Internet of Things-driven intelligent ventilation system for buildings in humanized designs

The concept of the Internet of Things was first proposed by scholars at MIT in 1999, that is, the Internet of Things is a network that uses radio frequency identification technology and sensing devices to connect all items to the Internet for intelligent identification and management. In 2005, the ITU proposed in its annual report that “by embedding short-range mobile transceivers into a variety of accessories and everyday items, a new mode of communication between people, people and things, and things and things has been formed, i.e., real-time interactions can be realized at any time and any place.” In 2008, IBM formally put forward the idea and concept of Smart Earth, namely “Internet of Things + Internet = Smart Earth”. In this regard, the Internet of Things has further become an important innovative weapon needed by countries around the world to face the financial crisis, leading to a third revolution in the field of information technology. In 2009, the European Union's Strategic Roadmap for IoT Research also stipulated the concept of IoT. In the same year, Chinese Premier Wen Jiabao also put forward the concept of “Sensing China”, a concept related to the Internet of Things. Nowadays, the development of Internet of Things (IoT) has entered China's emerging strategic industry planning map, becoming an important strategy for the development of national information industry.

The conceptual model of the Internet of Things (IoT) is shown in Figure 1, which cannot be separated from the traditional Internet and has the following characteristics: first, it is a kind of ubiquitous network based on the Internet. The Internet of Things is the integration and expansion of the Internet with various networks, while the Internet remains the important foundation and core of the Internet of Things technology. Meanwhile, in the process of transmitting massive information, IoT supports various protocols and heterogeneous networks in order to meet various requirements for data. Secondly, it is an extended application of various sensing technologies. A huge number and variety of sensors are deployed on the IoT, where each sensor is a source of information collection, which can collect various specific information according to a specified frequency and periodicity, and these data are extremely real-time. What's more, IoT realizes intelligent processing on the basis of sensor connectivity itself. The Internet of Things realizes intelligent data processing on sensors and utilizes various intelligent technologies, giving it the ability to control objects and the environment intelligently, while expanding both width and depth in the field of application.

Figure 1.

Conceptual model of iot

The range of data that can be captured by a single sensor is very limited, and multiple and multiple types of sensors must be used to obtain relatively accurate as well as wide coverage data. Multi-sensor data fusion, sometimes referred to as multi-sensor information fusion or multi-sensor data fusion, is used to integrate and fuse multiple data from different sensors, devices, or algorithms to obtain more comprehensive and reliable information.

Grubbshuaize is particularly applicable in measurement scenarios where the amount of data is between 3 and 50, especially when individual outliers need to be identified and ruled out. This method is not only effective, but also simple and easy to implement. It is important to note that when the significance level is set at 0.1, the number of measurements performed must exceed 10 to ensure the accuracy of the criterion.

The environmental data obtained from measurements in one cycle were sorted in order from smallest to largest. Equations (1), (2) and (3) were utilized to calculate the arithmetic mean and standard deviation of these measurements: 1xi¯=1m∑m=1mxim 2εim=xim−xi¯ 3σi=∑m=1mεim2

From the Grubbs criterion, the Grubbs of the home's indoor internal environmental data is shown in equation (4): 4gim=xim−xi¯σi

Let y be the value of the significant level parameter, which can usually be taken as 0.05 or 0.01. When g_ij ≥ g_i0(m, y) holds, it means that the data contains gross errors and should be removed. In addition, the Grubbs Criterion g_i0(m, y) can be determined by looking up the table.

The Grubbs Criterion input measurements are also measured home indoor environmental data over a period of time, and then this sequence is ordered to find its semi-mean and variance. Determine the exact distribution of the Grubbs statistic and compare the magnitude of g_ij and g_i0(m, y). If yes, the data contains gross errors and should be eliminated, then calculate the next data and compare again. If no, the data is normal and should be retained, then calculate the next data.

