SEARCH · Search NASA
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.
The biological motivation for Artificial Neural Net developments is briefly discussed, and the most popular paradigm, the feedforward supervised learning net with error back propagation training algorithm, is introduced. Possible approaches for utilization in structural optimization is illustrated through simple examples. Other currently ongoing developments for application in structural mechanics are also mentioned.
The inverse dynamic equation of a flexible manipulator was solved in the time domain. By dividing the inverse system equation into the causal part and the anticausal part, we calculated the torque and the trajectories of all state variables for a given end point trajectory. The interpretation of this method in the frequency domain was explained in detail using the two-sided Laplace transform and the convolution integral. The open loop control of the inverse dynamic method shows an excellent result in simulation. For real applications, a practical control strategy is proposed by adding a feedback tracking control loop to the inverse dynamic feedforward control, and its good experimental performance is presented.
The low gravity environment provided by space flight has afforded the science community a unique area for the study of fundamental and technological sciences. However, the dynamic environment observed on space shuttle flights and predicted for Space Station Freedom has complicated the analysis of prior 'microgravity' experiments and prompted concern for the viability of proposed space experiments requiring long term, low gravity environments. Thus, isolation systems capable of providing significant improvements to this random environment have been developed. This dissertation deals with the design constraints imposed by acceleration sensitive, microgravity experiment payloads in the unique environment of space. A theoretical background for the inertial feedback and feedforward isolation of a payload was developed giving the basis for two experimental active inertial isolation systems developed for the demonstration of these advanced active isolation techniques. A prototype six degree of freedom digital active isolation system was designed and developed for the ground based testing of an actively isolated payload in three horizontal degrees of freedom. A second functionally equivalent system was built for the multi-dimensional testing of an active inertial isolation system in a reduced gravity environment during low gravity aircraft trajectories. These multi-input multi-output control systems are discussed in detail with estimates on acceleration noise floor performance as well as the actual performance acceleration data. The attenuation performance is also given for both systems demonstrating the advantages between inertial and non-inertial control of a payload for both the ground base environment and the low gravity aircraft acceleration environment. A future goal for this area of research is to validate the technical approaches developed to the 0.01 Hz regime by demonstrating a functional active inertial feedforward/feedback isolation system during orbital flight. A NASA IN-STEP flight experiment has been proposed to accomplish this goal, and the expected selection for the IN-STEP program has been set for Jul. of 1993.
This report examines the management of human error in the cockpit. The principles probably apply as well to other applications in the aviation realm (e.g. air traffic control, dispatch, weather, etc.) as well as other high-risk systems outside of aviation (e.g. shipping, high-technology medical procedures, military operations, nuclear power production). Management of human error is distinguished from error prevention. It is a more encompassing term, which includes not only the prevention of error, but also a means of disallowing an error, once made, from adversely affecting system output. Such techniques include: traditional human factors engineering, improvement of feedback and feedforward of information from system to crew, 'error-evident' displays which make erroneous input more obvious to the crew, trapping of errors within a system, goal-sharing between humans and machines (also called 'intent-driven' systems), paperwork management, and behaviorally based approaches, including procedures, standardization, checklist design, training, cockpit resource management, etc. Fifteen guidelines for the design and implementation of intervention strategies are included.
A feedforward-based amplifier linearization technique is presented which is capable of yielding significant improvements in both linearity and power efficiency over conventional amplifier classes (e.g. class-A or class-AB). Theoretical and practical results are presented showing that class-C stages may be used for both the main and error amplifiers yielding practical efficiencies well in excess of 30 percent, with theoretical efficiencies of much greater than 40 percent being possible. The levels of linearity which may be achieved are required for most satellite systems, however if greater linearity is required, the technique may be used in addition to conventional pre-distortion techniques.
