Internet of Things (IoT)-Driven on Evaporative Cooling System for Tropical Greenhouse Environmental Control

The impact of climate change on tropical agriculture can be mitigated by using controlled environmental conditions in tropical greenhouse buildings with evaporative cooling. Precision agriculture can be applied by adopting technology based on the Internet of Things (IoT) with easy access and real-time monitoring. A study has been conducted to design and manufacture an IoT-based evaporative cooling control system for cultivating horticultural crops in tropical greenhouses. The system consists of an environmental monitoring node and an air cooler actuator control node. Data temperature, relative humidity, and the response of the control actuator can be monitored in real-time via a cloud server. The study also discusses Vapor Pressure Deficit (VPD) as an important factor that needs to be considered in controlling greenhouse environmental conditions. The study used a linear regression test, validation test, analysis of VPD, calculation of the accuracy of the evaporative cooler, as well as evaluation of packet loss.


Introduction
This study builds on the issue of climate change that has impacted the agricultural sector.Southeast Asia is one of the regions most susceptible to climate change.Climate change adaptation is needed in the agricultural sector to reduce climate change's impact on plant growth.Therefore, agricultural productivity is well maintained [1].Precision agriculture is an approach that can adapt to climate change and maintain crop quality to preserve productivity gains.Precision agriculture can increase productivity while reducing environmental impacts by optimizing resource utilization using commensurate technology.One of the more optimal adoptions of precision agriculture is the application of Internet of Things (IoT) technology [2].Horticulture plays a crucial role in Southeast Asian agriculture within the tropical climate zone.Challenges like inconsistent product quality and the impact of climate change must be addressed.Improving the horticulture production system is essential for ensuring a continuous and consistent supply.Thus, careful consideration of environmental factors, including temperature and relative humidity, is necessary to enhance the tropical agriculture production system [3].
Greenhouse technology aims to create optimal environmental conditions for plant growth and can be applied to climate change adaptation.Greenhouse conditions can be set up with temperature and humidity according to the best growth rate of crops.The obstacle to cultivating commodities in greenhouses is the high temperature inside.Therefore, it is necessary to have an appropriate cooling system to stabilize the temperature in the greenhouse by creating an automatic cooling system that can monitor and control the microclimate in the greenhouse.An evaporative cooling system has been used in a greenhouse for cultivation in a controlled environment.Greenhouse conditions are set with temperature and humidity according to the best growth rate of crops [4].Evaporative cooling systems can be an alternative to lowering the temperature of the plant growing space by utilizing water to cool and increase humidity in the airflow [5].
The conditioning process needs to adjust to the target or optimal conditions for plant growth, so it is necessary to apply an IoT-based environmental condition control system with easy access monitoring and control.The IoT system utilizes internet network technology and automation to increase greenhouse control, time, and labor efficiency, and optimize plant development and growth processes.Wireless Sensor Network technology can also be an alternative, allowing wireless transmission of greenhouse condition acquisition data from Arduino to a web server, which is then stored in a database so that operators can monitor greenhouse conditions in real time at any time.
Therefore, an evaporative cooling control system can be designed and developed by adopting the application of wireless sensor network technology in control device component, not only for on-off fan automation but also for fan speed regulation based on the readings of temperature and humidity sensors so that the cooling and environmental conditioning processes are optimal.An IoT technology should be adopted to observe temperature and humidity, control the environments, and send data to a cloud server to be accessed easily anywhere in real time.The objective of this study is to design and assess the model of IoT-driven evaporative cooling control system for tropical greenhouse environment.

Materials and methods
Figure 1.Framework of the design and implementation of monitoring and control system in a greenhouse with IoT-driven

Research Framework
The research framework and equipment design can be seen in Figure 1.The evaporative cooling system consists of three main components: sensors, microcontrollers, and actuators connected using IoT.The sensors installed include temperature and humidity sensors, while the actuator is an evaporative air cooler.The evaporative cooling system will be applied to greenhouse buildings to directly observe the plant growing room's temperature and humidity.The global sub-system, i.e., the cloud, is a storage, calculation, analysis, and deployment platform for intelligent systems.Then, it will

Monitoring and Control Function
The function of the system includes monitoring and controlling function.The system starts with initializing and program setup.The observation was conducted by wireless sensor network (WSN) to observe the temperature and humidity of greenhouse environment.Data of the environment was collected then it will be sent to a cloud server Agrieye for online storage.The control function utilizes the data that has been collected.To maintain temperature and humidity conditions in greenhouses or plant growing rooms, as well as the use of evaporative cooling systems.The evaporative cooling system as an actuator can be adjusted to respond the temperature changes so that environmental control and energy use can be optimized.The monitoring and controlling function are clearly described based on the activity sequence in Figure 2. The integration of sensors and actuators with the Agrieye Cloud system, enabling autonomous data retrieval and actuation for monitoring and control, has been successfully implemented in previous research for a mini-plant factory system dedicated to the production of horticultural crops [6].Precision agriculture systems are intrinsically linked to the surveillance and regulation of the environment.This framework is crucial for maintaining the ambient conditions or microclimate in accordance with the needs of the plants [7].