The environmental data in the home interior collected by the environmental collection node within the home interior for a period of time, after preprocessing by the previous Grubbs criterion, the environmental data in the home interior collected by the terminal i of a certain collection node during this period of time is regarded as an array Q and divided into 3 groups. Each group can be represented as a sequence, and each sequence has its own corresponding mean and variance, as shown in equations (5) and (6). It is assumed here that there are 12 preprocessed data, so the 3 groups are 4 data in each group, and the first group can be represented as X_i = {X₁₁, X₁₂, X₁₃, X₁₄}, the second group as X₂ = {X₂₁, X₂₂, X₂₃, X₂₄}, and the third group as X₃ = {X₃₁, X₃₂, X₃₃, X₃₄}, where the m th group is represented as X_m = {X_m1, X_m2, ….., X_mn}: 5X¯m=1n∑nXmn 6σ2m=1n−1∑n(Xmn−Xm¯)2

The mean and variance of each of the three groups of data divided by Eq. (5) and Eq. (6) are obtained, and the following intra-group fusion is carried out, and through this computational process of Eq. (7), the mean and variance are utilized to compute the new fused value, i.e., the optimal ambient value of the environmental data acquired by the collection node terminals within the interior of the home within the period of time: 7X=σ21X1¯+σ22X2¯+σ23X3¯σ21+σ22+σ23

The previous equation obtains the optimal environmental values of the collection node terminals inside the home interior during a certain period of time, in addition, the total variance of the environmental data collected by the collection node terminals inside the home after fusion can be obtained based on the variance of each group σ², and the calculation process is shown in Equation (8): 8σ2=σ12σ22σ32σ12+σ22+σ32

Adaptive weighted fusion algorithms are commonly used for multi-sensor data fusion, where the data collected from different sensors are weighted according to their reliability and accuracy, and then the weighted data are fused to obtain the final result. In order to get the values that can more accurately reflect the environmental information inside the home, add weight values to each sequence of environmental data from each acquisition node, multiply each sequence with its weight, and finally add them to get the final environmental values. For example, now there are two collection nodes, it should be the environmental data collected by the two collection nodes by intra-group fusion and then weighted fusion.

Next, the weighted fused values of the environmental data inside the home are calculated, now assuming that n acquisition node is arranged inside the home, then there are the same number and type of sensors. The optimal fusion value and the overall total variance of each collection node sensor inside the home interior can be calculated using equation (7) and equation (8) above. Here, the weights of n identical sensors are set to be ω₁, ω₂, …… ω_n respectively, that is, weights are assigned to each acquisition node sensor. Assuming that the overall total mean square deviation of the data after fusion of each acquisition node sensor inside the home interior is σ² and the measurement result is X, the calculation formula can be obtained as shown in equation (9) below: 9{ X=∑i=1nωiXi∑i=1nωi=1σ2=∑i=1nωi2σi2

The above equation (9) is three separate equations, which are combined below to obtain the functional representation. The formula for the functional representation obtained from equation (9) is shown in equation (10) below: 10f(ω1,ω2,……ωn)=σ2=∑i=1nωi2σi2

In order to obtain the optimal weights, the key is to solve the minimal value of equation (10). The specific formula is shown in equation (11) below: 11ωi=1σi2∑i=1n1σi2

From Eq. (11), the weight value of the environmental data collected by a particular sensor is ω_i. Substituting this weight value into Eq. (9) yields the optimal value of the fused environmental data within the home interior, which is calculated as shown in Eq. (12): 12X=∑i=1nωiXi

Under the premise of ensuring the safety and comfort of the residential environment, in order to make the ventilation operating costs as small as possible, it is necessary to use appropriate control strategies to control the fan on. The ventilation system is a nonlinear, large time lag system, ordinary control strategies are difficult to meet the requirements. The fuzzy control method is a good choice for ventilation control because of its small dependence on the mathematical model of the system and good robustness.

Fuzzy control systems for a variety of control objects were also successfully developed. In the next few decades, fuzzy control theory has been developed tremendously, and it has become the most active control method in the field of intelligent control. Its characteristics are as follows: 1)

Although the fuzzy engineering calculation method utilizes the fuzzy set theory for fuzzy calculation, the results obtained are deterministic conditional statements.