Upcoming NASA missions will require tracking of low-orbit satellites. As a consequence, NASA antennas will be required to track satellites at higher rates than for the current deep space missions. This article investigates servo design issues for the 34-m beam-waveguide antennas that track low-orbit satellites. This includes upgrading the servo with a feedforward loop, using a monopulse controller design, and reducing tracking errors through either proper choice of elevation pinion location, application of a notch filter, or adjustment of the elevation drive amplifier gain. Finally, improvement of the signal-to-noise ratio through averaging of the over-sampled monopulse signal is described.
The purpose of this research program is to design, build and test (in cooperation with NASA personnel from the NASA Langley Research Center) neural controllers for two different small air-gap magnetic suspension systems. The general objective of the program is to study neural network architectures for the purpose of control in an experimental setting and to demonstrate the feasibility of the concept. The specific objectives of the research program are: (1) to demonstrate through simulation and experimentation the feasibility of using neural controllers to stabilize a nonlinear magnetic suspension system; (2) to investigate through simulation and experimentation the performance of neural controllers designs under various types of parametric and nonparametric uncertainty; (3) to investigate through simulation and experimentation various types of neural architectures for real-time control with respect to performance and complexity; and (4) to benchmark in an experimental setting the performance of neural controllers against other types of existing linear and nonlinear compensator designs. To date, the first one-dimensional, small air-gap magnetic suspension system has been built, tested and delivered to the NASA Langley Research Center. The device is currently being stabilized with a digital linear phase-lead controller. The neural controller hardware is under construction. Two different neural network paradigms are under consideration, one based on hidden layer feedforward networks trained via back propagation and one based on using Gaussian radial basis functions trained by analytical methods related to stability conditions. Some advanced nonlinear control algorithms using feedback linearization and sliding mode control are in simulation studies.
This invention provides a new hierarchical approach for supervised neural learning of time dependent trajectories. The modular hierarchical methodology leads to architectures which are more structured than fully interconnected networks. The networks utilize a general feedforward flow of information and sparse recurrent connections to achieve dynamic effects. The advantages include the sparsity of units and connections, the modular organization. A further advantage is that the learning is much more circumscribed learning than in fully interconnected systems. The present invention is embodied by a neural network including a plurality of neural modules each having a pre-established performance capability wherein each neural module has an output outputting present results of the performance capability and an input for changing the present results of the performance capabilitiy. For pattern recognition applications, the performance capability may be an oscillation capability producing a repeating wave pattern as the present results. In the preferred embodiment, each of the plurality of neural modules includes a pre-established capability portion and a performance adjustment portion connected to control the pre-established capability portion.
A gain-scheduling neural network architecture is proposed to enhance the noise-filtering efficiency of feedforward neural networks, in terms of both nominal performance and robustness. The synergistic benefits of the proposed architecture are demonstrated and discussed in the context of the noise-filtering of signals that are typically encountered in aerospace control systems. The synthesis of such a gain-scheduled neurofiltering provides the robustness of linear filtering, while preserving the nominal performance advantage of conventional nonlinear neurofiltering. Quantitative performance and robustness evaluations are provided for the signal processing of pitch rate responses to typical pilot command inputs for a modern fighter aircraft model.
A new vehicle following controller is proposed for autonomous intelligent vehicles. The proposed vehicle following controller not only provides smooth transient maneuvers for unavoidable nonzero initial conditions but also guarantees the asymptotic platoon stability without the availability of feedforward information. Furthermore, the achieved asymptotic platoon stability is shown to be robust to sensor delays and an upper bound for the allowable sensor delays is also provided in this paper.
We present an architecture of an intelligent restructurable control system to automatically detect failure of system components, assess its impact on system performance and safety, and reconfigure the controller for performance recovery. Fault detection is based on neural network associative memories and pattern classifiers, and is implemented using a multilayer feedforward network. Details of the fault detection network along with simulation results on health monitoring of a dc motor have been presented. Conceptual developments for fault assessment using an expert system and controller reconfiguration using a neural network are outlined.
