Timothy L. Ruchti

Identification algorithm for systemic arterial parameters with application to total artificial heart control

A new algorithm for estimating systemic arterial parameters from systolic pressure and flow measurements at the root of the aorta is developed and tested through a systems identification approach. The resulting procedure has direct application to a total artificial heart (TAH) control system currently under development. Identification models, representing the systemic arterial system, are developed from existing work in the area of cardiovascular modeling. The resistive and compliance components of these models are physically significant, representing overall hydraulic properties of the systemic arterial system. A unique method of parameterizing the identification models is designed which operates on the basis of aortic pressure and flow measurements taken exclusively during systole. The estimator is a modified recursive least squares algorithm which utilizes covariance modification to track time-varying parameters and a dead-zone to improve the robustness. Performance of the estimation algorithm was tested on data generated by a higher-order distributed model of the systemic arterial bed using normal canine parameters. Results from model-to-model experiments verify the consistency of the estimates and the ability of the estimator to converge quickly and track dynamically varying parameters.

Clinical results from a noninvasive blood glucose monitor

Non-invasive blood glucose monitoring has long been proposed as a means for advancing the management of diabetes through increased measurement and control. The use of a near-infrared, NIR, spectroscopy based methodology for noninvasive monitoring has been pursued by a number of groups. The accuracy of the NIR measurement technology is limited by challenges related to the instrumentation, the heterogeneity and time-variant nature of skin tissue, and the complexity of the calibration methodology. In this work, we discuss results from a clinical study that targeted the evaluation of individual calibrations for each subject based on a series of controlled calibration visits. While the customization of the calibrations to individuals was intended to reduce model complexity, the extensive requirements for each individual set of calibration data were difficult to achieve and required several days of measurement. Through the careful selection of a small subset of data from all samples collected on the 138 study participants in a previous study, we have developed a methodology for applying a single standard calibration to multiple persons. The standard calibrations have been applied to a plurality of individuals and shown to be persistent over periods greater than 24 weeks.

Modeling torque in a switched reluctance motor for adaptive control purposes using self-organizing neural networks

Training paradigms for topology-preserving Kohonen neural networks are introduced for the purpose of identifying and controlling nonlinear systems. A procedure for locking neuron weights at specific locations in a region is presented. It exploits prior knowledge about the system of interest. As a result, superior representations of an arbitrary multivariable nonlinear mapping can be achieved. In addition, the common problem of twisted meshes in these neural networks is eliminated. The strategy introduced for preferentially training these networks at region boundaries overcomes the limitation of boundary contraction. As an example, a one-dimensional neural network is used to approximate a nonlinear function, although in general an n-dimensional mapping can be used to approximate an m-dimensional system for n<or=m. As a practical implementation, the modeling of the theoretical torque of a switched reluctance motor (SRM) as a function of position and current is presented. The topological torque representation is suitable for adaptive control of SRMs in high-performance applications.<<ETX>>

Artificial neural network identification of partially known dynamic nonlinear systems

This paper presents a method for incorporating a priori information about an uncertain nonlinear system into the structure of a multilayer feedforward artificial neural network. Known information is incorporated into the activation function of the network output layer. An algorithm is derived for backpropagating the error and updating adjustable parameters within this layer that is consistent with existing supervised learning techniques. The developed technique is applied to the identification of a dynamic system and compared with conventional feedforward artificial neural network identifier. Results exhibit an improvement in the quality of the identification model and an increase in the rate of convergence. As a practical application, a prior information is utilized for identification of switched reluctance motor characteristics on the basis of experimental measurements. The results further demonstrate that artificial neural networks employing a priori information converge faster, require fewer adjustable weights, and more accurately predict the system of interest.<<ETX>>

Gray layer technology: incorporating a priori knowledge into feedforward artificial neural networks

