Authors: O. Mikulchenko, A. Rasmussen and K. Mayaram
Affilation: Washington State University, United States
Pages: 540 - 543
Keywords: microflow sensors, macromodeling, neural networks, microfluidic simulation
A neural network based macromodel has been developed for a microflow sensor and implemented in the circuit simulator SPICE3f5. The model has been validated with numerical simulations and allows accurate and efficient simulation of microflow sensors in a microfluidic system. This model simulates both the steady-state and the dynamic operation of the flow sensor. The neural network design is based on a problem-based scaling and a combination of stochastic search and the genetic algorithm. The macromodel can be used for microfluidic system simulation and the optimal design of the flow sensor.