A Kalman fjlter algorithm to estimate Direction for automotive navigation is reported. An extended Kalman filter is used to combine a magnetic compass and a rate gyroscope and to compensate f o r the sensor errors; the result is an optimal estimate for heading direction. A mathematical model f o r magnetic compass errors caused by body magnetization and the body effect of magnetic material is proposed. An error model of the rate gyroscope is also established. Errors of both sensors are calibrated through a computer simulation. Finally, experimental navigation results are demonstrated.

No abstract

This paper presents an efficient pulse-width modulation (PWM) method to optimally reduce switching (SW) losses in three-phase neutral-point clamped (NPC) inverters in one switching cycle using a new two-phase modulation scheme. The proposed method digitally selects an optimum switching vector (SV) pattern, which minimizes an estimation function based on switching losses considering phase currents among possible reduction patterns for one switching cycle. The proposed modulation method is further extended to reduce common-mode voltages in three-phase NPC inverters. The basic principle involves elimination of the common-mode voltage by limiting the combinations of the output phase voltage levels of the three-level NPC inverter. A modified space vector generation technique based on the two-phase modulation method is proposed to reduce both common-mode voltages and switching losses. The proposed PWM method was experimentally verified for output voltages of the NPC inverter connected to a universal load. The method reduced the number of switching transitions for the largest or second largest current, and it can automatically adapt to different load conditions. The method was also experimentally validated for its EMC and switching loss reduction capability.