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.
Based on the intuition that is often easier to learn to solve difficult problems after similar, simpler problems have been learned, incremental evolution is a recently proposed extension to evolutionary algorithms in which the evaluation function against which a population is evolved is scaled over time.
A methodology for processing spectral images to retrieve information on underlying physical, chemical, and/or biological phenomena is based on evolutionary and related computational methods implemented in software. In a typical case, the solution (the information that one seeks to retrieve) consists of parameters of a mathematical model that represents one or more of the phenomena of interest. The methodology was developed for the initial purpose of retrieving the desired information from spectral image data acquired by remote-sensing instruments aimed at planets (including the Earth). Examples of information desired in such applications include trace gas concentrations, temperature profiles, surface types, day/night fractions, cloud/aerosol fractions, seasons, and viewing angles. The methodology is also potentially useful for retrieving information on chemical and/or biological hazards in terrestrial settings. In this methodology, one utilizes an iterative process that minimizes a fitness function indicative of the degree of dissimilarity between observed and synthetic spectral and angular data. The evolutionary computing methods that lie at the heart of this process yield a population of solutions (sets of the desired parameters) within an accuracy represented by a fitness-function value specified by the user. The evolutionary computing methods (ECM) used in this methodology are Genetic Algorithms and Simulated Annealing, both of which are well-established optimization techniques and have also been described in previous NASA Tech Briefs articles. These are embedded in a conceptual framework, represented in the architecture of the implementing software, that enables automatic retrieval of spectral and angular data and analysis of the retrieved solutions for uniqueness.
Asynchronous evolutionary algorithms are becoming increasingly popular as a means of making full use of many processors while solving computationally expensive search and optimization problems. These algorithms excel at keeping large clusters fully utilized, but may sometimes inefficiently sample an excess of fast-evaluating solutions at the expense of higher-quality, slow-evaluating ones. We introduce a steady-state parent selection strategy, SWEET (“Selection whilE EvaluaTing”), that sometimes selects individuals that are still being evaluated and allows them to reproduce early. This gives slow-evaluating individuals that have higher fitnesses an increased ability to multiply in the population. We find that SWEET appears effective in simulated take-over time analysis, but that its benefit is confined mostly to early in the run, and our preliminary study on an autonomous vehicle controller problem that involves tuning a spiking neural network proves inconclusive.
The most significant challenges facing the Space Transfer Vehicle (STV) program are making space basing affordable and developing an easily evolvable STV. Study efforts try minimizing the high cost of astronauts activities and Space Station modification to enhance affordability. Modularity, design margins, and simple interfaces are being evaluated to develop an evolutionary approach for space basing starting in the 1990's. In this STV study, the concept definition approach is presented, including concept optimization trades as well as concept selection and sensitivity analyses.
The paper presents the specific approach taken to formulate the problem in terms of gene encoding, fitness function, and genetic operations. The genome is encoded such that a subset of the scheduling constraints is automatically satisfied. Several fitness functions are formulated to emphasize different aspects of the scheduling problem. The optimal solutions of the different fitness functions demonstrate the trade-off of the scheduling problem and provide insight into a conflict resolution process.
An optimal active vibration absorber can provide guaranteed closed-loop stability and control for large flexible space structures with collocated sensors/actuators. The active vibration absorber is a second-order dynamic system which is designed to suppress any unwanted structural vibration. This can be designed with minimum knowledge of the controlled system. Two methods for optimizing the active vibration absorber parameters are illustrated: minimum resonant amplitude and frequency matched active controllers. The Controls-Structures Interaction Phase-1 Evolutionary Model at NASA LaRC is used to demonstrate the effectiveness of the active vibration absorber for vibration suppression. Performance is compared numerically and experimentally using acceleration feedback.
An optimal active vibration absorber can provide guaranteed closed-loop stability and control for large flexible space structures with collocated sensors/actuators. The active vibration absorber is a second-order dynamic system which is designed to suppress any unwanted structural vibration. This can be designed with minimum knowledge of the controlled system. Two methods for optimizing the active vibration absorber parameters are illustrated: minimum resonant amplitude and frequency matched active controllers. The Controls-Structures Interaction Phase-1 Evolutionary Model at NASA LaRC is used to demonstrate the effectiveness of the active vibration absorber for vibration suppression. Performance is compared numerically and experimentally using acceleration feedback.
This paper explores impediments to innovation in aerospace and suggests how successful pathways from other industries can be adopted to facilitate greater innovation. Because of its nature, space exploration would seem to be a ripe field of technical innovation. However, engineering can also be a frustratingly conservative endeavor when the realities of cost and risk are included. Impediments like the "find the fault" engineering culture, the treatment of technical risk as almost always evaluated in terms of negative impact, the difficult to account for expansive Moore's Law growth when making predictions, and the stove-piped structural organization of most large aerospace companies and federally funded research laboratories tend to inhibit cross-cutting technical innovation. One successful example of a multi-use cross cutting application that can scale with Moore's Law is the Evolutionary Computational Methods (ECM) technique developed at the Jet Propulsion Lab for automated spectral retrieval. Future innovations like computational engineering and automated design optimization can potentially redefine space exploration, but will require learning lessons from successful innovators.
