Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Iterative”

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.

At least 631 records · Page 35

A Method to Solve Interior and Exterior Camera Calibration Parameters for Image Resection

An iterative method is presented to solve the internal and external camera calibration parameters, given model target points and their images from one or more camera locations. The direct linear transform formulation was used to obtain a guess for the iterative method, and herein lies one of the strengths of the present method. In all test cases, the method converged to the correct solution. In general, an overdetermined system of nonlinear equations is solved in the least-squares sense. The iterative method presented is based on Newton-Raphson for solving systems of nonlinear algebraic equations. The Jacobian is analytically derived and the pseudo-inverse of the Jacobian is obtained by singular value decomposition.

Samtaney, Ravi↗

Parallel Preconditioning for CFD Problems on the CM-5

Up to today, preconditioning methods on massively parallel systems have faced a major difficulty. The most successful preconditioning methods in terms of accelerating the convergence of the iterative solver such as incomplete LU factorizations are notoriously difficult to implement on parallel machines for two reasons: (1) the actual computation of the preconditioner is not very floating-point intensive, but requires a large amount of unstructured communication, and (2) the application of the preconditioning matrix in the iteration phase (i.e. triangular solves) are difficult to parallelize because of the recursive nature of the computation. Here we present a new approach to preconditioning for very large, sparse, unsymmetric, linear systems, which avoids both difficulties. We explicitly compute an approximate inverse to our original matrix. This new preconditioning matrix can be applied most efficiently for iterative methods on massively parallel machines, since the preconditioning phase involves only a matrix-vector multiplication, with possibly a dense matrix. Furthermore the actual computation of the preconditioning matrix has natural parallelism. For a problem of size n, the preconditioning matrix can be computed by solving n independent small least squares problems. The algorithm and its implementation on the Connection Machine CM-5 are discussed in detail and supported by extensive timings obtained from real problem data.

Simon, Horst D.↗

A Modified Wall Matching Treatment to Account for Local Solid to Fluid Thermal Coupling

The wall-matching methodology of Wilcox is modified to include a solid-wall, thermal-conduction model. This coupled fluid-thermal-structure model is derived assuming that the wall thermal-structure behavior is locally one-dimensional and that structural deformations, due to thermally induced stresses, are not significant. The one-dimensional coupled fluid-thermal-structure model is derived such that the wall temperature is removed as an independent boundary condition variable. The one-dimensional coupled fluid-thermal-structure model is also derived for the general case of an arbitrary mixture of thermally prefect gases and a wall of arbitrary thickness and conductivity by using a compressible, streamwise-pressure-gradient-corrected, wall-matching function and Fourier's law of heat conduction. The resulting model was implemented in the VULCAN CFD code as a new boundary condition type. VULCAN was then used to simulate a two-dimensional Mach 6 wind tunnel facility nozzle flow to demonstrate/validate the one-dimensional coupled fluid-thermal-structure model. The nozzle internal-wall surface temperature and heat transfer distributions computed using the one-dimensional coupled fluid-thermal-structure model are compared to wall temperature and heat transfer distributions from an iterative multi-dimensional analysis obtained by coupling the VULCAN CFD code and the MSC/NASTRAN-thermal code. The one-dimensional coupled fluid-thermal-structure model analysis is shown to be very robust and in excellent agreement with the multi-dimensional iteratively coupled analysis. It is also shown that the one-dimensional analysis can be used as an initial guess for the multi-dimensional iteratively coupled analysis.

White, Jeffery A.↗

Glenn-ht/bem Conjugate Heat Transfer Solver for Large-scale Turbomachinery Models

A coupled Boundary Element/Finite Volume Method temperature-forward/flux-hack algorithm is developed for conjugate heat transfer (CHT) applications. A loosely coupled strategy is adopted with each field solution providing boundary conditions for the other in an iteration seeking continuity of temperature and heat flux at the fluid-solid interface. The NASA Glenn Navier-Stokes code Glenn-HT is coupled to a 3-D BEM steady state heat conduction code developed at the University of Central Florida. Results from CHT simulation of a 3-D film-cooled blade section are presented and compared with those computed by a two-temperature approach. Also presented are current developments of an iterative domain decomposition strategy accommodating large numbers of unknowns in the BEM. The blade is artificially sub-sectioned in the span-wise direction, 3-D BEM solutions are obtained in the subdomains, and interface temperatures are averaged symmetrically when the flux is updated while the fluxes are averaged anti-symmetrically to maintain continuity of heat flux when the temperatures are updated. An initial guess for interface temperatures uses a physically-based 1-D conduction argument to provide an effective starting point and significantly reduce iteration. 2-D and 3-D results show the process converges efficiently and offers substantial computational and storage savings. Future developments include a parallel multi-grid implementation of the approach under MPI for computation on PC clusters.

