Nowadays, cyber-physical systems (CPS) are becoming ubiquitous in various application domains. The variety of design and implementation methodologies utilized for cyber-physical systems, as well as the dynamic interaction of its components, make the resilience of these systems a major challenge. We aim to increase the resilience of these systems in a decentralized way by leveraging knowledge of the social sciences and humanities (SSH) and especially emotional processes. Both individual decision-making processes and social coordinating mechanisms are based on emotional inspiration. Our hypotheses and studies on resilience approaches, cyber-physical systems and emotional processes allowed us to choose the multi-agent paradigm. In this paper, we present the results of our research on resilience, which includes an emotion-inspired anomaly detection approach for improving CPS resilience. This approach is integrated into an agent architecture, compared to the literature, and validated through the development of proof-of-concept scenarios. The experimental results prove its advantages in terms of resilience properties.