<p>While the impact of a single electromagnetic radiation on neural dynamics is well established, the effects of multiple electromagnetic stimuli on the chaotic behavior of neural systems remain largely unexplored. In this paper, the impact of multiple electromagnetic radiation stimuli on the chaotic dynamics of a local neuronal system is investigated. To this end, a tri-cell Cellular neural network (TC-CNN) model under three electromagnetic radiations is proposed firstly, in which the first neuron receives two distinct electromagnetic radiations and the second neuron receives one. The electromagnetic effect is modeled as magnetic flux across the cell membrane, influencing membrane potential through memristive feedback. Under varying stimuli, the proposed TC-CNN exhibits abundant polymorphous chaotic dynamics, including butterfly attractors, single-directional and grid multi-butterfly attractors, as well as single-directional and planar initial-boosted coexisting butterfly attractors. In particular, these behaviors can be flexibly controlled by tuning model parameters and initial values. Specifically, the number of multi-butterfly attractors can be altered by adjusting the memristor’s control parameters, whereas the location of the initial-boosted behaviors can be regulated by modifying its initial values. Meanwhile, an analog circuit implementation of the TC-CNN is presented, and experimental results align with numerical simulations. Furthermore, a privacy protection scheme for bank facial recognition services is developed based on the TC-CNN, demonstrating strong security performance and practical applicability.</p>