An analytical model has been developed to predict the resulting far field radiation from a turbofan engine inlet. A feedforward control algorithm was simulated to predict the controlled far field radiation from the destructive combination of fan noise and secondary control sources. Numerical results were developed for two system configurations, with the resulting controlled far field radiation patterns showing varying degrees of attenuation and spillover. With one axial station of twelve control sources and error sensors with equal relative angular positions, nearly global attenuation is achieved. Shifting the angular position of one error sensor resulted in an increase of spillover to the extreme sidelines. The complex control inputs for each configuration was investigated to identify the structure of the wave pattern created by the control sources, giving an indication of performance of the system configuration. It is deduced that the locations of the error sensors and the control source configuration are equally critical to the operation of the active noise control system.
The advent of high resolution, physical model based computer graphics has left a gap in the design of input/output technology appropriate for interacting with such complex virtual world models. Since virtual worlds consist of physical models, it is appropriate to output the inherent force information necessary for the simulation to the user. The detailed design, construction, and control of a three degree freedom force output joystick will be presented. A novel kinematic design allows all three axes to be uncoupled, so that the system inertia matrix is diagonal. The two planar axes are actuated through an offset gimbal, and the third through a sleeved cable. To compensate for friction and inertia effects, this transmission is controlled by a force feedforward and a closed force feedback proportional loop. Workspace volume is a cone of 512 cubic inches, and the device bandwidth is maximized at 60 Hz for the two planar and 30 Hz for the third axis. Each axis is controlled by a motor/proportional magnetic particle brake combination fixed to the base. The innovative use of motors and brakes allows objects with high resistive torque requirements to be simulated without the stability and related safety issues involved with high torque, energy storing motors alone. Position, velocity, and applied endpoint force are sensed directly. Different control strategies are discussed and implemented, with an emphasis on how virtual environment force information, generated by the MIT Media Lab Computer Graphics and Animation Group BOLIO system, is transmitted to the device controller. The design constraints for a kinesthetic force feedback device can be summarized as: How can the symbiosis between the sense of presence in the virtual environment be maximized without compromising the interaction task under the constraints of the mechanical device limitations? Research in this field will yield insights to the optimal human sensory feedback mix for a wide spectrum of control and interaction problems. A flexible research tool that is designed as an easily reproducible product prototype has been constructed to explore the variety of possible force interaction.
Electronic and optoelectronic hardware implementations of highly parallel computing architectures address several ill-defined and/or computation-intensive problems not easily solved by conventional computing techniques. The concurrent processing architectures developed are derived from a variety of advanced computing paradigms including neural network models, fuzzy logic, and cellular automata. Hardware implementation technologies range from state-of-the-art digital/analog custom-VLSI to advanced optoelectronic devices such as computer-generated holograms and e-beam fabricated Dammann gratings. JPL's concurrent processing devices group has developed a broad technology base in hardware implementable parallel algorithms, low-power and high-speed VLSI designs and building block VLSI chips, leading to application-specific high-performance embeddable processors. Application areas include high throughput map-data classification using feedforward neural networks, terrain based tactical movement planner using cellular automata, resource optimization (weapon-target assignment) using a multidimensional feedback network with lateral inhibition, and classification of rocks using an inner-product scheme on thematic mapper data. In addition to addressing specific functional needs of DOD and NASA, the JPL-developed concurrent processing device technology is also being customized for a variety of commercial applications (in collaboration with industrial partners), and is being transferred to U.S. industries. This viewgraph p resentation focuses on two application-specific processors which solve the computation intensive tasks of resource allocation (weapon-target assignment) and terrain based tactical movement planning using two extremely different topologies. Resource allocation is implemented as an asynchronous analog competitive assignment architecture inspired by the Hopfield network. Hardware realization leads to a two to four order of magnitude speed-up over conventional techniques and enables multiple assignments, (many to many), not achievable with standard statistical approaches. Tactical movement planning (finding the best path from A to B) is accomplished with a digital two-dimensional concurrent processor array. By exploiting the natural parallel decomposition of the problem in silicon, a four order of magnitude speed-up over optimized software approaches has been demonstrated.