Sensor Calibration and Data Validation
A range of cost-effective sensors and devices have been created and employed within the agricultural sector.These technologies have been put into practice to enhance agricultural output by advancing the knowledge and techniques of farming for the progression of smart agriculture [8].This research conducted DHT22 temperature and humidity sensor.The DHT22 sensor was calibrated.The sensor calibration procedure aims to confirm the precision of sensor readings by employing a measuring instrument alongside a reference measurement.This calibration process yields an equation used in validating the sensor's accuracy.The calibration employed the linear regression test method.One validation step involves graphing the data, with the X-axis representing sensor readings and the Y-axis denoting data from a reliable reference tool.The R2 value indicates the relationship between X and Y variables.R2 ranges from 0 to 1.The higher value approaching one, indicating a better fit.The R2 value equation can be written as follows [9,10]: Data sensor readings were validated using the Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE).To determine the error value of the system, RMSE and MAPE were calculated.The RMSE method is a system validation test method based on the error value of the prediction results.The RMSE equation can be written as follows [10]: The MAPE method is a system validation test method with error values in percentage.MAPE is a method used to analyze the reading error values of the system that was built.The MAPE equation can be written as follows [10]: The value of mean the parameter values using reference tools.The value of  mean the parameter values using sensors.The value of n is equal to the amount of data.RMSE value close to 0, indicates that the data obtained is accurate or valid.A low MAPE percentage value indicates the system is increasingly accurate [10].

Assembly of Components
The assembly of components on the environmental monitoring node of the greenhouse is done by connecting the main components, namely the Node MCU ESP8266 and the DHT22 sensor.The assembly of components on the environmental monitoring and control node of the greenhouse is carried out by connecting the main components, namely the Node MCU 1.0 (ESP-12E Module) board ESP8266, DHT22 sensor, I2C LCD, and 4-channel relay module.Subsequently, the ESP8266 board is connected to the ESP32 through the RX-TX pins for data transfer and transmission to the cloud server.The schematic diagram of the system can be seen in Figure 3.

Implementation Scheme of the system
The environmental monitoring nodes consist of 3 transmitter nodes, each comprising an ESP 8266 microcontroller and a DHT22 sensor for collecting temperature and humidity data within the plant growth area at various points.Subsequently, they transmit this data to the main monitoring node (receiver) for data collection and processing, which is then sent to the cloud server.On the other hand, the outdoor environmental monitoring nodes consist of an ESP32 microcontroller and a DHT22 sensor for collecting environmental data outside the greenhouse, directly transmitting it to the cloud server.This temperature and humidity data is relayed to the cloud server and can be observed in real-time via the web at www.agrieye.tp.ugm.ac.id.The schematic layout of the control system, sensors, and evaporative cooling fan actuators refers to Figure 1.

Evaluation System Analysis
The evaluation system analysis was conducted to the accuracy system and the percentage of system work related to the data logging.The online data logging was evaluated by packet loss.A calculation of accuracy evaluated the cooling system's performance Evaluation results will be obtained, which can provide several recommendations for developing an IoT-based cooling system model for greenhouses.Actuator accuracy is defined as a precision actuator that controls the microclimate so that it matches the setting point value or is in the controlled zone.Speed of control is defined as the average time required for the control system actuator to control the microclimate to reach the setting point value.Calculation of system accuracy can be calculated using the following equation [11]: The SN is equal to difference between actual value and setting point ( 0 C).The SP is equal to setting point value ( 0 C).The AK is equal to the actual value of measurements ( 0 C).KTACC represents the degree of inaccuracy, denoted as a percentage, whereas ACC signifies accuracy, also quantified as a percentage.A higher accuracy percentage corresponds to an enhanced system performance.
Packet loss is the percentage of data that is lost/not sent during observation.Packet loss can occur because the system is not connected to the cloud server so data cannot be sent and stored and there are also problems with the server experiencing problems.Packet loss calculation can be calculated using the following equation: After the percentage of data loss is obtained, the system performance percentage is calculated using the following equation: As the proportion of data loss diminishes, there is a corresponding augmentation in the percentage of system performance metrics, signifying an improvement in system functionality.

