Farid U. Dowla

OPTIMIZATION OF GROUNDWATER REMEDIATION USING ARTIFICIAL NEURAL NETWORKS WITH PARALLEL SOLUTE TRANSPORT MODELING

A new approach to nonlinear groundwater management methodology is presented which optimizes aquifer remediation with the aid of artificial neural networks (ANNs). The methodology allows solute transport simulations, usually the main computational component of management models, to be run in parallel. The ANN technology, inspired by neurobiological theories of massive interconnection and parallelism, has been successfully applied to a variety of optimization problems. In this new approach, optimal management solutions are found by (1) first training an ANN to predict the outcome of the flow and transport code, and (2) then using the trained ANN to search through many pumping realizations to find an optimal one for successful remediation. The behavior of complex groundwater scenarios with spatially variable transport parameters and multiple contaminant plumes is simulated with a two-dimensional hybrid finite-difference/finite-element flow and transport code. The flow and transport code develops the set of examples upon which the network is trained. The input of the ANN characterizes the different realizations of pumping, with each input indicating the pumping level of a well. The output is capable of characterizing the objectives and constraints of the optimization, such as attainment of regulatory goals, value of cost functions and cleanup time, and mass of contaminant removal. The supervised learning algorithm of back propagation was used to train the network. The conjugate gradient method and weight elimination procedures are used to speed convergence and improve performance, respectively. Once trained, the ANN begins a search through various realizations of pumping patterns to determine whether or not they will be successful. The search is directed by a simple genetic algorithm. The resulting management solutions are consistent with those resulting from a more conventional optimization technique, which combines solute transport modeling and nonlinear programming with a quasi-Newton search. The results suggest that the ANN approach has the following advantages over the conventional technique for the test remediations: more independence of the flow and transport code from the optimization, greater influence of hydrogeologic insight, and less computational burden due to the potential for parallel processing of the flow and transport simulations and the ability to “recycle” these simulations. The ANN performance was observed upon variation of the problem formulation, network architecture, and learning algorithm.

Optimal field-scale groundwater remediation using neural networks and the genetic algorithm.

We present a new approach for field-scale nonlinear management of groundwater remediation. First, an artificial neural network (ANN) is trained to predict the outcome of a groundwater transport simulation. Then a genetic algorithm (GA) searches through possible pumping realizations, evaluating the fitness of each with a prediction from the trained ANN. Traditional approaches rely on optimization algorithms requiring sequential calls of the groundwater transport simulation. Our approach processes the transport simulations in parallel and ``recycles`` the knowledge base of these simulations, greatly reducing the computational and real-time burden, often the primary impediment to developing field-scale management models. We present results from a Superfund site suggesting that such management techniques can reduce cleanup costs by over a hundred million dollars.

Position estimation of transceivers in communication networks

With rapid developments in wireless sensor networks, there is a growing need for transceiver position estimation independent of GPS, which may not be available in indoor networks. Our approach is to use range estimates from time-of-flight (TOF) measurements, a technique well suited to large bandwidth physical links, such as in ultra-wideband (UWB) systems. In our UWB systems, pulse duration less than 200 psecs can easily be resolved to less than a foot. Assuming an encoded UWB physical layer, we first test positioning accuracy using simulations. We are interested in sensitivity to range errors and the required number of ranging nodes, and we show that in a high-precision environment, such as UWB, the optimal number of transmitters is four. Four transmitters with /spl plusmn/20 ft. range error can locate a receiver to within one or two feet. We then implement these algorithms on an 802.11 wireless network and demonstrate the ability to locate a network access point to approximately 20 feet.

Interference mitigation in transmitted-reference ultra-wideband (UWB) receivers

Transmitted-reference (TR) ultra-wideband transceivers have recently become increasingly popular for their simplicity, capability to reduce the stringent UWB timing requirements, and robust performance in multipath channels. However, the performance of TR receivers is considerably limited by the severity of noise-on-noise component introduced by various types of channel noise such as additive white Gaussian noise (AWGN) or narrowband interference (NBI) on the transmitted signal. It is expected that such receivers will perform poorly at low signal-to-noise ratio links, or in the presence of strong narrowband interferers. In this paper we propose a novel technique that maximizes the extraction of information from reference pulses for UWB-TR receivers. The scheme efficiently processes the incoming signal to suppress different types of interference prior to signal detection. The method described introduces a feedback loop mechanism to enhance the signal-to-noise ratio of reference pulses in a conventional TR receiver. The performance of a conventional TR receiver and a feedback loop TR receiver in the presence of AWGN and strong narrowband interference is investigated by analysis and computer simulations. Our studies show that the reference enhancing feedback loop mechanism greatly improves the robustness of the link performance of TR receivers in the presence of non-UWB interference with modest increase in complexity.

