Advanced Materials Research Vols. 143-144

Abstract: The dynamic model of Cournot game with bounded rationality in power market is established. The different dynamic behaviors are analyzed, such as Nash equilibrium, period and chaos. The economic performances of dynamic behaviors are analyzed and compared for duopoly dynamic Cournot game. For the chaotic market status which has lower economic efficiency, the parameter perturbation control method is proposed. Finally, by means of the proposed control method, the chaotic state is controlled to the stable equilibrium point so that the economic performance of power market is effectively improved.

Abstract: Viscosity-temperature effect is an important factor for journal bearing calculation, but many scholars ignored the effect of viscosity-temperature, which may produce some error. The simulation of journal bearing oil film flow field is done using CFD, and the viscosity-temperature and cavitation effect have been consider. The results shows that the capacity and temperature rise of the oil film will be overrated when using the constant viscosity of oil. The vapor distribution and static characteristics of the journal bearing have been calculate at last.

Abstract: This paper makes a research on dynamic fingerprints methods and model updating methods of identification theories of bridge damage. And it has a full study on and makes a comparison among the four basic theories of dynamic fingerprints and their applications in engineering, signifying the advantages and disadvantages. The general procedures of damage identification of model updating method are proposed. And finally it points out that the problems needing further studies and the directions of development in damage identification theories.

Abstract: Intelligent computing is the application of advanced computing methods to improve performance in areas such as complex representations that are clear to users and easily modifiable. The goal is to make decision making more reliable, spontaneous and creative. In this paper, we put forwards a novel heuristic genetic algorithm based on simulated annealing strategy. Experimental results show the effectiveness of the proposed algorithm.

Abstract: The ability of magnetorheological(MR) fluids to resist compression in the direction of magnetic field was tested to investigate the quasi-static squeezing process of MR fluids. A experiment setup was designed and fabricated to test the compression characteristics. Under the different magnetic flux density, the curves of the MR fluids were studied for yield stress versus compression stress, compression stress versus compression strain. The compression resistance of MRF was then measured for the different magnetic flux density for comparison with the shear yield strength of the same magnetic field. The results showed that the compression resistance of the MRF was much stronger than its shear yield stress for the same magnetic field strength conditions. The compression resistance of MRF can be utilized to design new magnetorheological devices.

Abstract: A method for solving the fault diagnosis problem of air brake system based on probabilistic approach is presented. The fault diagnosis model based on Bayesian network was built for the uncertainty characteristic of fault in the air brake system. Through evaluating the characteristic of Bayesian networks in the diagnosis inference and model expression, it is demonstrated that this method can solve the uncertain problems in fault diagnosis. The test result has shown that the Bayesian network model is effective in fault diagnosis of the air brake system.

Abstract: Dynamic Bayesian networks can be well dealt with the time-varying multivariable problem. The state model based on Dynamic Bayesian networks can more accurately describe the relationship between the system state and the influencing factors. In this paper, the width of the reasoning is used to simplify the amount of data in the reasoning process. Multi-step state prediction is achieved by extending time-slice. Experiment has shown that the proposed algorithm can achieve better prediction results.

Abstract: Cement-based nanocomposite filled with only 0.1 wt.% multi-wall carbon nanotube (MWNT) (MNFRC), or hybrid with 0.1 wt.% MWNT and 0.5 wt.% microsized short carbon fiber (SCF) (SF/MNFRC), were prepared employing surfactant ultrasonic dispersion and high-speed mixing process. The electrical resistivities (ρ), compressive stresses (σ), and longitudinal strains (εl) of these cured nanocomposites under cyclic uploading/unloading were simultaneously collected, to characterize their stress/strain-sensitive properties. There exists good piezoresisitivity and high strain sensitivity for MNFRC. The fractional change in ρ (Δρ) regularly descends or ascends following the σ, or the εl of MNFRC (the Δρ/εl sensitivity near 68). Hybrid of SCF is found to be more effective to improve the self-sensing repeatability and variation stability of the SF/MNFRC rather than the self-sensing sensitivity (the Δρ/εl sensitivity only about 64).

Abstract: Cementitious composite filled with only 0.1 wt% multi-walled carbon nanotube (MWCNT) (MNTCC), or hybrid with 0.1 wt% MWCNT and 0.5 wt% nanophase carbon black (NCB) (NB/MNTCC), were prepared employing surfactant ultrasonic dispersion and subsequently high-speed mixing process. The electrical resistivities (ρ), compressive stresses (σ), and longitudinal strains (εl) of these cured nanocomposites under cyclic loading (10 times) were simultaneously collected, to investigate their piezoresistivty properties. Results revealed that, hybrid of NCB is very effective for enhancement on the stress/strain sensitivity of NB/MNTCC (the strain sensitivity up to 107, much higher than that of MNTCC 67), but there is almost no improvement on the self-sensing repeatability and variation stability of the corresponding nanocomposite.

Abstract: We present a novel online unsupervised anomaly detection method for human activities. The proposed approach is based on one-class support vector machine (OCSVM) clustering, where the novelty detection SVM capabilities are used for the identification of anomalous activities. Particular attention is given to activity classification in absence of a priori information on the distribution of outliers. Activities are represented by variable-length event sequences, but the most commonly used kernels are defined on fixed-dimension spaces. To solve the problem, we develop a novel sequence-similarity kernel, the n-grams kernel. Our kernel is conceptually simple and efficient to compute and performs well in comparison with state-of-the-art methods for anomaly detection. Moreover, most SVM algorithms require large number of memory to store the kernel matrix, or repeated access to the training samples. This makes it infeasible for online anomaly detection. In this paper, we develop simple and computationally efficient online learning algorithms for anomaly detection.