Current reconstruction techniques for survivable three phase PWM converters

In high performance three phase pulse width modulated (PWM) converter systems, failure of current sensors will interrupt the system performance and even cause disastrous consequences without in-time maintenance. Redundancy design is one of the popular choices to overcome the issue. However, this usually leads to an undesirable increase in size and cost. This paper will present two novel model-independent genetic current reconstruction technique with digital implementation for three phase PWM converter systems. Using proposed sensor placement and software algorithm, phase currents within a three phase converter can be reconstructed to an acceptable level of accuracy under single-survived-sensor (S³) scenario. A hardware prototype is built and fault tolerant algorithms are implemented using a digital signal processor (TMS320F2812). Experiments are conducted under worst case scenario to verify the flexibility of sensor placement/converter layout and software algorithm.