This paper presents a new relocation method for Adaptive Weighted Aggregation (AWA) that is a powerful multi-start framework of scalarized decent methods for multi-objective continuous function optimization. AWA repeats two procedures; subdivision and relocation. Subdivision decides initial weight vectors and solutions for relocation. Relocation iteratively adapts weight vectors by repeating two procedures of optimization and weight adaptation in order to improve the coverage of an approximate solution set. AWA has been reported to find better approximate solution sets in terms of coverage than conventional multi-start methods. However, relocation has three serious problems when applied to problems with strong non-linearity. First, relocation often does not converge. Secondly, relocation often converges before obtaining the desired weight vector. Thirdly, the convergence speed often becomes very slow. In order to remedy these problems, we propose a new relocation method named the step size control weight adaptation method (SSCWA). In order to investigate the effectiveness of SSCWA, we compared the performance of AWA with SSCWA (AWA-SSCWA) with that of the original AWA on three to five objective benchmark problems with ten variables. As a result, we confirmed that AWA-SSCWA outperformed the original AWA on the benchmark problems.
Many human augmentation devices have been developed in the past few decades, but lots of them neglected the importance of well-timed assistance, which commonly leads to problems like lowered performance or interference with user's movement. In this research the authors focused in depth on the timing issue of powered assistance. As a pilot study for the proof of concept, the authors hypothesized that different timing of assistance may greatly affect the efficiency of movement for power-assisting devices, and verified the hypothesis through human experiment on sit-to-stand (STS) movement. By measuring the electromyogram (EMG) of lower limb muscles, the authors firstly confirmed the effectiveness of assistance, then compared the effect of assistance timing with other frequently investigated determinants of STS movement, and evaluated the efficiency of movement under different timing conditions. The results demonstrated the importance of timing in powered assistance, and showed that, rather than the common practice of providing assistance after the onset of movement, which is adopted by most traditional proportional-EMG-controlled human augmentation devices, it may improve the efficiency of movement for at least half of the subjects simply by adopting an earlier (before or at the same time of movement onset) but optimal timing for assistance. As a preparation for future works, the authors also quantitatively analyzed various early timing conditions, and the result indicated that if the authors could realize appropriate movement prediction mechanism, the performance of assistance may be improved further. Considering the generality of timing issue in many related works, these results could lead to a new direction of research on performance improvement of human augmentation devices.
In this article, we propose sampled-data design of digital filters that cancel the continuous-time effect of coupling waves in a single-frequency full-duplex relay station. In this study, we model a relay station as a continuous-time system while conventional researches treat it as a discrete-time system. For a continuous-time model, we propose digital feedback canceler based on the sampled-data H∞ control theory to cancel coupling waves taking intersample behavior into account. We also propose robust control against unknown multipath interference. Simulation results are shown to illustrate the effectiveness of the proposed method.
The dynamic model for HVAC (Heating, Ventilating and Air Conditioning) electricity consumption is estimated for a campus building in Daido University. In order to construct energy management system for a community, buildings play one of the major role to control power balance in the community, because they consume a lot of energy and are a very predictable demand response player due to certain electricity load patterns. Then building dynamic characteristics should be known to control their energy consumption and to estimate the influence of the demand response. In order to observe the feature of the dynamic building energy consumption, electricity consumption data is measured in an autumn-winter period. Analyzing these data, the building characteristics are observed. The analysis concludes (1) the correlation between ambient temperature and HVAC electricity consumption, and (2) that building dynamics consist of a time lag with 1.6 hour time constant. Then the building model is proposed based on the observations and its parameters are estimated. The model agrees with the measurement data with less than 10% error.
This paper is concerned with local state synchronization of linear agents subject to input saturation over a fixed undirected communication graph. The author first derives a sufficient condition for locally achieving the synchronization via a relative state feedback control law. Based on this analysis result, the author presents a linear matrix inequality (LMI) condition for designing the synchronizing state feedback gain. The present LMI condition is scalable as long as the eigenvalues of the associated graph Laplacian are available, and is efficiently solved by an existing convex programming algorithm.
Design methods of inverse optimal adaptive consensus control of multi-agent systems composed of the first-order and the second-order regression models are presented based on H∞ control criterion. The proposed control schemes are deduced as solutions of certain H∞ control problems, where estimation errors of tuning parameters and imperfect knowledge of leaders are regarded as external disturbances to processes. The resulting control systems are shown to be robust to uncertain system parameters and the desirable consensus tracking is achieved asymptotically or approximately via adaptation schemes and L2-gain design parameters.
This paper deals with analysis of fundamental limitation in tracking control problem. The authors have given the tracking performance limitation of two degree of freedom systems for a class of reference signals, explicitly. This obtained result includes a uniform description of reference signals. In this paper, the result is extended to one degree of freedom systems. In the case of unstable plants, the tracking performance for one degree of freedom systems has not been analyzed except for the step reference. The analysis results of this paper clearly separate the contributions of the plant and the reference characteristics. The optimal performances are illustrated by numerical example with the sinusoidal reference.
A parallel feedforward model that improves estimates of unknown disturbances to non-minimum-phase systems is presented. Such a model must ensure the minimum-phase property of the augmented system, which consists of the plant and the parallel feedforward model. In addition, the frequency response of the parallel feedforward model must match that of the plant. This paper proposes a design procedure of the disturbance observer using a parallel feedforward model based on the Kalman-Yakubovich-Popov (KYP) lemma and linear matrix inequality (LMI) conditions. The effectiveness of this approach is verified by numerical simulations.