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 new technique for locating simultaneous roots of a set of functions is described. The technique is based on the property of the Kreisselmeier-Steinhauser function which descends to a minimum at each root location. It is shown that the ensuing algorithm may be merged into any nonlinear programming method for solving optimization problems with equality constraints.
The nonlinear rescaling principle (NRP) consists of transforming the objective function and/or the constraints of a given constrained optimization problem into another problem which is equivalent to the original one in the sense that their optimal set of solutions coincides. A nonlinear transformation parameterized by a positive scalar parameter and based on a smooth scaling function is used to transform the constraints. The methods based on NRP consist of sequential unconstrained minimization of the classical Lagrangian for the equivalent problem, followed by an explicit formula updating the Lagrange multipliers. We first show that the NRP leads naturally to proximal methods with an entropy-like kernel, which is defined by the conjugate of the scaling function, and establish that the two methods are dually equivalent for convex constrained minimization problems. We then study the convergence properties of the nonlinear rescaling algorithm and the corresponding entropy-like proximal methods for convex constrained optimization problems. Special cases of the nonlinear resealing algorithm are presented. In particular a new class of exponential penalty-modified barrier functions methods is introduced.
The objective is to simulate and optimize distributed cooperation among a network of robots tasked with cooperative excavation on an extra-terrestrial surface. Additionally to examine the concept of directed Emergence among a group of limited artificially intelligent agents. Emergence is the concept of achieving complex results from very simple rules or interactions. For example, in a termite mound each individual termite does not carry a blueprint of how to make their home in a global sense, but their interactions based strictly on local desires create a complex superstructure. Leveraging this Emergence concept applied to a simulation of cooperative agents (robots) will allow an examination of the success of non-directed group strategy achieving specific results. Specifically the simulation will be a testbed to evaluate population based robotic exploration and cooperative strategies while leveraging the evolutionary teamwork approach in the face of uncertainty about the environment and partial loss of sensors. Checking against a cost function and 'social' constraints will optimize cooperation when excavating a simulated tunnel. Agents will act locally with non-local results. The rules by which the simulated robots interact will be optimized to the simplest possible for the desired result, leveraging Emergence. Sensor malfunction and line of sight issues will be incorporated into the simulation. This approach falls under Swarm Robotics, a subset of robot control concerned with finding ways to control large groups of robots. Swarm Robotics often contains biologically inspired approaches, research comes from social insect observation but also data from among groups of herding, schooling, and flocking animals. Biomimetic algorithms applied to manned space exploration is the method under consideration for further study.
There is currently renewed interest in robotic and crewed landers for a return to the lunar surface. Advanced guidance and navigation algorithms are essential to accurately delivering cargo and crew safely to the moon successfully. This paper reports the overall performance of an integrated set of navigation and guidance algorithms flown on a terrestrial suborbital rocket up to an altitude of approximately 100km. The navigation algorithm consists of an onboard extended Kalman Filter (EKF) that ingests multiple sensor measurements, one of which is the output from a terrain relative navigation (TRN) algorithm that cross-references camera images to on-board satellite imagery to perform feature correlation within the camera image. The guidance algorithm solves for a 6-degree-of-freedom (DoF) optimal trajectory using a successive convexification method during powered descent. The altitude range as well as the landing dynamics experienced during this test flight are realistic for an extraterrestrial landing and provide an invaluable data set to gauge the current development of these landing algorithms in an effort to advance the overall software readiness levels (SRL). This paper will delve into different aspects of each algorithm and present an analysis of the in-flight performance of the algorithms. This flight was conducted under the National Aeronautics and Space Administration (NASA) Safe and Precise Landing Integrated Capabilities Evolution (SPLICE) project focused on technology advancement for landing applications.
In this paper, the Manousiouthakis parametrization of all decentralized stabilizing controllers is employed in mathematically formulating the l(sup 1) optimal decentralized controller synthesis problem. The resulting optimization problem is infinite dimensional and therefore not directly amenable to computations. It is shown that finite dimensional optimization problems that have value arbitrarily close to the infinite dimensional one can be constructed. Based on this result, an algorithm that solves the l(sup 1) decentralized performance problems is presented. A global optimization approach to the solution of the infinite dimensional approximating problems is also discussed.
The results of optimal control theory are used to synthesize a feedback filter. The feedback filter is used to force the output of the filtered frequency response to match that of a desired optimal frequency response over a finite frequency range. This matching is accomplished by employing a nonlinear programing algorithm to search for the coefficients of the feedback filter that minimize the error between the optimal frequency response and the filtered frequency response. The method is applied to the synthesis of an active flutter-suppression control law for an aeroelastic wind-tunnel model. It is shown that the resulting control law suppresses flutter over a wide range of subsonic Mach numbers. This is a promising method for synthesizing practical control laws using the results of optimal control theory.