VPD Analysis
The Vapor Pressure Deficit (VPD) is a crucial environmental factor that impacts stomatal behavior, photosynthesis, and overall plant growth in crops and horticultural plants.In greenhouse cultivation, VPD experiences fluctuations due to the regulation of air temperature and relative humidity through mechanisms like turning on or off evaporative systems and opening or closing roof windows.However, the specific effects of VPD fluctuations on photosynthesis and plant growth performance are not yet fully understood [12].VPD can be calculated using the following formula [13]: The Euler's number (e) is a mathematical constant approximately equal to 2.71828.T is a current temperature in degrees Celsius, and RH is a current relative humidity in percentage.While the evaporative cooling system is implemented, the data obtained will be compared with the calculation of the VPD value.

Monitor Nodes and Actuator Control Nodes
The assembly result of the system consists of two types of nodes: the main monitor as receiver and control node for actuators, and the monitoring node for observing the environment inside as transmitter and outside the greenhouse.Each node has different programs, features, and components.This monitor node has its main program function: performing microclimate sensing in an indoor greenhouse plant growth chamber, then sending the data to the monitor main node and actuator controls.These feature nodes can perform sensing functions and then send the sensing result data (as a transmitter node) to the main node (as a receiver node).Delivery data on this monitor node is carried out using ESP-Now rules with a broadcast send message system.This monitor node (transmitter node) assembly uses several components, including the ESP8266 microcontroller, cables, battery, regulator, and the DHT22 temperature sensor, which are assembled and placed inside a panel box.The hardware result is an environmental monitoring node inside the greenhouse, as shown in Figure 4(a).
The main monitoring and control nodes have several program functions, including reading temperature and RH conditions, receiving data on temperature and RH conditions from other monitoring nodes, then averaging the data readings.Subsequently, it sends all the sensing data to the cloud.After that, this node will function to compare the average readings of the environmental conditions inside the growing space with the pre-set point values; then, it will regulate the actuators.The main monitoring and control node features include reading and accumulating readings from the sensors received from other monitoring nodes, temporary data storage and retrieval, averaging incoming data, and comparing the average sensor readings with the pre-set point values.The assembly of the main monitor and control node uses several components, including a relay module, ESP8266 microcontroller, cables, DHT22 Temperature Sensor, and I2C LCD, which are assembled on the base plate Node MCU ESP8266 and placed inside a panel box.The hardware result is a main monitor and control node, as shown in Figure 4

Sensor Evaluation Test of Monitor Nodes
Every sensor used must be calibrated to ensure its readings are more accurate.Calibration is the process of determining the value of a measuring instrument by comparing it to a standard reference value, thus obtaining a calibration equation used in every stage of sensor validation.During calibration, environmental temperature data is collected using the DHT22 sensor and the UNI-T UT320D Thermometer.The reading data from the UNI-T UT320D Thermometer is used as a reference in sensor validation.Linear regression testing is performed for sensor calibration by plotting the data on a graph to find the relationship between variables, using Microsoft Excel to obtain the R square (R 2 ) value.The R 2 is also known as the coefficient of determination, which explains how much independent data can explain dependent data.R2 values range from 0 to 1, with a higher value approaching one, indicating a better fit.There are three categorizations for R2 values: strong, moderate, and weak.An R 2 value of 0.75 falls into the strong category, an R2 value of 0.50 falls into the moderate category, and an R 2 value of 0.25 falls into the weak category [9].The validation test is conducted to analyze error values using the RMSE and MAPE calculation methods for each monitoring node of the greenhouse.Testing with the RMSE and MAPE methods can indicate the level of prediction error compared to actual values.The MAPE value is considered excellent when less than 10%.Meanwhile, the RMSE value is considered better when smaller or approaching 0 [10].The results of the DHT22 sensor evaluation for each node are summarized in Table 1 below.  1 shows that each monitoring node exhibits an R 2 value that has a number near 1 and a MAPE value below 10%, which is generally regarded as highly favorable.This observation signifies a low error rate and high predictive accuracy.Furthermore, a lower RMSE value leads to a reduction in the error rate and an enhancement in prediction accuracy.

