SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
A method for aerodynamic shape optimization based on a genetic algorithm approach is demonstrated. The algorithm is coupled with a transonic full potential flow solver and is used to optimize the flow about transonic wings including multi-objective solutions that lead to the generation of pareto fronts. The results indicate that the genetic algorithm is easy to implement, flexible in application and extremely reliable.
Wind tunnels use scale models to characterize aerodynamic coefficients, Wind tunnel testing can be slow and costly due to high personnel overhead and intensive power utilization. Although manual curve fitting can be done, it is highly efficient to use a neural network to define the complex relationship between variables. Numerical simulation of complex vehicles on the wide range of conditions required for flight simulation requires static and dynamic data. Static data at low Mach numbers and angles of attack may be obtained with simpler Euler codes. Static data of stalled vehicles where zones of flow separation are usually present at higher angles of attack require Navier-Stokes simulations which are costly due to the large processing time required to attain convergence. Preliminary dynamic data may be obtained with simpler methods based on correlations and vortex methods; however, accurate prediction of the dynamic coefficients requires complex and costly numerical simulations. A reliable and fast method of predicting complex aerodynamic coefficients for flight simulation I'S presented using a neural network. The training data for the neural network are derived from numerical simulations and wind-tunnel experiments. The aerodynamic coefficients are modeled as functions of the flow characteristics and the control surfaces of the vehicle. The basic coefficients of lift, drag and pitching moment are expressed as functions of angles of attack and Mach number. The modeled and training aerodynamic coefficients show good agreement. This method shows excellent potential for rapid development of aerodynamic models for flight simulation. Genetic Algorithms (GA) are used to optimize a previously built Artificial Neural Network (ANN) that reliably predicts aerodynamic coefficients. Results indicate that the GA provided an efficient method of optimizing the ANN model to predict aerodynamic coefficients. The reliability of the ANN using the GA includes prediction of aerodynamic coefficients to an accuracy of 110% . In our problem, we would like to get an optimized neural network architecture and minimum data set. This has been accomplished within 500 training cycles of a neural network. After removing training pairs (outliers), the GA has produced much better results. The neural network constructed is a feed forward neural network with a back propagation learning mechanism. The main goal has been to free the network design process from constraints of human biases, and to discover better forms of neural network architectures. The automation of the network architecture search by genetic algorithms seems to have been the best way to achieve this goal.
Many software systems have been successfully implemented using an agent paradigm which employs a number of independent entities that communicate with one another to achieve a common goal. The distributed nature of such a paradigm makes it an excellent candidate for use in high speed reconfigurable computing hardware environments such as those present in modem FPGA's. In this paper, a distributed genetic algorithm that can be applied to the agent based reconfigurable hardware model is introduced. The effectiveness of this new algorithm is evaluated by comparing the quality of the solutions found by the new algorithm with those found by traditional genetic algorithms. The performance of a reconfigurable hardware implementation of the new algorithm on an FPGA is compared to traditional single processor implementations.
A genetic-algorithm-based optimization approach is described and evaluated using a simple hill-climbing model problem. The model problem utilized herein allows for the broad specification of a large number of search spaces including spaces with an arbitrary number of genes or decision variables and an arbitrary number hills or modes. In the present study, only single objective problems are considered. Results indicate that the genetic algorithm optimization approach is flexible in application and extremely reliable, providing optimal results for all problems attempted. The most difficult problems - those with large hyper-volumes and multi-mode search spaces containing a large number of genes - require a large number of function evaluations for GA convergence, but they always converge.
A fast, reliable way of predicting aerodynamic coefficients is produced using a neural network optimized by a genetic algorithm. Basic aerodynamic coefficients (e.g. lift, drag, pitching moment) are modelled as functions of angle of attack and Mach number. The neural network is first trained on a relatively rich set of data from wind tunnel tests of numerical simulations to learn an overall model. Most of the aerodynamic parameters can be well-fitted using polynomial functions. A new set of data, which can be relatively sparse, is then supplied to the network to produce a new model consistent with the previous model and the new data. Because the new model interpolates realistically between the sparse test data points, it is suitable for use in piloted simulations. The genetic algorithm is used to choose a neural network architecture to give best results, avoiding over-and under-fitting of the test data.
This paper investigates the feasibility of applying genetic algorithms to solve optimization problems that are implemented entirely in reconfgurable hardware. The paper highlights the pe$ormance/design space trade-offs that must be understood to effectively implement a standard genetic algorithm within a modem Field Programmable Gate Array, FPGA, reconfgurable hardware environment and presents a case-study where this stochastic search technique is applied to standard test-case problems taken from the technical literature. In this research, the targeted FPGA-based platform and high-level design environment was the Starbridge Hypercomputing platform, which incorporates multiple Xilinx Virtex II FPGAs, and the Viva TM graphical hardware description language.