Fuzzy, rule, inference, and de-fuzzy together constitute a fuzzy controller, which combines the actual situation of home indoor environment monitoring system, and automatically adjusts the fan turn-on time according to the collected environmental data using fuzzy inference. Inputting CO2 concentration deviation and dust concentration deviation into the fuzzy controller starts fuzzification, enters the inference machine, and then defuzzifies to get the clear value which is the fan on time. After performing the whole process go back to start the cycle next time.

After the introduction of the previous section, to complete the construction of the entire intelligent Internet of Things home ventilation system and the laboratory of the function of each part of the relevant test, the test results show that the realization of the function of each part of the expected goal. But a successful system in addition to ensure that the functions to be realized, a very important point is the stability of the system indicators meet the requirements, when the stability of the system indicators meet the requirements, it shows that the system designed to meet the requirements. Based on the requirements of system stability, so the overall function of the system to do the stability test, the test mode through a number of measurements to observe the success rate to see whether it can meet the requirements, the test results shown in Table 1.

Through the above test data, it can be seen that the stability of the various parts of the entire system function is very high, the success rate are more than 95%, to meet the requirements of the system's normal operation of the reliability of the system can meet the requirements of the system design, the completion of the design of the laboratory system to meet the expected goals.

The comparison results are shown in Table 2, and it can be seen that the fuzzy inference-based PID controller has a steady state error of only 2% compared with the conventional PID controller. The fuzzy PID controller integrates the advantages of PID control and fuzzy control, thus completing the dynamic adjustment of the PID control parameters, the system's dynamic response curve is good, the response time is short, the overshooting amount is small, the steady state accuracy is high, and it can realize the automatic adjustment of the ventilation volume according to the change of the environment in time to ensure the safe, reliable and stable operation of the ventilation system. Using computer simulation software, the designed conventional PID controller and fuzzy PID controller comparison simulation experiments, compared with the conventional PID control, fuzzy PID control has a better control performance, but also for the design and debugging of the actual integrated ventilation system to provide a practical significance and theoretical basis.

In this paper, the environmental data (including temperature, carbon monoxide, oxygen, hydrogen sulfide, etc.) of a family for a period of time were collected, with a total of 90 sets of experimental data. The same environmental parameter is taken to be averaged after repeated measurements for many times to get the optimal value after the fusion of the environmental data within the family room, and it is taken as the real value with high reliability. Some of the environmental data collected are shown in Table 3.

5.2.1

Analysis of indoor air quality monitoring

In this test experimental scenario, due to the indoor equipment to realize the complete networking. During the experimental test, the monitoring and reporting function of each indoor environmental parameter is realized by each monitor, and the ventilation system is turned on and off, so as to verify the effectiveness of the IoT ventilation system.

The single-day CO2 concentration change curve in the living room-bedroom is shown in Figure 3, and the single-day PM2.5 concentration change curve in the living room-living room. During the night rest period (00:00-8:00), the CO2 concentration in the bedroom is above and below 800-2000 ppm, and some instantaneous values can exceed 2000 ppm, and the bedroom is closed at this time with the inner door and outer window closed. At around 8:00 a.m., the experimenter woke up and got up (point 1), at this time, the platform opened the ventilation system. 8:10 a.m., at this time, the indoor CO2 concentration of 1380 ppm, 10 minutes later, the indoor CO2 concentration dropped to about 900 ppm. At this time, the experimenter closed the outside window of the bedroom and opened the inside door of the bedroom, and the indoor CO2 concentration dropped to 785 ppm at 8:36 p.m. Subsequently, until the experimenter returned home at 2:40 p.m., the indoor area was unoccupied (the outside window of the bedroom was closed), and the CO2 concentration of the bedroom was maintained at about 400-500 ppm.

Figure 3.

Variation curve of single-day CO2 concentration in bedroom

The experimenter returned home at 2:40 p.m. and entered the bedroom (point 3), where he was working on his writing at the bedroom desk, and there was a sustained increase in the indoor CO2 concentration, as shown in points 3-4, during which there was an instantaneous value of 1,381 ppm.