Gray layer technology represents a novel method for incorporating prior information about an uncertain nonlinear system into the structure of a multilayer feedforward artificial neural network (ANN). A prescriptive technique is developed that sets or constrains the weights of a particular layer, designated the gray layer, according to a known or partially known function that is related in some manner to the system of interest. An algorithm is derived for backpropagating the error through the gray layer and adjusting the parameters of the gray layer that is consistent with other ANN learning techniques. Gray layer technology is applied to the identification of two different partially known dynamic systems. Results demonstrate a significant improvement in the quality of the identification model and an increase in the rate of convergence.<<ETX>>

Kalman based artificial neural network training algorithms for nonlinear system identification

The utility of artificial neural networks (ANNs) in nonlinear system identification and control is intimately linked with the ability to parameterize the ANN structure on the basis experimental observations. Four existing training algorithms are reviewed under a parameter estimation framework, and the method of target state backpropagation previously proposed by the authors is extended. The new algorithm follows a different approach to the generation of error signals in embedded layers by backpropagating target or desired states rather than partial derivatives. The target states are used in conjunction with a linear Kalman based update algorithm, and transients associated with initial conditions are eliminated through a time-varying method of covariance modification. Comparisons of the five algorithms are made through a system identification problem, and the error convergence associated with each algorithm versus actual training time is presented. The results demonstrate an increased rate of convergence in comparison with backpropagation.<<ETX>>

Nonlinear estimation of torque in switched reluctance motors using grid locking and preferential training techniques on self-organizing neural networks

The torque of a switched reluctance motor (SRM) can be estimated using a topology-preserving self-organizing neural network map. Since self-organizing maps tend to contract at region boundaries, a procedure for locking neuron weights at specific locations in a region is presented. A strategy for preferentially training neuron weights on the region boundaries is introduced. As an example of these training techniques, a one-dimensional neural network will approximate a nonlinear function. In general an n-dimension mapping can be used to approximate an m-dimensional system for n<or=m. As a practical implementation of this technique, the modeling of the theoretical torque of a SRM as a function of position and current is presented. A two-dimensional neural network estimates a three-dimensional highly nonlinear surface.<<ETX>>

A parameter estimation approach to artificial neural network weight selection for nonlinear system identification

A unified framework for artificial neural network (ANN) training algorithms applied to nonlinear system identification based on considering weight selection as a parameter estimation problem is presented. Three existing ANN training strategies are reviewed within this framework, including gradient-descent backpropagation, the extended Kalman algorithm, and the recursive-least-squares method. A strikingly different approach to error backpropagation is presented, resulting in the development of a novel method of backward signal propagation and target state generation for embedded layers. The technique is suitable for implementation with a linear Kalman-based update algorithm and is applied with a unique method of covariance modification for the elimination of transients associated with initial conditions. Experimental nonlinear identification results demonstrate a greatly increased rate of convergence in comparison with backpropagation. The new algorithm displayed similar rates of parameter convergence and a decreased computational overhead compared with the extended Kalman algorithm.<<ETX>>

Estimation of artificial neural network parameters for nonlinear system identification

A unified framework for representing ANN (artificial neural network) training algorithms is developed by considering weight selection as a parameter estimation problem. Three existing ANN training strategies are reviewed within this framework, i.e., gradient-descent backpropagation, the extended Kalman algorithm, and the recursive least squares method. A strikingly different approach to error backpropagation is presented, resulting in the development of a novel method of backward signal propagation and target state generation for embedded layers. The proposed technique is suitable for implementation with a linear-Kalman based update algorithm and is applied with a time-varying method of covariance modification for the elimination of transients associated with initial conditions. Results from a nonlinear identification experiment demonstrate an increased rate of convergence in comparison with backpropagation. The new algorithm displayed similar rates of parameter convergence and a decreased computational overhead compared to the extended Kalman algorithm.<<ETX>>

On identification of partially known dynamic nonlinear systems with neural network

A method for incorporating a priori information about an uncertain nonlinear system into the structure of a multilayer feedforward artificial neural network is presented. The result is an improved identification model and controller structures suitable for nonlinear system identification and control applications. The known information is incorporated into the activation function of the network output layer. An algorithm is derived for backpropagating the error and updating adjustable parameters within this layer that is consistent with existing supervised learning techniques. The developed technique is applied to the