基于RBF神经网络的非线性观察器设计

提出了一种新的非线性观察器设计方法。与一般方法采用神经网络逼近整个非线性系统不同，该方法用RBF神经网络逼近系统的非线性项，故提高了状态估计的精度。基于李亚普诺夫方法，证明了状态估计误差渐近稳定且渐近收敛到零。仿真结果表明，所提出的非线性观察器设计方法具有良好的性能。在故障检测、状态估计等领域具有广泛的应用前景。     

Intermediate-term periodicities in relativistic solar electron fluences during solar cycles 22 and 23

In this paper, we have investigated the intermediate-term periodicities of the relativistic (E > 10 MeV) solar electron flares measured by IMP-8 satellite of NASA for the time period of 1986–2001. This period of investigation includes the entire solar cycle 22; ascending, maximum and a part of descending phase of the current solar cycle 23. To determine accurately the occurrence rate of electron flux, we have employed three different spectral decomposition techniques, viz. fast Fourier transformation (FFT); maximum entropy method (MEM) and Lomb–Scargle periodogram analysis method. For solar cycle 22, in the low frequency range, power spectrum analysis exhibits statistically significant periodicities at ∼706, ∼504 and ∼392 days. In the intermediate frequency range, we have found a series of significant periodicities ∼294, ∼221, ∼153, ∼86, ∼73 and ∼66 days. For short term, periodicities of ∼21–23, ∼31 and ∼37 days were found in power spectrum. When solar cycle 23 is considered the significant periodicities are ∼20, ∼23, ∼29, ∼39, ∼54, ∼63, ∼118, ∼133 and ∼154 days. These results provide evidence that the best known Rieger period (∼153 days), appeared in the high energetic electron flux data for cycle 22 and also likely during maxima of cycle 23. The existence of these periodicities has been discussed in the light of earlier results.     

Comptonization efficiencies of the variability classes of GRS 1915 + 105

The Galactic microquasar GRS 1915 + 105 exhibits at least seventeen types of variability classes. Intra and inter class transitions are reported to be observed within seconds to hours. Since the observation was not continuous, these classes appeared to be exhibited in a random order. Our goal is to predict a sequence of these classes. In this paper, we compute the ratio of the photon counts obtained from the power-law component and the blackbody component of each class and call this ratio as the ‘Comptonizing efficiency’ (CE) of that class. We sequence the classes in the ascending order of CE and find that this sequence matches with a few class transitions observed by RXTE satellite and IXAE instruments on board IRS-P3. A change in CE corresponds to a change in the optical depth of the Compton cloud. Our result implies that the optical depth of the Compton cloud gradually rises as the variability class becomes harder.     

Humans to Mars: a feasibility and cost-benefit analysis

Mars is a compelling astrobiological target, and a human mission would provide an opportunity to collect immense amounts of scientific data. Exploration alone, however, cannot justify the increased risk. Instead, three factors drive a human mission: economics, education, and exploration. A human mission has a unique potential to inspire the next generation of young people to enter critically needed science and engineering disciplines. A mission is economically feasible, and the research and development program put in place for a human mission would propel growth in related high-technology industries. The main hurdles are human physiological responses to 1–2 years of radiation and microgravity exposure. However, enabling technologies are sufficiently mature in these areas that they can be developed within a few decade timescale. Hence, the decision of whether or not to undertake a human mission to Mars is a political decision, and thus, educational and economic benefits are the crucial factors.     

Helicopter aeroelastic analysis with spatially uncertain rotor blade properties

This paper investigates the effects of spatially uncertain material properties on the aeroelastic response predictions (e.g., rotating frequencies, vibratory loads, etc.) of composite helicopter rotor. Initially, the spatial uncertainty is modeled as discrete random variables along the blade span and uncertainty analysis is performed with direct Monte Carlo simulations (MCS). Uncertainty effects on the rotating frequencies vary with the higher-order modes in a non-linear way. Each modal frequency is found to be more sensitive to the uncertainty at certain sections of the rotor blade than uncertainty at other sections. Uncertainty effects on the vibratory hub load predictions are studied in the next stage. To reduce the computational expense of stochastic aeroelastic analysis, a high-dimensional model representation (HDMR) method is developed to approximate the aeroelastic response as functions of blade stiffness properties which are modeled as random fields. Karhunen–Loève expansion and a lower-order expansion are used to represent the input and outputs, respectively, in the HDMR formulation which is similar to the spectral stochastic finite element method. The proposed method involves the approximation of the system response with lower-dimensional HDMR, the response surface generation of HDMR component functions, and Monte Carlo simulation. The proposed approach decouples the computationally expensive aeroelastic simulations and uncertainty analysis. MCS, performed with computationally less expensive HDMR models, shows that spatial uncertainty has considerable influence on the vibratory hub load predictions.     

Optimal Design of Induction Generators for Space Applications

A high frequency electromechanical energy conversion option for space power applications is introduced. An induction generator (IG) operating at high frequencies is an excellent option for obtaining high power and energy densities. Various design aspects affect the operating characteristics of the machine. The effects and consequences of specific trade-offs in the design process are explored. The methodology is helpful in evaluating specific choices for materials as well as machine configurations. Use of power semiconductor devices and other power electronic devices is avoided to provide immunity from radiation and electromagnetic interference.