Smoke separation is accelerating towards the Tokyo Olympics and Paralympics. However, there is no method for monitoring harmful chemical substances in tobacco smoke in real-time, so there are scientific lacks against non-smokers health whether the smoke separated space have no impact or not to their health. The main chemical substance in tobacco smoke is nicotine. We decided to build up a model first for nicotine. In this study, we use a newly developing gas analyzer which measures nicotine in real-time. From that nicotine measurement result, analyzes the nicotine behavior with FEM in a simulated space. Then make it learn the parameters of the nicotine behavior model, and using a different modeling and learning algorithm than general multi-agent reinforcement learning, we visualize the nicotine behavior in tobacco smoke through the new achieving model. First, we confirmed the availability of nicotine analysis with the gas analyzer.
