Photovoltaic power system is taking a significant percentage of power system and the demands for accurate forecasting of the power outputs is surging. In prior works, the forecasting problem was formulated as a regression problem, however, which most cannot guarantee that the forecasted outputs is nonnegative. To solve this problem, we proposed a novel probabilistic model by using nonlinear regression and Bayesian learning method. In the paper, we present the detailed theoretical derivations and interpretations. The simulation results show the validity and feasibility of the proposed algorithm by comparing with the traditional SVM algorithm.
How to reduce the hardware cost and high power consumption of RF link of communication device is the key problem to be solved for multi-transmitting antenna and multi-receiving antenna system (MIMO). Always choose the best antennas connection a limited number of RF circuits, which is called antenna selection technology (AS), are a perfect solution to the problem, Assuming that the spatial range of the antenna meets the requirements of signal multiplexing and based on the maximum capacity criterion of the selected MIMO system, the manuscript proposes a low computational complexity (CC) and high performance joint transmitting and receiving antenna selection technique (JTRAS). Starting from the traditional capacity formula and the full matrix of MIMO channel, we utilize a simplified channel capacity expression through repeatedly iterating to delete a row and a column of the equivalent decrement channel matrix, which is to remove a pair of transmitting and receiving antennas. Based on the decreasing JTRAS (DJTRAS) algorithm, the capacity results of simulating calculation indicate that its median capacity overtakes other ones, such as optimum selection (OS), AS based on Frobenius 2 norm (NBS), and concise joint AS criterion (CJAS) etc., and the novel DJTRAS scheme can significantly reduce computational complexity (CC) compared to the exhaustive search method with maximum capacity, which defined as optimal algorithm in the curve graphs. This new technology of the AS is particularly suited to large number of selected antennas, such as Lt ≥ NT/2,Lr ≥ NR/2.
The paper is devoted to the problem of mining graph data. The goal of this process is to discover possibly certain sequences appearing in data. Both rough set flow graphs and fuzzy flow graphs are used to represent sequences of items originally arranged in tables representing information systems. Information systems are considered in the Pawlak sense, as knowledge representation systems. In the paper, an approach involving ant based clustering is proposed. We show that ant based clustering can be used not only for building possible large groups of similar objects, but also to build larger structures (in our case, sequences) of objects to obtain or preserve the desired properties.
Two approximate representations are proposed for distributed parameter systems described by two linear hyperbolic PDEs with two time- and space-dependent state variables and two collocated boundary inputs. Using the method of lines with the backward difference scheme, the original PDEs are transformed into a set of ODEs and expressed in the form of a finite number of dynamical subsystems (sections). Each section of the approximation model is described by state-space equations with matrix-valued state, input and output operators, or, equivalently, by a rational transfer function matrix. The cascade interconnection of a number of sections results in the overall approximation model expressed in finite-dimensional state-space or rational transfer function domains, respectively. The discussion is illustrated with a practical example of a parallel-flow double-pipe heat exchanger. Its steady-state, frequency and impulse responses obtained from the original infinite-dimensional representation are compared with those resulting from its approximate models of different orders. The results show better approximation quality for the “crossover” input–output channels where the in-domain effects prevail as compared with the “straightforward” channels, where the time-delay phenomena are dominating.
Satellite image classification is essential for many socio-economic and environmental applications of geographic information systems, including urban and regional planning, conservation and management of natural resources, etc. In this paper, we propose a deep learning architecture to perform the pixel-level understanding of high spatial resolution satellite images and apply it to image classification tasks. Specifically, we augment the spatial pyramid pooling module with image-level features encoding the global context, and integrate it into the U-Net structure. The proposed model solves the problem consisting in the fact that U-Net tends to lose object boundaries after multiple pooling operations. In our experiments, two public datasets are used to assess the performance of the proposed model. Comparison with the results from the published algorithms demonstrates the effectiveness of our approach.
The paper proposes a discrete-time sliding mode controller for single input linear dynamical systems, under requirements of the fast response without overshoot and strong robustness to matched disturbances. The system input saturation is imposed during the design due to inevitable limitations of most actuators. The system disturbances are compensated by employing nonlinear estimation by integrating the signum of the sliding variable. Hence, the proposed control structure may be regarded as a super-twisting-like algorithm. The designed system stability is analyzed as well as the sliding manifold convergence conditions are derived using a discrete-time model of the system in the δ-domain. The results obtained theoretically have been verified by computer simulations.
This paper proposes a methodology for observer-based fault estimation of leader-following linear multi-agent systems subject to actuator faults. First, a proportional-integral distributed fault estimation observer is developed to estimate both actuator faults and states of each follower agent by considering directed and undirected graph topologies. Second, based on the proposed quadratic Lyapunov equation, sufficient conditions for the asymptotic convergence of the observer are obtained as a set of linear matrix inequalities. Finally, a numerical example is provided to illustrate the proposed approach.
We introduce a novel fractional order adaptive control design based on the tube model reference adaptive control (TMRAC) scheme for a class of fractional order linear systems. By considering an adaptive state feedback control configuration, the main idea is to replace the classical reference model with a single predetermined trajectory by a fractional order performance tube guidance model allowing a set of admissible trajectories. Besides, an optimization problem is formulated to compute an on-line correction control signal within specified bounds in order to update the system performance while minimizing a control cost criterion. The asymptotic stability of the closed loop fractional order control system is demonstrated using an extension of the Lyapunov direct method. The dynamical performance of the fractional order tube model reference adaptive control (FOTMRAC) is compared with the standard fractional order model reference adaptive control (FOMRAC) strategy, and the simulation results show the effectiveness of the proposed control method.
The global (absolute) stability of nonlinear systems with fractional positive and not necessarily asymptotically stable linear parts and feedbacks is addressed. The characteristics u = f(e) of the nonlinear parts satisfy the condition k1e ≤ f(e) ≤ k2e for some positive k1 and k2. It is shown that the fractional nonlinear systems are globally asymptotically stable if the Nyquist plots of the fractional positive linear parts are located on the right-hand side of the circles (−1/k1, −1/k2).
Medical imaging tasks, such as segmentation, 3D modeling, and registration of medical images, involve complex geometric problems, usually solved by standard linear algebra and matrix calculations. In the last few decades, conformal geometric algebra (CGA) has emerged as a new approach to geometric computing that offers a simple and efficient representation of geometric objects and transformations. However, the practical use of CGA-based methods for big data image processing in medical imaging requires fast and efficient implementations of CGA operations to meet both real-time processing constraints and accuracy requirements. The purpose of this study is to present a novel implementation of CGA-based medical imaging techniques that makes them effective and practically usable. The paper exploits a new simplified formulation of CGA operators that allows significantly reduced execution times while maintaining the needed result precision. We have exploited this novel CGA formulation to re-design a suite of medical imaging automatic methods, including image segmentation, 3D reconstruction and registration. Experimental tests show that the re-formulated CGA-based methods lead to both higher precision results and reduced computation times, which makes them suitable for big data image processing applications. The segmentation algorithm provides the Dice index, sensitivity and specificity values of 98.14%, 98.05% and 97.73%, respectively, while the order of magnitude of the errors measured for the registration methods is 10−5.