Search

In search, exploration, and reconnaissance operations of autonomous ground vehicles, an image recognition capability is needed for specifically classifying targeted objects (relevant classes) and at the same time identifying as unknown (irrelevant and unseen) objects that do not belong to any known classes, as opposed to falsely classifying them in one of the relevant classes. This paper integrates an unsupervised learning feature extraction framework based on the Instance Discrimination method with an Open-Set Low-Shot (IDLS) classifier for creating the desired new capability. Unlabeled images from the vehicle's operating environment are used for training the feature extractor while a modest number (less than 40) images for each relevant class and unlabeled irrelevant images are used for training the Open-Set Low-Shot (OSLS) classifier in a manner that enables recognition of images unseen during training as irrelevant. The value and the accuracy of the new IDLS approach are demonstrated through a thorough comparison with alternative unsupervised and fully supervised methods. [ABSTRACT FROM AUTHOR]

This paper presents a continuum-based design sensitivity analysis and optimization of high-frequency radiation problems using the energy finite element method and energy boundary element method. The noise radiated from the vibrating structure at a high-frequency range is obtained through a sequential procedure. The structural energy finite element method calculates structural energy distribution, which is then used as the boundary condition for the energy boundary element method to calculate the energy density at a far-field observation point. For design sensitivity analysis, the direct differentiation method calculates the sensitivity of the exterior noise through the sensitivity of the structural energy density obtained from the energy finite element method. The adjoint variable method calculates the adjoint load from an acoustic energy boundary element method reanalysis, and the adjoint response is obtained from a structural energy finite element method reanalysis. The sensitivity information is obtained by carrying out numerical integration only on the structural finite element part. The proposed design sensitivity analysis approach has been applied in the design of automotive and naval structures to search for the best material layout to achieve the lowest noise level at high frequency.

A probabilistic analysis is presented for studying the variation effects on the main bearing performance of an 1.C. engine system, under structural dynamic conditions. For computational efficiency, the probabilistic analysis is based on surrogate models (metamodels), which are developed using the kriging method. An optimum symmetric Latin hypercube algorithm is used for efficient 'space-filling' sampling of the design space. The metamodels provide an efficient and accurate substitute to the actual engine bearing simulation models. The bearing performance is based on a comprehensive engine system dynamic analysis which couples the flexible crankshaft and block dynamics with a detailed main bearing elastohydrodynamic analysis. The clearance of all main bearings and the oil viscosity comprise the random variables in the probabilistic analysis. The maximum oil pressure and the percentage of time within each cycle that a bearing operates with oil film thickness below a threshold value of 0.27 [micro]m at each main bearing constitute the system performance measures. Probabilistic analyses are first performed to calculate the mean, standard deviation and probability density function of the bearing performance measures. Subsequently, a probabilistic sensitivity analysis is described for identifying the important random variables. Finally, a reliability-based design optimization study is conducted for optimizing the main bearing performance under uncertainty. Results from a V6 engine are presented. [DOI: 10.1115/1.2000268]

A comprehensive formulation is presented for the dynamics of a rotating flexible crankshaft coupled with the dynamics of an engine block-through a finite difference elastohydrodynamic main bearing lubrication algorithm. The coupling is based on detailed equilibrium conditions at the bearings. The component mode synthesis is employed for modeling the crankshaft and block dynamic behavior. A soecialized algorithm for coupling the rigid and flexible body dynamics of the crankshaft within the framework of the component mode synthesis has been developed A finite difference lubrication algorithm is used for computing the oil film elastohydrodynamic characteristics. A computationally accurate and efficient mapping algorithm has been developed for transferring information between a high-density computational grid for the elastohydrodynamic bearing solver and a low-density structural grid utilized in computing the crankshaft and block structural dynamic response. The new computational capability is used to compute the vibratory response of an automotive V6 engine due to combustion and inertia loading. [DOI: 10.1115/1.1645297]

This article compares the numerical solutions of the new hybrid finite element formulation with those of closed-form analytical solutions for midfrequency range and bean assembly combinations. Results show consistent correlation between the two analyses.

Byline: Zhidong Zhang, Nickolas Vlahopoulos, T. Allen, K.Y. Zhang The Indirect Boundary Element Analysis is employed for developing a computational pass-by noise simulation capability. An acoustic field reconstruction process is developed in order to generate the definition of the main pass-by noise sources in a numerical model. Numerical boundary element models that characterise the individual sources are combined with a vehicle model for developing a system model for pass-by noise simulation. The numerical algorithm for defining generic sources is validated initially by employing analytical solutions. The numerical techniques are also validated through comparison between numerical results and test data for component level and system level analyses. Specifically, the generic source definition capability is validated by comparing the actual and the computationally reconstructed acoustic field for an engine intake manifold. The overall pass-by noise simulation capability is validated by computing the maximum overall sound pressure level for a vehicle under two different driving conditions.