Study of odor recorder

我们的目标是建立嗅觉信息记录和再现的基本阶段，利用气味传感系统。如果建立记录和再现气味的技术，并将其应用到多媒体中，那么与五种感官的交流可能会变得更加可行。对苹果和柑橘的香味进行了记录和再现实验。感官测试表明，我们开发的系统所产生的气味与目标苹果的味道相同。然后，在气味记录仪中实现了每隔几秒交替切换目标气味和混合气味的实时参考方法。它有助于增强对温度和湿度变化等环境变化的鲁棒性。此外，利用实时参考方法，即使在环境变化的情况下，也可以记录气味配方的动态变化。当我们同时使用两个相同的传感器阵列时，可以捕获更多的时间。然后，引入气味近似的概念来减少成分气味的数量。我们还使用自动进样器制作了混合多达96种成分气味的系统。我们尝试分别使用3种成分气味来近似由10-20种成分气味组成的苹果、香蕉、橙子和甜瓜口味。人体感官测试表明，主成分分析散射图上位置相近的气味接近。此外，还对不同底调配方的香水进行了测量。由于传感器对自动混合气味的响应与人工混合气味的响应接近，因此使用自动进样器可以获得足够的精度。我们将从现在开始进行大规模的实验，期望这两种方法，如气味近似和自动采样器，将大大扩大气味记录仪的气味范围。最后，实现了由喷墨装置和网状加热器组成的气味混合器，解决了低挥发性气味移动速度慢的问题。使用所提出的气味混合器，气味迅速产生并消失。利用该混合器和梯度下降法，得到了茉莉香精的二元配方。少

We aimed to establish the fundamental stage of recording and reproducing olfactory information using an odor sensing system. The communication with five senses might become more feasible if the technology for recording and reproducing smells is established and is implemented into a multimedia.The experiment on recording and reproducing the smells of the apple and citrus flavors was performed. The sensory test revealed that the smell reproduced by our developed system was identical to the target apple flavor. Then, real-time reference method, which alternately switches the target and blended odors at the interval of several seconds, was implemented into the odor recorder. It contributed to the enhancement of the robustness against the environmental change such as temperature and humidity changes. Moreover, it was found that the dynamical change of the odor recipe could be recorded even under the environmental change using the real-time reference method. The odor recipe change over 5 sec … More onds could be captured when we simultaneously used the two same sensor arrays.Then, the concept of the odor approximation was introduced to reduce the number of the component odors. We also made the system for blending up to 96 component odors using an auto sampler. We tried to approximate apple, banana, orange and melon flavors made up of 10-20 component odors using 3 component odors, respectively. The human sensory test revealed that the odors with similar locations on the scattering diagram of Principal Component Analysis were close. In addition, the perfumes with various recipes of base notes were measured. Since the sensor responses to the automatically blended odor was close to those to the manually blended one, it was found that the sufficient accuracy was obtained using the auto sampler. We will do the large-scale experiment from now on and expect that these two methods such as odor approximation and the auto sampler will drastically expand the range of odor for the odor recorder.Finally, the odor blender composed of inkjet devices and the mesh heater was realized to solve the problem of the slow speed for the movement of low-volatile odor. The odor was rapidly generated and vanished using the proposed odor blender. The binary recipe of jasmine flavor was obtained using this odor blender and the gradient descent method. Less