Search

This paper presents two main novelties concerning power curve modeling of wind turbines. First novelty lies in the hybridization of 5 widely-used parametric functions and 8 recently-developed metaheuristic optimization algorithms. While constructing new hybrid power curve models, design coefficients of 4-parameter and 5-parameter logistic, 5th-order and 6th-order polynomial and modified hyperbolic tangent functions are fitted with ant lion, grey wolf, moth-flame and multi-verse optimizers and whale optimization, sine cosine, salp swarm and dragonfly algorithms. The best hybrid power curve model is achieved by the grey wolf optimizer-based modified hyperbolic tangent function in terms of the goodness-of-fit indicators. Second novelty lies in the integration of a well-known partitional clustering method to the best hybrid power curve model developed. While building a novel integrative power curve model, design coefficients of grey wolf optimizer-based modified hyperbolic tangent function are solved using only the highly representative data points identified by the Squared Euclidean-based k-means clustering algorithm. The operational characteristics of the wind turbine power curve are reflected with a higher accuracy. As a crucial result, the proposed power curve modeling framework is shown to be superior for wind turbines.

In the past few years, China's space science and application has entered the stage of the space station with large-scale space science experiments. The number of flight missions related to space science grows rapidly, meanwhile, the corresponding payloads from various fields of science become more complicated. This paper breaks through the traditional management mode in which the payloads of space science are usually monitored by space-ground telemetry links and proposes an approach to utilize workflow net models to monitor the experimental processes of on-orbit payloads in real time. This approach can effectively offload the pressure of space-ground communication as well as the pressure of ground monitor stations. In this approach, spectral graph clustering is used in decomposing a workflow net model when the model is too huge to be processed in parallel on one single GPU (graphic processing units) device, and hybrid parallel computing algorithms are designed to carry out conformance checking operations based on observed events from on-orbit payloads in real time. The algorithms are implemented with a gather-apply-scatter model in CUDA 9.0 (compute unified device architecture) on Jetson TX2i module and are benchmarked. The performance of the algorithms is acceptable for practical use on orbit.

Traffic network partitioning is of great importance in regional coordinated traffic signal control in urban areas. Most partitioning algorithms only use a single traffic parameter to represent dynamic traffic information, which will lead to inaccurate results. Moreover, traditional clustering and heuristic partitioning algorithms are not practical in applications. Thus, in this paper, we first propose a new combinatorial characteristic parameter for clustering-based partitioning algorithm by using the Pearson correlation coefficient and data normalization. Then, we refer to the idea of “snake” algorithm and use a linear programming model to obtain the exact partitioning result, and such algorithm avoids local optimum of heuristic algorithms. Finally, based on the real traffic data of a Chinese city, we conduct the experiments and verify the effectiveness of the new combinatorial parameter.

Games became popular, within the formal verification community, after their application to automatic synthesis of circuits from specifications, and they have been receiving more and more attention since then. This paper focuses on coding the "Sokoban" puzzle, i.e., a very complex single-player strategy game. We show how its solution can be encoded and represented as a Bounded Model Checking problem, and then solved with a SAT solver. After that, to cope with very complex instances of the game, we propose two different ad-hoc divide-and-conquer strategies. Those strategies, somehow similar to state-of-the-art abstraction-and-refinement schemes, are able to decompose deep Bounded Model Checking instances into easier subtasks, trading-off between efficiency and completeness. We analyze a vast set of difficult hard-to-solve benchmark games, trying to push forward the applicability of state-of-the-art SAT solvers in the field. Those results show that games may provide one of the next frontier for the SAT community.

Though cloud computing has become an attractive technology due to its openness and services, it brings several security hazards towards cloud storage. Since the distributed nature of clouds is achieved through internetworking technologies, clouds suffer from all the vulnerabilities by which networking also suffers. In essence, data stored in clouds are vulnerable to attacks from intruders. But, no single technique can provide efficient intrusion detection. In this paper, we propose fuzzy self-classifying clustering based cloud intrusion detection system which is intelligent to gain knowledge of fuzzy sets and fuzzy rules from data to detect intrusions in a cloud environment. Its efficiency is explained by comparing with other three cloud intrusion detection systems. Using a standard benchmark data from a CIDD (Cloud Intrusion Detection Dataset), experiments are conducted and tested. The results are presented in terms of success rate accuracy.

Modern Utility Management Systems which utilize multiprocessor systems for efficient processing of large data model are considered in this paper. The necessary preconditions for the efficient calculation are optimal load balancing of processors and data model partitioning among processors. The novel multilevel Super-Roots (SR) algorithm was developed to improve existing algorithms (e. i. METIS) for initial partitioning of data model. The proposed algorithms are applied on data model describing large electricity power distribution network. Experiments show that SR algorithm achieves better results than METIS multilevel algorithm in many cases.