A novel, real-valued Genetic Algorithm (GA) was designed and implemented to minimize the reflectivity and/or transmissivity of an arbitrary number of homogeneous, lossy dielectric or magnetic layers of arbitrary thickness positioned at either the center of an infinitely long rectangular waveguide, or adjacent to the perfectly conducting backplate of a semi-infinite, shorted-out rectangular waveguide. Evolutionary processes extract the optimal physioelectric constants falling within specified constraints which minimize reflection and/or transmission over the frequency band of interest. This GA extracted the unphysical dielectric and magnetic constants of three layers of fictitious material placed adjacent to the conducting backplate of a shorted-out waveguide such that the reflectivity of the configuration was 55 dB or less over the entire X-band. Examples of the optimization of realistic multi-layer absorbers are also presented. Although typical Genetic Algorithms require populations of many thousands in order to function properly and obtain correct results, verified correct results were obtained for all test cases using this GA with a population of only four.
The placement of actuators on a wing determines the control effectiveness of the airplane. One approach to placement maximizes the moments about the pitch, roll, and yaw axes, while minimizing the coupling. For example, the desired actuators produce a pure roll moment without at the same time causing much pitch or yaw. For a typical wing, there is a large set of candidate locations for placing actuators, resulting in a substantially larger number of combinations to examine in order to find an optimum placement satisfying the mission requirements and mission constraints. A genetic algorithm has been developed for finding the best placement for four actuators to produce an uncoupled pitch moment. The genetic algorithm has been extended to find the minimum number of actuators required to provide uncoupled pitch, roll, and yaw control. A simplified, untapered, unswept wing is the model for each application.
Our anatomical and behavioral studies of embryonic rats that developed in microgravity suggest that the vestibular sensory system, like the visual system, has genetically mediated processes of development that establish crude connections between the periphery and the brain. Environmental stimuli also regulate connection formation including terminal branch formation and fine-tuning of synaptic contacts. Axons of vestibular sensory neurons from gravistatic as well as linear acceleration receptors reach their targets in both microgravity and normal gravity, suggesting that this is a genetically regulated component of development. However, microgravity exposure delays the development of terminal branches and synapses in gravistatic but not linear acceleration-sensitive neurons and also produces behavioral changes. These latter changes reflect environmentally controlled processes of development.
Numerous problems exist with the current thinking of RNA as the first genetic material. No plausible prebiotic processes have yet been demonstrated to produce the nucleosides or nucleotides or for efficient two-way nonenzymatic replication. Peptide nucleic acid (PNA) is a promising precursor to RNA, consisting of N-(2-aminoethyl)glycine (AEG) and the adenine, uracil, guanine, and cytosine-N-acetic acids. However, PNA has not yet been demonstrated to be prebiotic. We show here that AEG is produced directly in electric discharge reactions from CH(4), N(2), NH(3), and H(2)O. Electric discharges also produce ethylenediamine, as do NH(4)CN polymerizations. AEG is produced from the robust Strecker synthesis with ethylenediamine. The NH(4)CN polymerization in the presence of glycine leads to the adenine and guanine-N(9)-acetic acids, and the cytosine and uracil-N(1)-acetic acids are produced in high yield from the reaction of cyanoacetaldehyde with hydantoic acid, rather than urea. Preliminary experiments suggest that AEG may polymerize rapidly at 100 degrees C to give the polypeptide backbone of PNA. The ease of synthesis of the components of PNA and possibility of polymerization of AEG reinforce the possibility that PNA may have been the first genetic material.
The purpose of this paper is to utilize the genetic algorithm (GA) optimization method for structural design of a nuclear propulsion vehicle. Genetic algorithms provide a guided, random search technique that mirrors biological adaptation. To verify the GA capabilities, other traditional optimization methods were used to generate results for comparison to the GA results, first for simple two-dimensional structures, and then for full-scale three-dimensional truss designs.
Nuclear electric propulsion (NEP) vehicles will be needed for future manned missions to Mars and beyond. Candidate designs must be identified for further detailed design from a large array of possibilities. Genetic algorithms have proven their utility in conceptual design studies by effectively searching a large design space to pinpoint unique optimal designs. This research combined analysis codes for NEP subsystems with a genetic algorithm. The use of penalty functions with scaling ratios was investigated to increase computational efficiency. Also, the selection of design variables for optimization was considered to reduce computation time without losing beneficial design search space. Finally, trend analysis of a reference mission to the asteroids yielded a group of candidate designs for further analysis.
As part of the NASA Aviation Safety Program, a unique model-based diagnostics method that employs neural networks and genetic algorithms for aircraft engine performance diagnostics has been developed and demonstrated at the NASA Glenn Research Center against a nonlinear gas turbine engine model. Neural networks are applied to estimate the internal health condition of the engine, and genetic algorithms are used for sensor fault detection, isolation, and quantification. This hybrid architecture combines the excellent nonlinear estimation capabilities of neural networks with the capability to rank the likelihood of various faults given a specific sensor suite signature. The method requires a significantly smaller data training set than a neural network approach alone does, and it performs the combined engine health monitoring objectives of performance diagnostics and sensor fault detection and isolation in the presence of nominal and degraded engine health conditions.
