نشریه مهندسی برق
سال پنجاه و یکم شماره 3 (پیاپی 97، Autumn 2021)

In this paper, optimal robust controllers are developed for tracking control of a ducted fan engine of a thrust-vectored aircraft in the presence of external disturbances. First, by applying a nonlinear regulator based on the state-dependent differential Riccati equation (SDDRE) approach, an optimal control law is designed that is not robust against external disturbances. The second design, a VSC (variable structure control) with an NTV (nonlinear time-varying) sliding sector, is proposed. The sliding sector is a subset of the state space which is obtained by the SDDRE. As the final design, to guarantee the system's robustness against external disturbances and achieve optimal performance, a robust optimal sliding mode controller based on SDDRE (ROSMC) is designed, which integrates the sliding mode control (SMC) theory with the SDDRE approach. For each design, the global asymptotic stability is proved using the Lyapunov stability theorem. Also, the SDDRE is solved by a change of variable and converting it to a differential Lyapunov equation (DLE). Numerical simulations are presented considering different types of external disturbances and several scenarios. Simulation results show that ROSMC has stronger robustness and demonstrates optimal performance compared to SDDRE and VSC designs.

EvoStream is a stream clustering algorithm which gradually clusters data in the idle times of the stream. In comparison with other algorithms in this field, EvoStream has a lower computation overload in the offline phase and has better accuracy. Also, in this algorithm, the number of clusters is taken as constant whereas in an authentic stream this number varies with the complexity of input data. In this work, we present DynamicEvoStream as an improved version of the original EvoStream. In this algorithm, we detect and exploit variations in the distribution and speed of the stream.  Also, we modified the cleanup function to merge overlapping clusters. Therefore, in contrast to the basic EvoStream, DynamicEvoStream identifies the number of clusters in a dynamic manner. Also, the speed of evolutionary steps is increased while improving the quality of the clusters. Finally, experiments using DynamicEvoStream on different streams showed that it can cluster the stream up to four times faster than the original EvoStream with fewer computation and memory resources. In the worst case, the quality of clusters is competitive to the original EvoStream, however improves the quality of clusters up to 30% in the majority of cases.

In this paper, we study an uplink multi-unmanned aerial vehicle (UAV) wireless communication network where multi-UAV are deployed to serve the ground users by utilizing the non-orthogonal multiple access (NOMA) technology. The goal is to minimize the total transmit power of users by jointly optimizing the user association, sub-channel assignment, power allocation and UAVs’ position. The formulated problem is a mixed integer non-convex optimization that is difficult to solve in optimal approach. By applying the convex optimization tools, successive convex approximation (SCA) and Lagrange dual approaches, we solve the optimization problem then we propose an efficient iterative algorithm. Numerical results confirm that the proposed scheme can provide a better performance compared to the orthogonal multiple access (OMA), random position of UAVs in NOMA (RP-NOMA) and OMA (RP-OMA) schemes in both of the transmit power and sum-rate performance metrics.

The congestion control problem is one of the most essential subjects in the Transmission Control Protocol (TCP) Network because of the complex nonlinear model, uncertainties, and external disturbances.  This paper extends the adaptive robust H∞ control finite-time approach to TCP network and presents a new solution to solve the congestion control problem employing Active Queue Management (AQM). Firstly, a modified nonlinear model of TCP network system with parametric uncertainties and external disturbance is given. Then, by several variable changes based on the backstepping method and Lyapunov function, adaptation and control laws were derived. Stability analysis is given to prove that all the signals of the closed-loop system are finite-time bounded. In addition, the results show that the proposed controller can guarantee both the transient and steady-state performance of the system, the queue of the TCP network system can track the desired queue and the disturbance is rejected satisfactory based on H∞ control part of the controller. Finally, a comparison example is considered to demonstrate the feasibility and superiority of the presented scheme.

Simulation-based test pattern generation methods are an interesting alternative to deterministic methods because of lower time complexity. In these methods, test patterns are evaluated and those with higher efficiency are selected. Traditionally, test pattern selection is based on fault coverage, which is an accurate merit indicator, but its calculation is time-consuming. Instead of fault coverage, approximate indicators can be used to assess efficiency of test patterns. In this paper, an approximate indicator called APXD is proposed, which is more efficient than existing approximate methods. Our experimental results show that APXD indicator has a strong correlation with fault coverage. In addition, APXD simulation is 1900x, 63x, and 56x faster than serial, sampling, and parallel fault simulation, respectively. Exploiting APXD indicator instead of fault coverage, in a pruning-based test generation method, leads to about 700x, 24.2x, and 18.4x speedup, respectively compared to pruning based methods that use serial, sampling, or parallel fault simulation for test pattern evaluation, at fault coverage of 80%. Speedup at fault coverage of 95% is about 111.3x, 11.1, and 3.6x, respectively. While, the use of APXD indicator instead of fault coverage increases the number of test vectors by 2% at most, confirming the efficiency of APXD indicator compared with probabilistic and statistical approximate indicators.

With the extension of social networks, research on influence maximization (IM) in time-sensitive graphs has increased in recent years. IM is a problem to find a seed set with k nodes to maximize the information propagation range in the graph. Most of the research in this area consists of greedy, heuristic, meta-heuristic methods. However, most of these methods ignore the time-sensitivity to propagation delay and duration. The preceding time-sensitive centrality measures as a part of heuristic approaches take the propagation delay but only consider the nodes locally so that each graph node considers only the direct neighbors. Based on the above analysis, this article focuses on the time-sensitive IM problem. Here, a propagation value for each path in the graph is defined in terms of the probability of affecting through the edge and freshness amount of the edge. To solve the problem, we propose time-sensitive centrality measures that consider propagation value and both the direct and the indirect neighbors. Therefore, four measures of time-sensitive closeness centrality (TSCloseness), time-sensitive harmonic (TSHarmonic), time-sensitive decay centrality (TSDecay), and time-sensitive eccentricity centrality (TSEccentricity) were proposed. The experiments on five datasets demonstrate the efficiency and influence performance of the TSHarmonic measure on evaluation metrics.

In this paper, an automatic outdoor fire detection method is proposed for the fire videos recorded by a fixed camera. First, a new set of color rules is introduced to eliminate the non-fire pixels as much as possible while detecting the fire zone pixels completely. Next, the texture and flicker effect features are extracted from the detected fire zone, to remove the remainder of non-fire pixels if still any non-fire pixel exists. The texture feature is extracted by using the angular second moment. To extract the flicker effect feature, the time prehistory signal of color components of each fire zone pixel is obtained and passed through a half band high pass filter. Finally, the Ward classifier clusters the fire features to separate the fire zone pixels from the non-fire. At the various steps of the proposed method, the morphology process is also used to improve the accuracy of fire detection. The proposed method is applied to the 200 different fire videos including the fire-like objects. Simulation results indicate 6% to 56% improvement on performance of the proposed method in comparison to the similar ones.