<p>This paper proposes a new method to regulate the voltage generated by a vehicle alternator using a fuzzy neural network (FNN) regulator. Variations of alternator loads and rotation speeds in a vehicle and different characteristics of alternators influence the regulated voltages of an alternator. To address this problem, this paper employs an FNN-based automatic voltage regulation (AVR) method to reduce regulated voltage fluctuations. The FNN regulator has three inputs: alternator speed expressed as revolutions per minute, load voltage, and ripple amplitude from an alternation rectifier bridge. Based on the three inputs, the FNN determines the duty cycle of pulse width modulation that switches on and off the alternator field to maintain a constant regulated voltage. The FNN comprises Takagi-Sugeno-type fuzzy rules built through structure and parameter learning. After learning, the FNN is implemented in a microcontroller unit. An alternator testing equipment is used to verify actual voltage regulation performances of the proposed FNN-based AVR. Experimental results show that the FNN helps determine optimal duty cycles to regulate the regulator voltages of different alternators under varying operating conditions, ensuring robust charging voltage. Comparisons with different AVR methods show the advantage of the proposed FNN-based AVR method.</p>