At McDonnell Douglas Aerospace (MDA), an artificial neural network based control system has been developed and implemented to control laser heating for the fiber placement composite manufacturing process. This neurocontroller learns an approximate inverse model of the process on-line to provide performance that improves with experience and exceeds that of conventional feedback control techniques. When untrained, the control system behaves as a proportional plus integral (PI) controller. However after learning from experience, the neural network feedforward control module provides control signals that greatly improve temperature tracking performance. Faster convergence to new temperature set points and reduced temperature deviation due to changing feed rate have been demonstrated on the machine. A Cerebellar Model Articulation Controller (CMAC) network is used for inverse modeling because of its rapid learning performance. This control system is implemented in an IBM compatible 386 PC with an A/D board interface to the machine.
A hybrid neurocomputing/numerical strategy is presented for geometrically nonlinear analysis of structures. The strategy combines model-free data processing capabilities of computational neural networks with a Pade approximants-based perturbation technique to predict partial information about the nonlinear response of structures. In the hybrid strategy, multilayer feedforward neural networks are used to extend the validity of solutions by using training samples produced by Pade approximations to the Taylor series expansion of the response function. The range of validity of the training samples is taken to be the radius of convergence of Pade approximants and is estimated by setting a tolerance on the diverging approximants. The norm of residual vector of unbalanced forces in a given element is used as a measure to assess the quality of network predictions. To further increase the accuracy and the range of network predictions, additional training data are generated by either applying linear regression to weight matrices or expanding the training data by using predicted coefficients in a Taylor series. The effectiveness of the hybrid strategy is assessed by performing large-deflection analysis of a doubly-curved composite panel with a circular cutout, and postbuckling analyses of stiffened composite panels subjected to an in-plane edge shear load. In all the problems considered, the hybrid strategy is used to predict selective information about the structural response, namely the total strain energy and the maximum displacement components only.
Optical data processing techniques have the inherent advantage of high data throughout, low weight and low power requirements. These features are particularly desirable for onboard spacecraft in-situ real-time data analysis and data compression applications. the proposed multi-layer optical holographic neural net pattern recognition technique will utilize the nonlinear photorefractive devices for real-time adaptive learning to classify input data content and recognize unexpected features. Information can be stored either in analog or digital form in a nonlinear photofractive device. The recording can be accomplished in time scales ranging from milliseconds to microseconds. When a system consisting of these devices is organized in a multi-layer structure, a feedforward neural net with bifurcating data classification capability is formed. The interdisciplinary research will involve the collaboration with top digital computer architecture experts at the University of Southern California.
Space applications for telerobotics requires a manipulator that is robust to the large payload variations that occur in a zero-g environment. One approach to provide the required performance for this large payload variation is to provide inertial decoupling to the controller. The equations to generate the required joint torques can be obtained in several manners but the independent parameters required to perform the calculations are difficult to obtain. This presentation consists of an overview of the inertial decoupling control system and discusses how to reduce the inertial parameter set to a minimum and obtain the required parameter set to a minimum and obtain the required parameter vector using sensed joint positions and torques. In addition, as telerobotic technology becomes more prevalent in the space environments the ability to emulate the effect of zero-g in a one-g environment is critical in learning how to perform tasks in space. To emulate the zero-g environment it is necessary to remove the effect of gravity on the manipulator by determining the feedforward joint torques. A subset of the algorithm developed to obtain the decoupling parameters is used to obtain the gravity offload parameters. These parameters are used to eliminate the effect of gravity. The algorithm is validated on Martin Marietta/NASA Langley's, 7 Degrees-Of-Freedom (DOF) Flight Telerobotic Servicer Hydraulic Manipulator Test Bed. A video portraying actual results of the algorithm is provided.