Divo, E.↗

Software Estimates Costs of Testing Rocket Engines

Simulation-Based Cost Model (SiCM) is a computer program that simulates pertinent aspects of the testing of rocket engines for the purpose of estimating the costs of such testing during time intervals specified by its users. A user enters input data for control of simulations; information on the nature of, and activity in, a given testing project; and information on resources. Simulation objects are created on the basis of this input. Costs of the engineering-design, construction, and testing phases of a given project are estimated from numbers and labor rates of engineers and technicians employed in each phase, the duration of each phase; costs of materials used in each phase; and, for the testing phase, the rate of maintenance of the testing facility. The three main outputs of SiCM are (1) a curve, updated at each iteration of the simulation, that shows overall expenditures vs. time during the interval specified by the user; (2) a histogram of the total costs from all iterations of the simulation; and (3) table displaying means and variances of cumulative costs for each phase from all iterations. Other outputs include spending curves for each phase.

Source record↗

Reducing Design Cycle Time and Cost Through Process Resequencing

In today's competitive environment, companies are under enormous pressure to reduce the time and cost of their design cycle. One method for reducing both time and cost is to develop an understanding of the flow of the design processes and the effects of the iterative subcycles that are found in complex design projects. Once these aspects are understood, the design manager can make decisions that take advantage of decomposition, concurrent engineering, and parallel processing techniques to reduce the total time and the total cost of the design cycle. One software tool that can aid in this decision-making process is the Design Manager's Aid for Intelligent Decomposition (DeMAID). The DeMAID software minimizes the feedback couplings that create iterative subcycles, groups processes into iterative subcycles, and decomposes the subcycles into a hierarchical structure. The real benefits of producing the best design in the least time and at a minimum cost are obtained from sequencing the processes in the subcycles.

Rogers, James L.↗

A Modified Wall Matching Treatment to Account for Local Solid to Fluid Thermal Coupling

The wall-matching methodology of Wilcox is modified to include a solid-wall, thermal-conduction model. This coupled fluid-thermal-structure model is derived assuming that the wall thermal-structure behavior is locally one-dimensional and that structural deformations, due to thermally induced stresses, are not significant. The one-dimensional coupled fluid-thermal-structure model is derived such that the wall temperature is removed as an independent boundary condition variable. The one-dimensional coupled fluid-thermal-structure model is also derived for the general case of an arbitrary mixture of thermally prefect gases and a wall of arbitrary thickness and conductivity by using a compressible, streamwise-pressure-gradient-corrected, wall-matching function and Fourier's law of heat conduction. The resulting model was implemented in the VULCAN CFD code as a new boundary condition type. VULCAN was then used to simulate a two-dimensional Mach 6 wind tunnel facility nozzle flow to demonstrate/validate the one-dimensional coupled fluid-thermal-structure model. The nozzle internal-wall surface temperature and heat transfer distributions computed using the one-dimensional coupled fluid-thermal-structure model are compared to wall temperature and heat transfer distributions from an iterative multi-dimensional analysis obtained by coupling the VULCAN CFD code and the MSC/NASTRAN-thermal code. The one-dimensional coupled fluid-thermal-structure model analysis is shown to be very robust and in excellent agreement with the multi-dimensional iteratively coupled analysis. It is also shown that the one-dimensional analysis can be used as an initial guess for the multi-dimensional iteratively coupled analysis.

