Search

A method, being based on multi-class support vector machine (SVM) classifier and hybrid particle swarm optimization (PSO) and gravity search algorithm (GSA), is presented to diagnose the faults in electrical motor drive system. In this method, the global search ability of PSO and the local search ability of GSA are integrated to combine the advantages of PSO and GSA, and the multi-class SVM classifier is optimized by the hybrid GSAPSO algorithm to improve classification performance. To test the presented method, a series of simulation and experiment are studied. The diagnostic results display that the presented method can gain more precise classification accuracy than multi-class SVM with PSO and multi-class SVM with GSA. • A multi-class support vector machine classifier constructed by one-versus-one method is used for fault diagnosis of Electrial machine drive system. • Radial basis kernel function is adopted as the kernel function of support vector machine, where Hybrid GSAPSO lgorithm is applied to solve optimization problems. • Fault diagnosis steps are presented, and the effectiveness of the proposed method is validated. [ABSTRACT FROM AUTHOR]

When it concerns the reliable working of the modular multilevel converter (MMC), the power loss balance control cannot be ignored. Under the submodule (SM) malfunction condition, each arm of the MMC operates in an asymmetrical status, resulting in unbalanced power loss in every arm of MMCs, which affects the converter operation's reliability. This essay proposes a variable capacitor voltage deviation predefined value-based power loss balance control (VCVDPV-PLBC), which could optimize faulty arms' power losses by regulating capacitor voltage deviation predefined value to change power devices' switching loss in MMCs under SM malfunction. On this basis, the coupling relationship between SM's switching frequency and the capacitor voltage deviation predefined value is revealed. The raised VCVDPV-PLBC could effectually prolong the service life of power devices in faulty arms and strengthen the reliability of MMC system under the malfunction of the SM. The validity of the raised VCVDPV-PLBC was proved by simulation results with specialized software PSCAD/EMTDC and experimental studies with small-scale MMC setup. [ABSTRACT FROM AUTHOR]

• 1. To propose a quasi-quantum model, which can reasonably divide and guide the planning period to reduce uncertainty. • 2. To propose an evaluation model, which can analyze the uncertainty potential energy function before and after the planning to assist the quasi quantum model, so as to reasonably divide the planning cycle. • 3. To propose a three-layer planning framework, including planning cycle division, stochastic planning, and rolling planning. With the rapid development in social informatization, more and more factors, such as regional economy, technological development, and people's living needs, will affect the supply–demand relationship of regional integrated energy systems (RIES), which involve multiple energy forms. These factors turn the supply–demand relationship of an energy system into a nonlinear and time-varying chaotic system. This makes it difficult to predict and balance these relationships, which poses a huge challenge to the prediction, planning, operation, etc. Existing conventional methods attempt to quantify the prediction bias caused by various external environmental factors of energy systems and to weaken the uncertainty caused by multiple energy loads through random programming. However, uncertainty factors increase with the development of an energy system, thereby inreasing the requirements of such uncertainty quantification methods. Furthermore, in conventional methods, the prediction or planning decisions are often made by an observer while neglecting the impact of the observer's decision-making behavior on prediction and planning. Therefore, this paper proposes a three-layer planning framework for to solve the above problems. This architecture includes quasi-quantum uncertainty periodic evaluation, stochastic planning based on information gap decision theory, and rolling planning based on model predictive control. First, we establish a quasi-quantum model of multi-energy system prediction error based on the quasi-quantum wave function model to qualitatively analyze prediction errors affected by uncertainty before and after planning. Simultaneously, combined with the evaluation model of the quasi-quantum potential energy function, a quasi-quantum uncertainty period evaluation model is proposed. Based on the minimum equivalent planning entropy, the planning cycle of multistage rolling planning is divided to minimize uncertainty. Second, the information gap decision theory is used to quantify the influence range of source and load uncertainty in the planning process and the multistage planning cycle is randomly planned. Then, the model predictive control method is used to control the actual error, and the planning strategy is changed in time to reduce the planning deviation caused by the prediction error. Finally, adopting a Beichen demonstration area in Tianjin, China, as a case study, the effectiveness of the proposed method in uncertainty analysis and long-term planning improvement is verified. The three-layer planning framework can improve the adaptability of the regional integrated energy system in the long-term planning process, and can timely adjust the planning scheme to cope with the impact of unpredictable uncertainties in the planning process. © 2017 Elsevier Inc. All rights reserved. [ABSTRACT FROM AUTHOR]