To see the performance of the proposed adaptive comfort models under variations of the outdoor temperature, a real measurement of the outdoor temperature outside an office building is obtained during a working time of 8:00-18:00 hrs. Fig. 10 shows the dotted plots of the actual outdoor temperature that are sampled every 5 minutes. The weather for that whole day is mostly cloudy and warm.
   It can be seen in Fig. 10 that the outdoor temperature rises from 29.7°C at time 8:00 hrs to a peak of 32.4°C at time of 12:30 hrs and drops to 29.5°C at 14:00 hrs, due to raining. After the rain stops, the outdoor temperature increases to 32.2°C at 15:30 hrs and it decreases afterward. The solid line presents quite good predictions of the outdoor temperature determined from the grey forecasting model in Eq. (10), although the initial guess of the outdoor temperature is set to be 30°C. After a while, the estimated outdoor temperature can track along the actual outdoor temperature. Fig. 11 illustrates plots of the indoor temperature and outdoor temperature against the time when the simulated air-conditioning unit is switched on at 8:00 hrs. The PI controller has the capability to keep the temperature of the indoor air at the reference temperature of 25°C, even though the outdoor temperature changes and the thermal loads act on the air-conditioned space.
   At each sampling interval of real-time implementation, the indoor comfort temperature is determined from sampled signals of the outdoor temperature while the sampling time is designed for effective controller performance of air-conditioning systems. In Fig. 12, the indoor comfort temperature is determined to be the reference temperature in each sampling time according to dynamic changes of the outdoor temperature as given in Fig. 10.
   Under the same thermal conditions, Fig. 13 presents that the temperature of the indoor air is controlled to follow the reference temperature, which is recommended by the model of the adaptive indoor comfort temperature. According to the obtained results, the simulation studies indicate the strong viability of the proposed methodology for the real-time implementation of the embedded system as described in the previous sections.
   To show the effectiveness of the proposed technique in a practical use, a field study is conducted in an air-conditioned space for a reading room of the school library. The embedded devices as mentioned in section 2 are used daily for real-time control implementation of adaptive indoor comfort temperature with a typical air-conditioning system. Two hundred and twenty-five questionnaires from potential students and staffs were collected for two months.
   Table 3 indicates personal data of interviewees, who filled out the questionnaire forms during the survey on the satisfaction of the thermal comfort.
   In overall characteristics of library users, most interviewees are male students of 11-15 years that have a previous activity of indoor walking, and then spend time in the reading room of the school library later. Fig. 14 shows a graphical display of tabulated frequencies of the outdoor air temperature, the corresponding indoor comfort temperature and the indoor air temperature during the survey. It can be observed that the temperature of the outdoor air around the library building has a high frequency at 30.2°C and 31.4°C. The minimum temperature and maximum temperature are 28.2°C and 33.6°C, respectively. The distribution of the indoor comfort temperature is similar to that of outdoor air temperature due to the linear relation in the adaptive comfort model as expected. By observation, although there were extensive variations in the outdoor air temperature, the embedded system is capable of maintaining the indoor air temperature close to the indoor comfort temperature with an accuracy of approximately ± 1°C, as seen in Fig. 14.
   As mentioned in section 3, the reference temperature is set at a given indoor comfort temperature. The task of the temperature controller is to maintain the actual indoor temperature at the desired indoor comfort temperature. The variation of the indoor temperature observed in Fig. 14 takes place due to external disturbances, such as changes in numbers of occupants, air infiltration or air exfiltration, usages of electrical appliances, etc. Table 4 reports the percentage results of thermal comfort, satisfaction of interviewees and expectation on improvement of thermal comfort. It can be noticed that few interviewees are dissatisfied, with more than 95% of the interviewees having responses close to neutrality (slightly cool, neutral, and slightly warm). The investigation reflects that the proposed implementation can be well satisfactory to most subjects in thermal comfort. 嵌入式系统对空调舒适温度控制的英文文献和翻译(4):http://www.751com.cn/fanyi/lunwen_3023.html