Traditional methods of control allocation optimization have shown difficulties in exploiting the full potential of controlling large arrays of control devices on innovative air vehicles. Artificial neutral networks are inspired by biological nervous systems and neurocomputing has successfully been applied to a variety of complex optimization problems. This project investigates the potential of applying neurocomputing to the control allocation optimization problem of Hybrid Wing Body (HWB) aircraft concepts to minimize control power, hinge moments, and actuator forces, while keeping system weights within acceptable limits. The main objective of this project is to develop a proof-of-concept process suitable to demonstrate the potential of using neurocomputing for optimizing actuation power for aircraft featuring multiple independently actuated control surfaces. A Nastran aeroservoelastic finite element model is used to generate a learning database of hinge moment and actuation power characteristics for an array of flight conditions and control surface deflections. An artificial neural network incorporating a genetic algorithm then uses this training data to perform control allocation optimization for the investigated aircraft configuration. The phase I project showed that optimization results for the sum of required hinge moments are improved by more than 12% over the best Nastran solution by using the neural network optimization process.
We present optimization results obtained for two type of antennas using evolutionary algorithms. A quadrifilar helical UHF antenna is currently flying aboard NASA's Mars Odyssey spacecraft and is due to reach final Martian orbit insertion in January, 2002. Using this antenna as a benchmark, we ran experiments employing a coevolutionary genetic algorithm to evolve a quadrifilar helical design in-situ - i.e., in the presence of a surrounding structure. Results show a 93% improvement at 400 MHz and a 48% improvement at 438 MHz in the average gain. The evolved antenna is also one-fourth the size. Yagi-Uda antennas are known to be difficult to design and optimize due to their sensitivity at high gain and the inclusion of numerous parasitic elements. Our fitness calculation allows the implicit relationship between power gain and sidelobe/backlobe loss to emerge naturally, a technique that is less complex than previous approaches. Our results include Yagi-Uda antennas that have excellent bandwidth and gain properties with very good impedance characteristics. Results exceeded previous Yagi-Uda antennas produced via evolutionary algorithms by at least 7.8% in mainlobe gain.
Limited spacecraft communication links call for users to investigate the potential use of video compression and multimedia technologies to optimize bandwidth allocations. The objective was to determine the transmission characteristics of multimedia data - motion video, text or bitmap graphics, and files transmitted independently and simultaneously over an ethernet local area network. Commercial desktop video teleconferencing hardware and software and Intel's proprietary Digital Video Interactive (DVI) video compression algorithm were used, and typical task scenarios were selected. The transmission time, packet size, number of packets, and network utilization of the data were recorded. Each data type - compressed motion video, text and/or bitmapped graphics, and a compressed image file - was first transmitted independently and its characteristics recorded. The results showed that an average bandwidth of 7.4 kilobits per second (kbps) was used to transmit graphics; an average bandwidth of 86.8 kbps was used to transmit an 18.9-kilobyte (kB) image file; a bandwidth of 728.9 kbps was used to transmit compressed motion video at 15 frames per second (fps); and a bandwidth of 75.9 kbps was used to transmit compressed motion video at 1.5 fps. Average packet sizes were 933 bytes for graphics, 498.5 bytes for the image file, 345.8 bytes for motion video at 15 fps, and 341.9 bytes for motion video at 1.5 fps. Simultaneous transmission of multimedia data types was also characterized. The multimedia packets used transmission bandwidths of 341.4 kbps and 105.8kbps. Bandwidth utilization varied according to the frame rate (frames per second) setting for the transmission of motion video. Packet size did not vary significantly between the data types. When these characteristics are applied to Space Station Freedom (SSF), the packet sizes fall within the maximum specified by the Consultative Committee for Space Data Systems (CCSDS). The uplink of imagery to SSF may be performed at minimal frame rates and/or within seconds of delay, depending on the user's allocated bandwidth. Further research to identify the acceptable delay interval and its impact on human performance is required. Additional studies in network performance using various video compression algorithms and integrated multimedia techniques are needed to determine the optimal design approach for utilizing SSF's data communications system.
Multi-Reward Proximal Policy Optimization (MRPPO) is a multi-objective rein- forcement learning algorithm used to construct low-thrust transfers between pe- riodic orbits in multi-body systems. Previous implementations of MRPPO have relied on a predefined reference transfer to successfully train each policy. In this paper, an algorithmic modification labeled the ‘moving reference’, is introduced to autonomously construct these reference trajectories during training. With this modification, MRPPO is used to recover various low-thrust transfers between two periodic orbits in the Earth-Moon circular restricted three-body problem to solve a multi-objective optimization problem. These results are then compared with the solutions recovered via a gradient descent optimization scheme to validate the performance of MRPPO with the moving reference modification.
Multi-Reward Proximal Policy Optimization (MRPPO) is a multi-objective rein- forcement learning algorithm used to construct low-thrust transfers between pe- riodic orbits in multi-body systems. Previous implementations of MRPPO have relied on a predefined reference transfer to successfully train each policy. In this paper, an algorithmic modification labeled the ‘moving reference’, is introduced to autonomously construct these reference trajectories during training. With this modification, MRPPO is used to recover various low-thrust transfers between two periodic orbits in the Earth-Moon circular restricted three-body problem to solve a multi-objective optimization problem. These results are then compared with the solutions recovered via a gradient descent optimization scheme to validate the performance of MRPPO with the moving reference modification.