White, Jeffery A.↗

On-board autonomous attitude maneuver planning for planetary spacecraft using genetic algorithms

A key enabling technology that leads to greater spacecraft autonomy is the capability to autonomously and optimally slew the spacecraft from and to different attitudes while operating under a number of celestial and dynamic constraints. The task of finding an attitude trajectory that meets all the constraints is a formidable one, in particular for orbiting or fly-by spacecraft where the constraints and initial and final conditions are of time-varying nature. This paper presents an approach for attitude path planning that makes full use of a priori constraint knowledge and is computationally tractable enough to be executed on-board a spacecraft. The approach is based on incorporating the constraints into a cost function and using a Genetic Algorithm to iteratively search for and optimize the solution. This results in a directed random search that explores a large part of the solution space while maintaining the knowledge of good solutions from iteration to iteration. A solution obtained this way may be used 'as is' or as an initial solution to initialize additional deterministic optimization algorithms. A number of example simulations are presented including the case examples of a generic Europa Orbiter spacecraft in cruise as well as in orbit around Europa. The search times are typically on the order of minutes, thus demonstrating the viability of the presented approach. The results are applicable to all future deep space missions where greater spacecraft autonomy is required. In addition, onboard autonomous attitude planning greatly facilitates navigation and science observation planning, benefiting thus all missions to planet Earth as well.

genetic algorithm↗

Human Factors Assessment and Redesign of the ISS Respiratory Support Pack (RSP) Cue Card

The Respiratory Support Pack (RSP) is a medical pack onboard the International Space Station (ISS) that contains much of the necessary equipment for providing aid to a conscious or unconscious crewmember in respiratory distress. Inside the RSP lid pocket is a 5.5 by 11 inch paper procedural cue card, which is used by a Crew Medical Officer (CMO) to set up the equipment and deliver oxygen to a crewmember. In training, crewmembers expressed concerns about the readability and usability of the cue card; consequently, updating the cue card was prioritized as an activity to be completed. The Usability Testing and Analysis Facility at the Johnson Space Center (JSC) evaluated the original layout of the cue card, and proposed several new cue card designs based on human factors principles. The approach taken for the assessment was an iterative process. First, in order to completely understand the issues with the RSP cue card, crewmember post training comments regarding the RSP cue card were taken into consideration. Over the course of the iterative process, the procedural information was reorganized into a linear flow after the removal of irrelevant (non-emergency) content. Pictures, color coding, and borders were added to highlight key components in the RSP to aid in quickly identifying those components. There were minimal changes to the actual text content. Three studies were conducted using non-medically trained JSC personnel (total of 34 participants). Non-medically trained personnel participated in order to approximate a scenario of limited CMO exposure to the RSP equipment and training (which can occur six months prior to the mission). In each study, participants were asked to perform two respiratory distress scenarios using one of the cue card designs to simulate resuscitation (using a mannequin along with the hardware). Procedure completion time, errors, and subjective ratings were recorded. The last iteration of the cue card featured a schematic of the RSP, colors, borders, and simplification of the flow of information. The time to complete the RSP procedure was reduced by approximately three minutes with the new design. In an emergency situation, three minutes significantly increases the probability of saving a life. In addition, participants showed the highest preference for this design. The results of the studies and the new design were presented to a focus group of astronauts, flight surgeons, medical trainers, and procedures personnel. The final cue card was presented to a medical control board and approved for flight. The revised RSP cue card is currently onboard ISS.

Byrne, Vicky↗

Wavefront Control Toolbox for James Webb Space Telescope Testbed

We have developed a Matlab toolbox for wavefront control of optical systems. We have applied this toolbox to the optical models of James Webb Space Telescope (JWST) in general and to the JWST Testbed Telescope (TBT) in particular, implementing both unconstrained and constrained wavefront optimization to correct for possible misalignments present on the segmented primary mirror or the monolithic secondary mirror. The optical models implemented in Zemax optical design program and information is exchanged between Matlab and Zemax via the Dynamic Data Exchange (DDE) interface. The model configuration is managed using the XML protocol. The optimization algorithm uses influence functions for each adjustable degree of freedom of the optical mode. The iterative and non-iterative algorithms have been developed to converge to a local minimum of the root-mean-square (rms) of wavefront error using singular value decomposition technique of the control matrix of influence functions. The toolkit is highly modular and allows the user to choose control strategies for the degrees of freedom to be adjusted on a given iteration and wavefront convergence criterion. As the influence functions are nonlinear over the control parameter space, the toolkit also allows for trade-offs between frequency of updating the local influence functions and execution speed. The functionality of the toolbox and the validity of the underlying algorithms have been verified through extensive simulations.

