetration of power drives makes their reliability, robustness, and early diagnosis a central point of attention especially in planning, designing, and financing. This book explores fault diagnosis of inverter drives to enable early diagnosis and robust design for efficient long life operation. Fault Diagnosis for Robust Inverter Power Drives focuses on early diagnosis, prognosis, and intrinsic reliability of inverter power drives and their applications. Topics include material degradation, materials, semiconductors, inverter topologies, and early diagnosis as well as fault tolerant software strategies. This work is highly relevant to researchers, power electronics professionals, and system designers in aerospace, hybrid and electrical cars, and power systems.
Inspec keywords: electric motors; semiconductor device reliability; fault diagnosis; power capacitors; electric drives; power semiconductor devices; invertors; fault tolerance; cells (electric)
Other keywords: battery storage; reliability; DC-link capacitors; power semiconductors; GaN; embedded fault diagnosis; invertor power drive; MOSFETs; power convertors; motor diagnostics; power component; aging; fault tolerance; failure; SiC; IGBTs; power drive components; life expectancy; fault prognosis; protection
Subjects: Power electronics, supply and supervisory circuits; d.c. machines; Reliability; General electrical engineering topics; Power semiconductor devices; DC-AC power convertors (invertors); Other power apparatus and electric machines; a.c. machines; Drives; Electrochemical conversion and storage
Several decades ago, the prevalent concept of the community was that there were fundamentally different approaches to diagnosing and maintaining mechanical versus electrical/electronic devices. For mechanical devices, the approaches were more concentrated on wear and tear, based on life-based model estimation; for electrical/electronic devices, the approaches were concentrated only on probabilistic and random phenomena. In other words, electronic approaches were mainly confined to statistics, assigning a probabilistic value of failure to each of the components to determine the overall reliability of the equipment. With some few exceptions, while the statistical approach is extremely valuable, we can do more in understanding reliability because the process of electromagnetic energy conversion of a device requires matter. More specifically, matter is the enabler of the process, channeling and regulating the conversion of one form of energy to another, usually from electric to magnetic or vice versa. The fact that the converters require matter as “the enabler” exposes the fundamental principle: that aging governs electrical elements in the same fashion as do the mechanical parts. Consequently, solid-state materials such as conductors, insulators, or semiconductors transfer energy from molecule to molecule, atom to atom, degrading during the process, which is modified by the level of temperature, electromagnetic fields, humidity, and other factors. Therefore, small cracks that appear due to impurities or “hot collisions” progress over time and are manifested as aging in the material, which usually shows as a loss of elasticity or an increase in the losses associated with energy transfer. This process creates a degradation “marker” that can be identified in many cases at the early stages of the degradation process. The understanding, identification, and progression of these degradation markers are the focal point of this chapter.
The reliability of a power converter depends mainly on the endurance of its main component, the power semiconductor. Therefore, particular attention is paid in this chapter to understand the models that allow the identification of features and early indicators of problems that establish the groundwork for early fault diagnosis in inverters. The two major switch technologies that control the market are field-effect transistors (FETs) and insulated gate bipolar transistors (IGBTs), which are bipolar devices integrating concepts from bipolar junction transistors (BJTs; IGBTs) which are bipolar devices integrating concepts from BJTs and FETs. From the beginning of power electronics, silicon-based semiconductors have been the undisputed king, but modern silicon-based power semiconductors are being challenged by silicon carbide (SiC) and more recently by gallium nitride (GaN). Regardless of the semiconductor material, early diagnostics in semiconductors are based not only on the understanding of the failure mechanisms but also on the “parasitic structures” that are intrinsically associated with the fabrication of the device. The identification and characterization of these “parasitic or associated structures” allow for the development of unified models with general characteristics across different materials and structures of power semiconductors. The aging effects are, in general, reflected in the parasitic structures early in the process of degradation, creating an ideal approach for understanding failure propagation. Fortunately, semiconductor devices share a similar structure, and a standard model across all power semiconductors is used in this chapter to understand aging and to enable early diagnostics.
