Summary The new power system stabilizer PSS4B demonstrates good performance on suppressing low‐frequency power oscillation. However, the numerous interdependent parameters make their tunings rather difficult. In this paper, a modified particle swarm optimization algorithm is applied to tune the parameters of PSS4B. The influence level of each block for the frequency characteristic is different. Hence, there are 2 steps for the optimizing parameters in this paper. The first step optimizes the parameters of hybrid block and gain block, and the parameters of lead‐lag block are optimized in the second step. To evaluate the effectiveness of the proposed method, several simulations are performed under very challenging conditions by MATLAB/Simulink. Moreover, the simulation results are compared with the conventional tuning method. The comparison indicates that the proposed method can provide damping characteristics and suppress low‐frequency power oscillation effectively.

No abstract

Simple experimental setups are designed for observing liquid surface waves, studying the dispersion relation and obtaining the liquid surface tension coefficient. In addition, the relationship between temperature and the liquid surface tension coefficient is verified. We take pictures of the surface wave patterns with a smartphone camera and measure the wavelength with software analysis based on image recognition. The experiment is performed not only with common devices and simple operations, but also with the student's own smartphone. As a result, the experiment itself is easy and convenient to carry out, which stimulates undergraduates' interest in this experiment.

Natural answer generation (NAG) is more and more popular in real-world knowledge base question answering (KBQA) systems for being able to automatically generate natural language answers with structured KB. Large-scale community QA-pairs crawled from the Internet could be directly used to train NAG models. However, it is pervasive in these datasets that one question may contain multiple answers of varied quality, and NAG models suffer from the simple principle of equal treatment of these answers. To address this problem, we propose two kinds of attention-based algorithms to handle all answers to a question at a time. Selective attention and self-attention mechanisms are used to dynamically weight the answers to one question during the training process. Specifically, selective attention methods weight the answers using the relationships between all the KB objects the question needs and the generated answers, and self-attention methods weight them according to the generating difficulty. The experiments on the public open-domain community QA dataset demonstrate that Selective-ATT outperforms the state-of-the-art by 10.53% in the entity accuracy, 9.34% in the BLEU score, and 1.19% in the Rouge score. INDEX TERMS Attention mechanism, knowledge based systems, machine learning algorithms, natural answer generation, natural language processing.

The crack propagation behavior of rock during compression involves complex mechanisms. Describing the growth behavior of a large number of cracks with conventional mechanical models is a major challenge. Therefore, in this work, we propose a new method to describe crack growth behavior by considering crack bodies as free voxels that can expand and coalesce within a rock sample according to certain rules. Specifically, we first propose a crack growth model that quantitatively describes the crack growth ratio and crack growth rate, which are integrally related to the loading rate, internal friction angle, cohesion, initial porosity, and confining stress. Second, to avoid the complex analytical process of the traditional mechanical model in solving the propagation directions of multiple cracks, we introduce a method for determining the crack growth directions of shearing failure based on the colony growth assumption. This method defines the crack propagation direction as a synthetic vector of the inertial direction, the attractive direction, the Coulomb direction, and the edge direction. Moreover, a new mathematical description method of fracture energy and plastic energy is proposed to calculate the crack growth at each time step. The simulation results show that our crack growth model for shearing failure agrees well with the experimental results and explains the fracture behavior and transformation law of cracks to some extent.