Shiri, Ron↗

Protograph LDPC Codes with Node Degrees at Least 3

In this paper we present protograph codes with a small number of degree-3 nodes and one high degree node. The iterative decoding threshold for proposed rate 1/2 codes are lower, by about 0.2 dB, than the best known irregular LDPC codes with degree at least 3. The main motivation is to gain linear minimum distance to achieve low error floor. Also to construct rate-compatible protograph-based LDPC codes for fixed block length that simultaneously achieves low iterative decoding threshold and linear minimum distance. We start with a rate 1/2 protograph LDPC code with degree-3 nodes and one high degree node. Higher rate codes are obtained by connecting check nodes with degree-2 non-transmitted nodes. This is equivalent to constraint combining in the protograph. The condition where all constraints are combined corresponds to the highest rate code. This constraint must be connected to nodes of degree at least three for the graph to have linear minimum distance. Thus having node degree at least 3 for rate 1/2 guarantees linear minimum distance property to be preserved for higher rates. Through examples we show that the iterative decoding threshold as low as 0.544 dB can be achieved for small protographs with node degrees at least three. A family of low- to high-rate codes with minimum distance linearly increasing in block size and with capacity-approaching performance thresholds is presented. FPGA simulation results for a few example codes show that the proposed codes perform as predicted.

low density parity check (LDPC)↗

The ALI-ARMS Code for Modeling Atmospheric non-LTE Molecular Band Emissions: Current Status and Applications

The Accelerated Lambda Iteration (ALI) technique was developed in stellar astrophysics at the beginning of 1990s for solving the non-LTE radiative transfer problem in atomic lines and multiplets in stellar atmospheres. It was later successfully applied to modeling the non-LTE emissions and radiative cooling/heating in the vibrational-rotational bands of molecules in planetary atmospheres. Similar to the standard lambda iterations ALI operates with the matrices of minimal dimension. However, it provides higher convergence rate and stability due to removing from the iterating process the photons trapped in the optically thick line cores. In the current ALI-ARMS (ALI for Atmospheric Radiation and Molecular Spectra) code version additional acceleration of calculations is provided by utilizing the opacity distribution function (ODF) approach and "decoupling". The former allows replacing the band branches by single lines of special shape, whereas the latter treats non-linearity caused by strong near-resonant vibration-vibrational level coupling without additional linearizing the statistical equilibrium equations. Latest code application for the non-LTE diagnostics of the molecular band emissions of Earth's and Martian atmospheres as well as for the non-LTE IR cooling/heating calculations are discussed.

Kutepov, A. A.↗

Board Saver for Use with Developmental FPGAs

A device denoted a board saver has been developed as a means of reducing wear and tear of a printed-circuit board onto which an antifuse field programmable gate array (FPGA) is to be eventually soldered permanently after a number of design iterations. The need for the board saver or a similar device arises because (1) antifuse-FPGA design iterations are common and (2) repeated soldering and unsoldering of FPGAs on the printed-circuit board to accommodate design iterations can wear out the printed-circuit board. The board saver is basically a solderable/unsolderable FPGA receptacle that is installed temporarily on the printed-circuit board. The board saver is, more specifically, a smaller, square-ring-shaped, printed-circuit board (see figure) that contains half via holes one for each contact pad along its periphery. As initially fabricated, the board saver is a wider ring containing full via holes, but then it is milled along its outer edges, cutting the via holes in half and laterally exposing their interiors. The board saver is positioned in registration with the designated FPGA footprint and each via hole is soldered to the outer portion of the corresponding FPGA contact pad on the first-mentioned printed-circuit board. The via-hole/contact joints can be inspected visually and can be easily unsoldered later. The square hole in the middle of the board saver is sized to accommodate the FPGA, and the thickness of the board saver is the same as that of the FPGA. Hence, when a non-final FPGA is placed in the square hole, the combination of the non-final FPGA and the board saver occupy no more area and thickness than would a final FPGA soldered directly into its designated position on the first-mentioned circuit board. The contact leads of a non-final FPGA are not bent and are soldered, at the top of the board saver, to the corresponding via holes. A non-final FPGA can readily be unsoldered from the board saver and replaced by another one. Once the final FPGA design has been determined, the board saver can be unsoldered from the contact pads on the first-mentioned printed-circuit board and replaced by the final FPGA.

