Evidence‐Gated Hierarchical Control of Fuel Cell Hybrid Electric Vehicles Under Degraded Perception: A Neutrosophic Formulation Tested Against a Matched Fuzzy Control

Health‐aware energy management for fuel cell hybrid electric vehicles assumes that the signals it acts upon are trustworthy. In an automated vehicle the power demand comes from a radar‐fed adaptive cruise control layer, and radar is intermittently corrupted by occlusion, so a conventional energy manager ramps the stack in response to sensing artifacts and thereby incurs additional load‐cycling degradation. This paper proposes an evidence‐gated hierarchical architecture that freezes the estimate of the sustained load, and hence the stack reference, when an independent perception‐quality channel reports untrustworthy evidence, and it is expressed in single‐valued neutrosophic terms. A matched conventional‐fuzzy realization of the same gate performs within about 1% of the neutrosophic one, so the measured benefit is attributable to the gated architecture rather than to the neutrosophic grades. Against seven baselines minimizing the identical objective, each with a filter tuned to its own optimum, the gated architecture reduces stack voltage decay by 23.7%–35.8% relative to the fuzzy and ECMS families at a 10% radar‐occlusion duty cycle, at parity of operating cost; the reduction persists under alternative load‐cycling damage models, from an aged initial state and on three unseen drive cycles. Model‐predictive control attains lower decay at about 12% higher cost and 3.5 times the computation, and the jerk reduction comes from the gated cruise‐control layer. Under clean perception the controller offers no cost advantage, and the regimes in which its advantage holds are mapped.