The platform turned on the ventilation system after noticing the exceedance of the standard, and the indoor CO2 concentration dropped rapidly at 4:30 p.m. (point 4), falling to 510 ppm at 4:43 p.m. The graphs show that the indoor CO2 concentration dropped to 510 ppm. Points 5 and 6 in the figure show the rapid decrease of indoor CO2 concentration as a result of the platform's operation to turn on the ventilation system. All of the above shows that the CO2 concentration in the bedroom can be reduced to below the safe level or even lower by turning on the ventilation system and reminding operation of the platform.

The pollutant situation in the living room is shown in Figure 4, where the PM2.5 concentration is mainly influenced by the behavioral activities of the experimenter. For example, during the rest period (00:00-8:00) the PM2.5 concentration in the living room was maintained above and below 50 μg/m³, and rarely exceeded 50 μg/m³. During the evening break, the outdoor windows in the living room were closed and the kitchen window was open. On the day of the test, the outdoor AQI was 42, excellent. The experimenter woke up at 8-9 a.m. and moved around the living room, and the experimenter smoked inside the living room, and the PM2.5 concentration rose rapidly due to smoking at point 1-point 2 in the figure, and the PM2.5 concentration rose from 42 μg/m³ to about 200 μg/m³. The platform reminds the real and turns on the ventilation system. PM2.5 concentrations dropped to 35 μg/m³ after a period of climbing. All the experimenters ate in the living room around 11:30 p.m. (point 3) and turned on the natural gas heating stove, at which time the PM2.5 concentration in the living room due to the increased human activities welcomed a climb, as shown in point 3-point 4, from 53 μg/m³ to about 140 μg/m³ and reached the highest value of 150 μg/m³ in point 4. The platform turned on the ventilation system and the PM2.5 concentration was decreasing to about 50 μg/m³. The same is true for points 5-6, where PM2.5 concentrations were exceeded due to behavioral activities such as heating and smoking in the room by the experimenter. As can be seen from the above, when there is no activity in the living room, the PM2.5 concentration can be maintained at a low level range (around 50 μg/m³), and once there is activity (especially smoking) there is a spike in the PM2.5 concentration, but after the operation of turning on the ventilation system as prompted by the platform, a short period of significant decline can be found.

Figure 4.

Variation curve of single-day PM2.5 concentration in living room

5.2.2

Analysis of kitchen air quality and hazardous gas monitoring

The pattern of change of pollutants in the kitchen is also very obvious, the pollutants monitored in the kitchen are CO2 and PM2.5 concentration, but only PM2.5 concentration is used as a control object. PM2.5 is the main source of cooking activities in the kitchen, and the change curve of PM2.5 concentration in the kitchen in a single day is shown in Figure 5. Area 1, Area 2, and Area 3 show the changes of pollutants during cooking time for breakfast, lunch, and dinner, respectively.

Figure 5.

Variation curve of single-day pollutant concentration in kitchen

During the breakfast cooking time, its PM2.5 concentration reached a maximum of 225 μg/m³ because the fan was not activated. During the lunch cooking time, the production started at 11:45 a.m. (point 1) An extremely rapid increase in PM2.5 concentration then occurred. The platform issued a recommendation to then turn on the fan in the kitchen for exhaust ventilation (point 2). The PM2.5 concentration in the kitchen dropped from 80.4 μg/m³ to 46 μg/m³ in 10 minutes. When the experimenter was stir-frying a dish, the PM2.5 concentration reached 101.4 μg/m³, but it was able to fall to 71.6 μg/m³ in 5 minutes without the fan operating and continuing to rise. The platform turned on the fan during the dinner cooking time, but the experimenter turned it off. During the cooking period, the PM2.5 concentration increased from 44.4 μg/m³ to 152.2 μg/m³ (point 4 - point 5). Points 6 - 7, where the experimenter boiled water in the kitchen and did not turn on the fan, revealed that the PM2.5 concentration significantly exceeded the standard (up to 115.8 μg/m³).

The hazardous gas alarm thresholds are shown in Figure 6. In addition to the pollutant monitoring function in the kitchen space, the hazardous gas alarm function is implemented and provides for an immediate alarm when the collected voltage value is greater than or equal to approximately 0.6.

Figure 6.

Dangerous gas alarm threshold