We compare Genetic Algorithms (GA's) with Probability Collectives (PC), a new framework for distributed optimization and control. In contrast to GA's, PC-based methods do not update populations of solutions. Instead they update an explicitly parameterized probability distribution p over the space of solutions. That updating of p arises as the optimization of a functional of p. The functional is chosen so that any p that optimizes it should be p peaked about good solutions. The PC approach works in both continuous and discrete problems. It does not suffer from the resolution limitation of the finite bit length encoding of parameters into GA alleles. It also has deep connections with both game theory and statistical physics. We review the PC approach using its motivation as the information theoretic formulation of bounded rationality for multi-agent systems. It is then compared with GA's on a diverse set of problems. To handle high dimensional surfaces, in the PC method investigated here p is restricted to a product distribution. Each distribution in that product is controlled by a separate agent. The test functions were selected for their difficulty using either traditional gradient descent or genetic algorithms. On those functions the PC-based approach significantly outperforms traditional GA's in both rate of descent, trapping in false minima, and long term optimization.
Generic algorithms have gained popularity as an effective procedure for obtaining solutions to traditionally difficult space mission optimization problems. In this paper, a brief survey of the use of genetic algorithms to solve astrodynamics problems is presented and is followed by new results obtained from applying a Pareto genetic algorithm to the optimization of low-thrust interplanetary spacecraft missions.
This work predicts whether a retrieval algorithm will usefully determine CO2 concentration from an input spectrum of GOSAT (Greenhouse Gases Observing Satellite). This was done to eliminate needless runtime on atmospheric soundings that would never yield useful results. A space of 50 dimensions was examined for predictive power on the final CO2 results. Retrieval algorithms are frequently expensive to run, and wasted effort defeats requirements and expends needless resources. This algorithm could be used to help predict and filter unneeded runs in any computationally expensive regime. Traditional methods such as the Fischer discriminant analysis and decision trees can attempt to predict whether a sounding will be properly processed. However, this work sought to detect a subsection of the dimensional space that can be simply filtered out to eliminate unwanted runs. LDAs (linear discriminant analyses) and other systems examine the entire data and judge a "best fit," giving equal weight to complex and problematic regions as well as simple, clear-cut regions. In this implementation, a genetic space of "left" and "right" thresholds outside of which all data are rejected was defined. These left/right pairs are created for each of the 50 input dimensions. A genetic algorithm then runs through countless potential filter settings using a JPL computer cluster, optimizing the tossed-out data s yield (proper vs. improper run removal) and number of points tossed. This solution is robust to an arbitrary decision boundary within the data and avoids the global optimization problem of whole-dataset fitting using LDA or decision trees. It filters out runs that would not have produced useful CO2 values to save needless computation. This would be an algorithmic preprocessing improvement to any computationally expensive system.
In this paper, genetic algorithm based and gradient-based topology optimization is presented in application to a real hardware design problem. Preliminary design of a planetary lander mockup structure is accomplished using these methods that prove to provide major weight savings by addressing the structural efficiency during the design cycle. This paper presents two alternative formulations of the topology optimization problem. The first is the widely-used gradient-based implementation using commercially available algorithms. The second is formulated using genetic algorithms and internally developed capabilities. These two approaches are applied to a practical design problem for hardware that has been built, tested and proven to be functional. Both formulations converged on similar solutions and therefore were proven to be equally valid implementations of the process. This paper discusses both of these formulations at a high level.
In this paper, a unique system level spacecraft design optimization will be presented. A Genetic Algorithm is used to design the global pattern of the reinforcing structure, while a gradient routine is used to adequately stiffen the sub-structure. The system level structural design includes determining the optimal physical location (and number) of reinforcing beams of a lunar pallet lander deck structure. Design of the substructure includes determining placement of secondary stiffeners and the number of rivets required for assembly.. In this optimization, several considerations are taken into account. The primary objective was to raise the primary natural frequencies of the structure such that the Pallet Lander primary structure does not significantly couple with the launch vehicle. A secondary objective is to determine how to properly stiffen the reinforcing beams so that the beam web resists the shear buckling load imparted by the spacecraft components mounted to the pallet lander deck during launch and landing. A third objective is that the calculated stress does not exceed the allowable strength of the material. These design requirements must be met while, minimizing the overall mass of the spacecraft. The final paper will discuss how the optimization was implemented as well as the results. While driven by optimization algorithms, the primary purpose of this effort was to demonstrate the capability of genetic algorithms to enable design automation in the preliminary design cycle. By developing a routine that can automatically generate designs through the use of Finite Element Analysis, considerable design efficiencies, both in time and overall product, can be obtained over more traditional brute force design methods.