Continual learning, the capability to learn new knowledge from streaming data without forgetting the previous knowledge, is a critical requirement for dynamic learning systems, especially for emerging edge devices such as self-driving cars and drones. However, continual learning is still facing the catastrophic forgetting problem. Previous work illustrate that model performance on continual learning is not only related to the learning algorithms but also strongly dependent on the inherited model, i.e., the model where continual learning starts. The better stability of the inherited model, the less catastrophic forgetting and thus, the inherited model should be elaborately selected. Inspired by this finding, we develop an evolutionary neural architecture search (ENAS) algorithm that emphasizes the Stability of the inherited model, namely ENAS-S. ENAS-S aims to find optimal architectures for accurate continual learning on edge devices. On CIFAR-10 and CIFAR-100, we present that ENAS-S achieves competitive architectures with lower catastrophic forgetting and smaller model size when learning from a data stream, as compared with handcrafted DNNs.
The propulsion system of rotorcraft vehicles is the most critical system to the vehicle in terms of safety and performance. The propulsion system must provide both vertical lift and forward flight propulsion during the entire mission. Whereas propulsion is a critical element for all flight vehicles, it is particularly critical for rotorcraft due to their limited safe, un-powered landing capability. This unparalleled reliability requirement has led rotorcraft power plants down a certain evolutionary path in which the system looks and performs quite similarly to those of the 1960 s. By and large the advancements in rotorcraft propulsion have come in terms of safety and reliability and not in terms of performance. The concept of the optimized propulsion system is a means by which both reliability and performance can be improved for rotorcraft vehicles. The optimized rotorcraft propulsion system which couples an oil-free turboshaft engine to a highly loaded gearbox that provides axial load support for the power turbine can be designed with current laboratory proven technology. Such a system can provide up to 60% weight reduction of the propulsion system of rotorcraft vehicles. Several technical challenges are apparent at the conceptual design level and should be addressed with current research.
Close-in transiting sub-Neptunes are abundant in our Galaxy. Planetary interior models based on their observed radius–mass relationship suggest that sub-Neptunes contain a discernible amount of either hydrogen (dry planets) or water (wet planets) blanketing a core composed of rocks and metal. Water-rich sub-Neptunes have been believed to form farther from the star and then migrate inwards to their present orbits. Here we report experimental evidence of reactions between warm, dense hydrogen fluid and silicate melt that release silicon from the magma to form alloys and hydrides at high pressures. We found that oxygen liberated from the silicate melt reacts with hydrogen, producing an appreciable amount of water up to a few tens of weight per cent, which is much greater than previously predicted based on low-pressure ideal gas extrapolation. Consequently, these reactions can generate a spectrum of water contents in hydrogen-rich planets, with the potential to reach water-rich compositions for some sub-Neptunes, implying an evolutionary relationship between hydrogen-rich and water-rich planets. Therefore, detection of a large amount of water in exoplanet atmospheres may not be the optimal evidence for planet migration in the protoplanetary disk, calling into question the assumed link between composition and planet formation location.
Single neuron models are fundamental for computational modeling of the brain's neuronal networks, and understanding how ion channel dynamics mediate neural function. A challenge in defining such models is determining biophysically realistic channel distributions. Here, we present an efficient, highly parallel evolutionary algorithm for developing such models, named NeuroGPU-EA. NeuroGPU-EA uses CPUs and GPUs concurrently to simulate and evaluate neuron membrane potentials with respect to multiple stimuli. We demonstrate a logarithmic cost for scaling the stimuli used in the fitting procedure. NeuroGPU-EA outperforms the typically used CPU based evolutionary algorithm by a factor of 10 on a series of scaling benchmarks. We report observed performance bottlenecks and propose mitigation strategies. Finally, we also discuss the potential of this method for efficient simulation and evaluation of electrophysiological waveforms.
Probability Collectives (PC) provides the information-theoretic extension of conventional full-rationality game theory to bounded rational games. Here an explicit solution to the equations giving the bounded rationality equilibrium of a game is presented. Then PC is used to investigate games in which the players use bounded rational best-response strategies. Next it is shown that in the continuum-time limit, bounded rational best response games result in a variant of the replicator dynamics of evolutionary game theory. It is then shown that for team (shared-payoff) games, this variant of replicator dynamics is identical to Newton-Raphson iterative optimization of the shared utility function.
Kill switches provide a biocontainment strategy in which unwanted growth of an engineered microorganism is prevented by expression of a toxin gene. A major challenge in kill switch engineering is balancing evolutionary stability with robust cell killing activity in application relevant host strains. Understanding host-specific containment dynamics and modes of failure helps to develop potent yet stable kill switches. To guide the design of robust kill switches in the agriculturally relevant strain Pseudomonas fluorescens SBW25, we present a comparison of lethality, stability, and genetic escape of eight different toxic effectors in the presence of their cognate inactivators (i.e., toxin–antitoxin modules, polymorphic exotoxin–immunity systems, restriction endonuclease–methyltransferase pair). We find that cell killing capacity and evolutionary stability are inversely correlated and dependent on the level of protection provided by the inactivator gene. Decreasing the proteolytic stability of the inactivator protein can increase cell killing capacity, but at the cost of long-term circuit stability. By comparing toxins within the same genetic context, we determine that modes of genetic escape increase with circuit complexity and are driven by toxin activity, the protective capacity of the inactivator, and the presence of mutation-prone sequences within the circuit. Here, the results of our study reveal that circuit complexity, toxin choice, inactivator stability, and DNA sequence design are powerful drivers of kill switch stability and valuable targets for optimization of biocontainment systems.