Response and Performance Test of Actuator Control Nodes
The actuator control node will regulate the response of the Evaporative Air Cooler in controlling the temperature inside the greenhouse.The actuator's response to the environmental air temperature inside the greenhouse can be seen in Figure 5.
Based on the above graph, it is known that the air cooler actuator can respond to changes in the environmental temperature inside the greenhouse.The actuator can operate at its highest speed (status 3) when the environmental temperature reaches 32°C.When the temperature drops below 32°C and reaches 30°C, the actuator stat decreases to status 2. When the temperature falls below 30°C and reaches 28°C, the actuator status drops to status 1.When the temperature reaches 28°C or below, the actuator becomes inactive.The temperature inside the greenhouse is relatively easier to control than outside the greenhouse.Therefore, if the temperature inside becomes too high, the evaporative cooling system can help lower the greenhouse's environmental temperature.The accuracy of the actuator can be defined as the precision of the actuator in controlling the micro-environmental conditions according to the set point value.The evaporative cooling control system actuator's accuracy in greenhouse is 92.86%.Based on this calculation result, the accuracy value of the evaporative cooling control system actuator is already good for controlling the microclimate in the controlled zone.For further development, to accelerate temperature and humidity control and improve accuracy, it is necessary to increase the number or capacity of the evaporative cooling actuators to expedite the microclimate control process, thus achieving the desired set point value.

Evaluation of Online Data Logging
During online data logging testing, 671 data points were missing out of a total of 2385 that should have been collected.Therefore, it can be determined that the percentage of lost data is 28.1% and the percentage of system performance in sending data to the online logger is 71.9%.The amount of missing or intransmitted data to the cloud server is still quite substantial.The failure to transmit data may be attributed to microcontrollers encountering errors during the data transmission process or due to an unstable internet connection.

Implementation of the System in the Mini Tropical Greenhouse
A temperature and humidity monitoring system, as well as evaporative cooling control, was implemented inside the greenhouse for six days.The measured parameters are the environment's temperature and RH, which are then compared with the VPD value.The data on the environmental conditions inside the greenhouse can be seen in Figure 6.
Based on the graph, it is evident that the environmental temperature rises during the day and falls at night.Conversely, RH decreases during the day and increases at night.The higher the air temperature, the lower the relative humidity.Temperature and air humidity are related to VPD, which measures the capacity of the space to accommodate moisture or water vapor at the current actual temperature.Ideal VPD conditions can optimize plant growth.According to the graph, VPD values increase during the day and decrease at night due to a corresponding drop in air humidity.The adjustment of VPD values can affect the greenhouse environment, plant growth, stomata traits, leaf gas exchange, plant biomass, yield, and fruit quality [14].Another study underscores the significance of VPD regulation in the cultivation of plants within plant factories and greenhouses [12].The other study also demonstrates the importance of VPD control during plant cultivation in plant factories and greenhouses.Elevated VPD levels result in heightened evaporation demands, disrupting the equilibrium between water availability and evaporative needs.This condition, prevalent under high VPD, leads to leaf desiccation and water stress, manifested as wilting and xylem cavitation.In this experiment, the greenhouse effectively sustained low VPD conditions below 2 kPa, which proved conducive for the growth of tomato plants [13].
If we refer to previous research, that the greenhouse effectively sustained low VPD conditions below 2 kPa, VPD values approaching a certain condition occur during the late afternoon, night, and early morning.The VPD value will increase during the daytime with rising temperatures and decreasing humidity.Therefore, a cooling system like an evaporative cooling system is necessary to help maintain the environmental temperature and humidity levels.

Conclusions
The IoT-based evaporative cooling system was designed and tested to monitor and control the environmental conditions for cultivating horticultural crops in a tropical greenhouse.This system includes monitoring and control nodes that can integrate with each other and with the cloud.The monitoring node is responsible for sensing temperature and RH environment at various points around the greenhouse.The control node compares the average temperature value with the set point to regulate the evaporative air cooler actuator.Monitoring and control data can be accessed remotely through a cloud server.The implementation of this system in our research introduces the concept of VPD as a crucial factor in greenhouse environmental control.This system has undergone evaluation and can provide additional insights if further development is pursued.Therefore, research in this field is still at a small-scale and demands more comprehensive studies for future advancement.

Figure 2 .
Figure 2. Flowchart of the monitoring and controlling system

Figure 3 .
Figure 3.The schematic diagram of the monitoring and evaporative air cooler control system based on Internet of Things for greenhouse (b) and Figure 4(c).

Figure 4 .
Figure 4. (a) The assembly of environmental monitoring nodes inside the greenhouse (as transmitter) and outside the greenhouse (b) The assembly of environmental monitoring main node (as receiver) and control node inside the greenhouse (c) Display of the main monitor node in the greenhouse

8 Figure 5 .
Figure 5.The results of monitoring the temperature inside the greenhouse and the actuator response

9 Figure 6 .
Figure 6.Observations of environmental conditions and VPD values inside the greenhouse

Table 1 .
The result of the sensor evaluation test for monitoring the greenhouse