Novel technology independent neural network approach on device modelling interface

A novel, fast and accurate neural network tool is proposed for efficient technology independent realisation of the interface between device modelling and circuit simulation. Enhanced back-propagation neural network based algorithms are applied to the problem of modelling various device characteristics. These algorithms include the modified back-propagation algorithm, the conjugate gradient algorithm and the Levenberg-Marquardt algorithm. Also, the radial basis function neural network is tested in the device modelling problem. Simulations show fast convergence or learning rate and an excellent fit of recalled characteristics to the measured device data. The algorithm utilised is robust and capable of presenting the entire device characteristics unaltered even with largely reduced amount of the teaching material. The good monotonicity of the neural network generated device data facilitates the usage of the method in circuit simulation purposes. Possible further applications of implementing circuit level macromodels with this technique are discussed