Berkun, Andrew↗

Constructing LDPC Codes from Loop-Free Encoding Modules

A method of constructing certain low-density parity-check (LDPC) codes by use of relatively simple loop-free coding modules has been developed. The subclasses of LDPC codes to which the method applies includes accumulate-repeat-accumulate (ARA) codes, accumulate-repeat-check-accumulate codes, and the codes described in Accumulate-Repeat-Accumulate-Accumulate Codes (NPO-41305), NASA Tech Briefs, Vol. 31, No. 9 (September 2007), page 90. All of the affected codes can be characterized as serial/parallel (hybrid) concatenations of such relatively simple modules as accumulators, repetition codes, differentiators, and punctured single-parity check codes. These are error-correcting codes suitable for use in a variety of wireless data-communication systems that include noisy channels. These codes can also be characterized as hybrid turbolike codes that have projected graph or protograph representations (for example see figure); these characteristics make it possible to design high-speed iterative decoders that utilize belief-propagation algorithms. The present method comprises two related submethods for constructing LDPC codes from simple loop-free modules with circulant permutations. The first submethod is an iterative encoding method based on the erasure-decoding algorithm. The computations required by this method are well organized because they involve a parity-check matrix having a block-circulant structure. The second submethod involves the use of block-circulant generator matrices. The encoders of this method are very similar to those of recursive convolutional codes. Some encoders according to this second submethod have been implemented in a small field-programmable gate array that operates at a speed of 100 megasymbols per second. By use of density evolution (a computational- simulation technique for analyzing performances of LDPC codes), it has been shown through some examples that as the block size goes to infinity, low iterative decoding thresholds close to channel capacity limits can be achieved for the codes of the type in question having low maximum variable node degrees. The decoding thresholds in these examples are lower than those of the best-known unstructured irregular LDPC codes constrained to have the same maximum node degrees. Furthermore, the present method enables the construction of codes of any desired rate with thresholds that stay uniformly close to their respective channel capacity thresholds.

Divsalar, Dariush↗

Automated Knowledge Discovery From Simulators

A computational method, SimLearn, has been devised to facilitate efficient knowledge discovery from simulators. Simulators are complex computer programs used in science and engineering to model diverse phenomena such as fluid flow, gravitational interactions, coupled mechanical systems, and nuclear, chemical, and biological processes. SimLearn uses active-learning techniques to efficiently address the "landscape characterization problem." In particular, SimLearn tries to determine which regions in "input space" lead to a given output from the simulator, where "input space" refers to an abstraction of all the variables going into the simulator, e.g., initial conditions, parameters, and interaction equations. Landscape characterization can be viewed as an attempt to invert the forward mapping of the simulator and recover the inputs that produce a particular output. Given that a single simulation run can take days or weeks to complete even on a large computing cluster, SimLearn attempts to reduce costs by reducing the number of simulations needed to effect discoveries. Unlike conventional data-mining methods that are applied to static predefined datasets, SimLearn involves an iterative process in which a most informative dataset is constructed dynamically by using the simulator as an oracle. On each iteration, the algorithm models the knowledge it has gained through previous simulation trials and then chooses which simulation trials to run next. Running these trials through the simulator produces new data in the form of input-output pairs. The overall process is embodied in an algorithm that combines support vector machines (SVMs) with active learning. SVMs use learning from examples (the examples are the input-output pairs generated by running the simulator) and a principle called maximum margin to derive predictors that generalize well to new inputs. In SimLearn, the SVM plays the role of modeling the knowledge that has been gained through previous simulation trials. Active learning is used to determine which new input points would be most informative if their output were known. The selected input points are run through the simulator to generate new information that can be used to refine the SVM. The process is then repeated. SimLearn carefully balances exploration (semi-randomly searching around the input space) versus exploitation (using the current state of knowledge to conduct a tightly focused search). During each iteration, SimLearn uses not one, but an ensemble of SVMs. Each SVM in the ensemble is characterized by different hyper-parameters that control various aspects of the learned predictor - for example, whether the predictor is constrained to be very smooth (nearby points in input space lead to similar output predictions) or whether the predictor is allowed to be "bumpy." The various SVMs will have different preferences about which input points they would like to run through the simulator next. SimLearn includes a formal mechanism for balancing the ensemble SVM preferences so that a single choice can be made for the next set of trials.

