No abstract

The variable reactor (L) and capacitor capacity (C) of the passive dynamic harmonic filter have influence on the filtering performance. The method of the L and C selecting is advanced in this paper. To select the optimal C, two conditions about the reactive power compensation and harmonic filtering need to satisfy. To analyze L, T-type equivalent circuit is constructed in this paper. With the optimized L and C of the passive dynamic harmonic filter, the harmonic filtering and the reactive power compensation are finished reliably.

According to the low power factor and low running efficiency, a dynamic reactive power compensation method of the super high-power and high-voltage motor is proposed in this paper. The following works have been done in the study: topology of the dynamic reactive power compensation device; principle of the dynamic reactive power compensation method; control system of the dynamic reactive power compensation device; implementation of the dynamic reactive power compensation method. The amount of reactive power compensation can be adjusted smoothly and dynamically in the process of the super high-power and high-voltage motor soft-starting. The research of this paper has laid a theoretical foundation for this compensator in industrial applications. The novel design is characterized by flexible parameter setting, excellent soft starting performance of the super high-power and high-voltage motor and bright prospect in application.

In recent years, deep neural network is widely used in machine learning. The multi-class classification problem is a class of important problem in machine learning. However, in order to solve those types of multi-class classification problems effectively, the required network size should have hyper-linear growth with respect to the number of classes. Therefore, it is infeasible to solve the multi-class classification problem using deep neural network when the number of classes are huge. This paper presents a method, so called Label Mapping (LM), to solve this problem by decomposing the original classification problem to several smaller sub-problems which are solvable theoretically. Our method is an ensemble method like errorcorrecting output codes (ECOC), but it allows base learners to be multi-class classifiers with different number of class labels. We propose two design principles for LM, one is to maximize the number of base classifier which can separate two different classes, and the other is to keep all base learners to be independent as possible in order to reduce the redundant information. Based on these principles, two different LM algorithms are derived using number theory and information theory. Since each base learner can be trained independently, it is easy to scale our method into a large scale training system. Experiments show that our proposed method outperforms the standard one-hot encoding and ECOC significantly in terms of accuracy and model complexity.

A large number of inductive loads need the reactive power, and the amount of the reactive power compensation is different in a state of work. The traditional static reactive power compensator is difficult to meet the needs of the reactive power compensation for the composite load reactive. Therefore, a dynamic reactive power compensation method of the composite power load is proposed in this paper. The following works have been done in the study: principle of dynamic reactive power compensation control method, topological structure of dynamic reactive power compensation device, and implementation of the dynamic reactive power compensation control method.