Vector quantization of ECG wavelet coefficients

An improved wavelet compression algorithm for ECG signals has been developed with the use of vector quantization on wavelet coefficients. Vector quantization on scales of long duration and low dynamic range retains feature integrity of the ECG with a very low bit-per-sample rate. Preliminary results indicate that the proposed method excels over standard techniques for high fidelity compression.<<ETX>>

Self organization of wireless sensor networks using ultra-wideband radios

Ultra-wideband (UWB) technology has proven to be useful in short range, high data rate, robust, and low power communications. These features can make UWB systems ideal candidates for reliable data communications between nodes of a wireless sensor network (WSN). However, the low powered UWB pulses can be significantly degraded by channel noise, inter-node interference, and intentional jamming. In This work we present a novel interference suppression technique for UWB based WSN that promises self-organization in terms of power conservation, scalability, and channel estimation for the entire distributed network.

Neural Networks in Seismic Discrimination

Neural networks are powerful and elegant computational tools that can be used in the analysis of geophysical signals. At Lawrence Livermore National Laboratory, we have developed neural networks to solve problems in seismic discrimination, event classification, and seismic and hydrodynamic yield estimation. Other researchers have used neural networks for seismic phase identification. We are currently developing neural networks to estimate depths of seismic events using regional seismograms. In this paper different types of network architecture and representation techniques are discussed. We address the important problem of designing neural networks with good generalization capabilities. Examples of neural networks for treaty verification applications are also described.

Wideband multichannel time-reversal processing for acoustic communications in highly reverberant environments

The development of multichannel time-reversal (T/R) processing techniques continues to progress rapidly especially when the need to communicate in a reverberant environment is critical. The underlying T/R concept is based on time-reversing the Green’s function characterizing the uncertain communications channel mitigating the deleterious dispersion and multipath effects. In this paper, attention is focused on two major objectives: (1) wideband communications leading to a time-reference modulation technique; and (2) multichannel acoustic communications in two waveguides: a stairwell and building corridors with many obstructions, multipath returns, severe background noise, disturbances, and long propagation paths (∼180ft) including disruptions (bends). It is shown that T/R receivers are easily extended to wideband designs. Acoustic information signals are transmitted with an eight-element array to two receivers with a significant loss in signal levels due to the propagation environment. The results of the n...

A Dynamic Programming Algorithm for Name Matching

In many database and data mining applications concerning people, name matching plays a key role. Many algorithms to match names have been proposed. These algorithms must take into account spelling and transcription errors, name abbreviations, nicknames, out of order names, and missing or extra names. The existing algorithms typically fall along the lines of sound based, edit distance based, or token based algorithms which can use other methods in matching each part of the name separately. In this article, we propose a dynamic programming approach that includes a substring matching algorithm. The algorithms performance is compared against two often used algorithms by testing on a random sample of names from a database. The data used for the testing comes from the DHS US-VISIT Arrival and Departure Information System database, which includes names from all over the world. The performance on this data set was compared with that of the Damerau-Levenshtein algorithm and the Jaro-Winkler algorithm. The dynamic programming algorithm with substring matching performed better than both of these algorithms on the data tested

Application of neural networks for seismic phase identification

The effectiveness of a multilayered feedforward neural network for seismic phase identification was investigated. The database consisted of seismograms from 75 earthquakes and 75 underground nuclear explosions. For learning, the conjugate gradient error backpropagation algorithm with a weight-elimination method was used. Results indicate that feedforward neural networks appear to outperform a conventional Bayesian classifier in a problem where the task was restricted to identifying only two of the principal regional phases, Pg and Lg, on earthquake and explosion seismograms of the western United States.<<ETX>>