This chapter presents an RUL prediction algorithm based on accelerated life test data and derived physics models for electrolytic capacitors. The main elements are (1) development of the first principles based degradation models; (2) the implementation of a Bayesian-based health state tracking and RUL prediction algorithm based on the Kalman filtering framework. One major advancement reported here is the prediction of RUL for capacitors as new measurements become available. The key contribution of this work is the prediction of RUL for capacitors as new measurements become available. The derived degradation models can be updated and developed at a finer granularity to be implemented for detailed prognostic implementation. This capability increases the technology readiness level ofprognostics applied to electrolytic capacitors.The results presented here are based on accelerated life test data and on the accelerated life timescale. Further research will focus on development of functional mappings that will translate the accelerated life timescale into real usage conditions timescale, where the degradation process dynamics will be slower and subject to several types of stresses. The performance of the proposed exponential-based degradation model is satisfactorily based on the quality of the model fit to the experimental data and the RUL prediction performance as compared to ground truth.
The increasing use of power electronics in industry applications, energy conversion, electric vehicles (EVs), aircrafts, vessels, etc. propels efforts at early detection of potential abnormalities, which can degrade system performance, in order to reduce profit losses and even risk to human lives. In the onset of the power electronics digital control era, microprocessors were used exclusively for control tasks, mainly due to the lack in computing power needed by diagnosis algorithms. Nowadays, powerful low-cost microprocessors have become standard options for the design engineer, and embedded routines that perform real-time fault diagnosis are easily added to the design. This chapter gathers the most relevant embedded techniques for real-time condition monitoring (CM) and fault diagnosis in power drives based systems that have been reported recent years, with special emphasis on those methods using electric variables (i.e., voltage and current) as health indicators of the involved devices.
The widespread use of power electronics-based converters has resulted in continuous improvement in the power switches' reliability. The previous authors shows a failure rate of medium-power IGBT modules of 0.1 failures per 106 h, many critical applications need further assurance of operation. This chapter has shown a sample of the techniques used to increase the system operating range under adverse conditions. Some techniques are applied before the onset of a device's fault, by allowing the device to move out of line for “recovery.” Other techniques modify the topology, either by disabling the faulty devices or by also including extra elements to replace the faulty ones. In most cases, this operation needs a corresponding change in the control strategy, as well as in the amount of power that the system can handle during this fault-tolerant operation.
This chapter outlines the state of the art on motor protection, monitoring, and failure prediction using inverter capabilities (sensing, data processing, and signal application) while feeding the electric motor. The methods included in this comprehensive review are thermal monitoring and protection schemes, monitoring schemes for insulation related issues, and methods that analyze the mechanical health of the motor or an asset connected to the motor. Other topics of interest include thermal modeling of electric machines, spectral analysis, expert systems, neural networks, and parameter estimation. The techniques described in this chapter are applied to induction machines, brushless direct current (DC) machines, and reluctance and synchronous motors.
This chapter focuses on lithium-ion batteries. It also covers relevant aspects of this technology, focusing on the impedance measurement as a diagnostic tool mechanism for detecting aging on this type of batteries. Furthermore, we divide such mechanisms into online and offline methods based on their means of operation during the service time of the battery.
FDP methods were developed in RS framework with great achievement for its simplicity. With the trend of distributed FDP, LSM is introduced with a philosophy of “execution when needed" to reduce the computation and make the long-term online prognosis possible on embedded systems. In this chapter, EKF algorithm is developed in the LS framework. An experiment of lithium-ion battery SOH diagnosis and prognosis with comparison against traditional RS-based approach is presented. It is demonstrated that the proposed approach is able to reduce the requirement on computational sources. This proposed approach combines the advantages of EKF and LS method, which results in low computation and small uncertainty accumulation.