| Abstract |
A new control strategy for induction motors in variable-speed applications using a resonant link inverter is proposed. Standard melhods for generation of the inverter's gate control signals are replaced by an algorithm residing in a floating-point digital signal processor (DSP). The algorithm can be represented as a decision tree, testing measured quantities to produce the binary gate control signals for each phase. The algorithm is automatically generated and coded for the DSP by machine learning software residing in a second processor. An expert system in the second processor monitors the inverter's performance using, for example, measurements of the input current's harmonic content, and can modify theDSP algorithm during on-line operation, adapting inverter operation to changing conditions, and providing initial tuning of the inverter to a specific application.
|
