Variations in dynamic properties are commonly used in Structural Health Monitoring to assess the conditions of a structural system, being these parameters sensitive to damage-induced changes. Yet, such variations can also be due to changes in environmental parameters, like fluctuations in temperature, humidity, etc. By performing a continuous monitoring, the correlation between those factors appears and their variations, if no damage exists, result in a cyclic phenomenon. Negative selection, a bio-inspired classification algorithm, can be exploited to distinguish anomalous from normal changes, thus eliminating the influence of environmental effects on the assessment of the structural condition. This algorithm can be trained to relate specific extracted features (e.g. modal frequencies) and other monitored parameters (e.g. environmental conditions), allowing to identify damage when the registered value oversteps the confidence interval defined around the predicted value. Negative selection draws inspiration from the mammalian immune system, whose physiology demonstrates the efficiency of this process in discriminating non-self elements, despite the restricted number of receptors available to face a vast amount of aggressors. In this paper, a negative-selection algorithm based on a non-random strategy for detector generation is optimized and tested on the monitoring data of a prominent monument of the Portuguese architecture.