Burl, Michael↗

Kalman Filter for Calibrating a Telescope Focal Plane

The instrument-pointing frame (IPF) Kalman filter, and an algorithm that implements this filter, have been devised for calibrating the focal plane of a telescope. As used here, calibration signifies, more specifically, a combination of measurements and calculations directed toward ensuring accuracy in aiming the telescope and determining the locations of objects imaged in various arrays of photodetectors in instruments located on the focal plane. The IPF Kalman filter was originally intended for application to a spaceborne infrared astronomical telescope, but can also be applied to other spaceborne and ground-based telescopes. In the traditional approach to calibration of a telescope, (1) one team of experts concentrates on estimating parameters (e.g., pointing alignments and gyroscope drifts) that are classified as being of primarily an engineering nature, (2) another team of experts concentrates on estimating calibration parameters (e.g., plate scales and optical distortions) that are classified as being primarily of a scientific nature, and (3) the two teams repeatedly exchange data in an iterative process in which each team refines its estimates with the help of the data provided by the other team. This iterative process is inefficient and uneconomical because it is time-consuming and entails the maintenance of two survey teams and the development of computer programs specific to the requirements of each team. Moreover, theoretical analysis reveals that the engineering/ science iterative approach is not optimal in that it does not yield the best estimates of focal-plane parameters and, depending on the application, may not even enable convergence toward a set of estimates.

Kang, Bryan↗

Spacecraft Attitude Maneuver Planning Using Genetic Algorithms

A key enabling technology that leads to greater spacecraft autonomy is the capability to autonomously and optimally slew the spacecraft from and to different attitudes while operating under a number of celestial and dynamic constraints. The task of finding an attitude trajectory that meets all the constraints is a formidable one, in particular for orbiting or fly-by spacecraft where the constraints and initial and final conditions are of time-varying nature. This approach for attitude path planning makes full use of a priori constraint knowledge and is computationally tractable enough to be executed onboard a spacecraft. The approach is based on incorporating the constraints into a cost function and using a Genetic Algorithm to iteratively search for and optimize the solution. This results in a directed random search that explores a large part of the solution space while maintaining the knowledge of good solutions from iteration to iteration. A solution obtained this way may be used as is or as an initial solution to initialize additional deterministic optimization algorithms. A number of representative case examples for time-fixed and time-varying conditions yielded search times that are typically on the order of minutes, thus demonstrating the viability of this method. This approach is applicable to all deep space and planet Earth missions requiring greater spacecraft autonomy, and greatly facilitates navigation and science observation planning.

Kornfeld, Richard P.↗

Real-Time Adaptive Color Segmentation by Neural Networks

Artificial neural networks that would utilize the cascade error projection (CEP) algorithm have been proposed as means of autonomous, real-time, adaptive color segmentation of images that change with time. In the original intended application, such a neural network would be used to analyze digitized color video images of terrain on a remote planet as viewed from an uninhabited spacecraft approaching the planet. During descent toward the surface of the planet, information on the segmentation of the images into differently colored areas would be updated adaptively in real time to capture changes in contrast, brightness, and resolution, all in an effort to identify a safe and scientifically productive landing site and provide control feedback to steer the spacecraft toward that site. Potential terrestrial applications include monitoring images of crops to detect insect invasions and monitoring of buildings and other facilities to detect intruders. The CEP algorithm is reliable and is well suited to implementation in very-large-scale integrated (VLSI) circuitry. It was chosen over other neural-network learning algorithms because it is better suited to realtime learning: It provides a self-evolving neural-network structure, requires fewer iterations to converge and is more tolerant to low resolution (that is, fewer bits) in the quantization of neural-network synaptic weights. Consequently, a CEP neural network learns relatively quickly, and the circuitry needed to implement it is relatively simple. Like other neural networks, a CEP neural network includes an input layer, hidden units, and output units (see figure). As in other neural networks, a CEP network is presented with a succession of input training patterns, giving rise to a set of outputs that are compared with the desired outputs. Also as in other neural networks, the synaptic weights are updated iteratively in an effort to bring the outputs closer to target values. A distinctive feature of the CEP neural network and algorithm is that each update of synaptic weights takes place in conjunction with the addition of another hidden unit, which then remains in place as still other hidden units are added on subsequent iterations. For a given training pattern, the synaptic weight between (1) the inputs and the previously added hidden units and (2) the newly added hidden unit is updated by an amount proportional to the partial derivative of a quadratic error function with respect to the synaptic weight. The synaptic weight between the newly added hidden unit and each output unit is given by a more complex function that involves the errors between the outputs and their target values, the transfer functions (hyperbolic tangents) of the neural units, and the derivatives of the transfer functions.

Duong, Tuan A.↗