Two distinct parametrization procedures of generating free-form surfaces to represent aerospace vehicles are presented. The first procedure is the representation using spline functions such as nonuniform rational b-splines (NURBS) and the second is a novel (geometrical) parametrization using solutions to a suitably chosen partial differential equation. The main idea is to develop a surface which is more versatile and can be used in an optimization process. Unstructured volume grid is generated by an advancing front algorithm and solutions obtained using an Euler solver. Grid sensitivity with respect to surface design parameters and aerodynamic sensitivity coefficients based on potential flow is obtained using an automatic differentiator precompiler software tool. Aerodynamic shape optimization of a complete aircraft with twenty four design variables is performed. High speed civil transport aircraft (HSCT) configurations are targeted to demonstrate the process.
On-orbit system identification (ID) of large space systems is essential for various reasons. For example, the complex composite structure of such systems cannot be ground-tested; their structural dynamic characteristics must be known accurately in order to accomplish active control. Furthermore, such capability can be used to characterize/identify various disturbances. The identification process is consisted of four principal elements: (1) modeling, (2) the estimation algorithm, (3) input system, and (4) measurement system. These elements are highly correlated and all togerher determine the success of the identification problem. Accurate modeling of large space systems is the most important element of the identification process. Large flexible structures are non-linear and infinite dimensional systems with highly coupled parameters and low frequency packed modes. In addition, these systems are subject to stochastic and time-varying disturbances, they have structural parameters which can vary due to on-orbit assembly deployment, and operations. These systems are generally; however, represented by constant coefficient, finite order differential equations. The non-linearities, coupling and noise effects are also often neglected. Moreover, identification experiment designs which lead to highly complex optimization problems usually require the simultaneous choice of ID algorithm, sensor, and actuator type and placement. On-orbit bandwidth and power restrictions on excitation, limited data window, and restrictions on sensor/actuator type, placement and number, has led to practical questions of implementations.
A new strategy for altering existing sector boundaries in response to blocking convective weather is presented. This method seeks to improve the reduced capacity of sectors directly affected by weather by moving boundaries in a direction that offers the greatest capacity improvement. The boundary deformations are shared by neighboring sectors within the region in a manner that preserves their shapes and sizes as much as possible. This reduces the controller workload involved with learning new sector designs. The algorithm that produces the altered sectors is based on a force-deflection mesh model that needs only nominal traffic patterns and the shape of the blocking weather for input. It does not require weather-affected traffic patterns that would have to be predicted by simulation. When compared to an existing optimal sector design method, the sectors produced by the new algorithm are more similar to the original sector shapes, resulting in sectors that may be more suitable for operational use because the change is not as drastic. Also, preliminary results show that this method produces sectors that can equitably distribute the workload of rerouted weather-affected traffic throughout the region where inclement weather is present. This is demonstrated by sector aircraft count distributions of simulated traffic in weather-affected regions.
The design and simulation of an on-line algorithm which estimates the safe maneuvering envelope of aircraft is discussed in this paper. The trim envelope is estimated using probabilistic methods and efficient high-fidelity model based computations of attainable equilibrium sets. From this trim envelope, a robust reachability analysis provides the maneuverability limitations of the aircraft through an optimal control formulation. Both envelope limits are presented to the flight crew on the primary flight display. In the results section, scenarios are considered where this adaptive algorithm is capable of computing online changes to the maneuvering envelope due to impairment. Furthermore, corresponding updates to display features on the primary flight display are provided to potentially inform the flight crew of safety critical envelope alterations caused by the impairment.
A procedure is presented which provides an algorithm for the special case where the unknown system has n inputs, and n finite length realizations for each of the n input signals are available. Given an n input system and n experimental records for each input, it is shown that the optimum Wiener filter is easily computed using only the experimental data. No assumptions concerning the analytical structure of the spectral matrix are necessary.
The problem of obtaining an optimal or near optimal schedule for scientific experiments to be performed on Spacelab missions is addressed. The current capabilities in this regard are examined and a method of ranking experiments in order of difficulty is developed to support the existing software. Experimental data is obtained from applying this method to the sets of experiments corresponding to Spacelab mission 1, 2, and 3. Finally, suggestions are made concerning desirable modifications and features of second generation software being developed for this problem.
In this paper, an algorithm is introduced to find a minimum phase transfer function of specified order whose magnitude "tightly" overbounds a specified real-valued nonparametric function of frequency. This method has direct application to transforming nonparametric uncertainty bounds (available from system identification experiments and/or plant modeling) into parametric representations required for modern robust control design software (i.e., a minimum-phase transfer function multiplied by a norm-bounded perturbation).