Conceptual efforts to develop a Space Station based system for docking and/or berthing the NSTS Orbiter are described. Past docking and berthing systems are reviewed, the general requirements and options for mating the Orbiter and Space Station are discussed, and the rationale for locating the system on the Station is established. One class of Station-based system is developed in several variations and evaluated with respect to weight distribution, loads, safety, reliability, viewing, and maintainability. An evolutionary presentation of the variations provides insight into the development process and the problems encountered. An overall evaluation of the Station-based variations compared to an optimized Orbiter-based system demonstrates the potential benefits of this approach as well as the issues that must be resolved to realize the benefits.
Abstract Leaf CO 2 uptake ( A ) in C 4 photosynthesis is limited by the maximum apparent rate of PEPc carboxylation ( V pmax ) at low intercellular [CO 2 ] ( c i ) with a sharp transition to a c i ‐saturated rate ( V max ) due to co‐limitation by ribulose‐1:5‐bisphosphate carboxylase/oxygenase (Rubisco) and regeneration of PEP. The response of A to c i has been widely used to determine these two parameters. V max and V pmax depend on different enzymes but draw on a shared pool of leaf resources, such that resource distribution is optimized, and A maximized, when V max and V pmax are co‐limiting. We collected published A / c i curves in 49 C 4 species and assessed variation in photosynthetic traits between phylogenetic groups, and as a function of atmospheric [CO 2 ]. The balance of V max ‐ V pmax varied among evolutionary lineages and C 4 subtypes. Operating A was strongly V max ‐limited, such that re‐allocation of resources from V pmax towards V max was predicted to improve A by 12% in C 4 crops. This would not require additional inputs but rather altered partitioning of existing leaf nutrients, resulting in increased water and nutrient‐use efficiency. Optimal partitioning was achieved only in plants grown at pre‐industrial atmospheric [CO 2 ], suggesting C 4 crops have not adjusted to the rapid increase in atmospheric [CO 2 ] of the past few decades.
We present a simple multi-dimensional exhaustive search method to obtain, in a reasonable time, the optimal solution of a nonlinear programming problem. It is more relevant in the present day non-mainframe computing scenario where an estimated 95% computing resources remains unutilized and computing speed touches petaflops. While the processor speed is doubling every 18 months, the band width is doubling every 12 months, and the hard disk space is doubling every 9 months. A randomized search algorithm or, equivalently, an evolutionary search method is often used instead of an exhaustive search algorithm. The reason is that a randomized approach is usually polynomial-time, i.e., fast while an exhaustive search method is exponential-time i.e., slow. We discuss the increasing importance of exhaustive search in optimization with the steady increase of computing power for solving many real-world problems of reasonable size. We also discuss the computational error and complexity of the search algorithm focusing on the fact that no measuring device can usually measure a quantity with an accuracy greater than 0.005%. We stress the fact that the quality of solution of the exhaustive search - a deterministic method - is better than that of randomized search. In 21 st century computing environment, exhaustive search cannot be left aside as an untouchable and it is not always exponential. We also describe a possible application of these algorithms in improving the efficiency of solar cells - a real hot topic - in the current energy crisis. These algorithms could be excellent tools in the hands of experimentalists and could save not only large amount of time needed for experiments but also could validate the theory against experimental results fast.
Satellites provide essential services such as networking and weather tracking, and the number of near-earth and deep space satellites are expected to grow rapidly in the coming years. Communications with terrestrial ground stations is one of the critical functionalities of any space mission. Satellite scheduling is a problem that has been scientifically investigated since the 1970s. A central aspect of this problem is the need to consider resource contention and satellite visibility constraints as they require line of sight. Due to the combinatorial nature of the problem, prior solutions such as linear programs and evolutionary algorithms require extensive compute capabilities to output a feasible schedule for each scenario. Machine learning based scheduling can provide an alternative solution by training a model with historical data and generating a schedule quickly with model inference. We present SatNet, a benchmark for satellite scheduling optimization based on historical data from the NASA Deep Space Network. We propose formulation of the satellite scheduling problem as a Markov Decision Process and use reinforcement learning (RL) policies to generate schedules. The nature of constraints imposed by SatNet differ from other combinatorial optimization problems such as vehicle routing studied in prior literature. Our initial results indicate that RL is an alternative optimization approach that can generate candidate solutions of comparable quality to existing state-of-the-practice results. However, we also find that RL policies overfit to the training dataset and do not generalize well to new data, thereby necessitating continued research on reusable and generalizable agents.