<![CDATA[Fire is a problem that can occur anywhere, whether in office buildings, housing or public facilities. Delays in handling result in losses, whether it be loss of life or material. The occupants of the building must receive information or early warning when a fire occurs so they can evacuate immediately. The fire alarm system that is commonly used is a conventional fire system, but due to several reasons and certain factors, the fire alarm often does not work properly and actually sends a false alarm. In this research, there are 3 sensors, namely the MQ-2 sensor, TMP102 sensor, and flame sensor which are connected to the Arduino Uno microcontroller. The Arduino Uno microcontroller is embedded with a decision tree method as an output decision provider. There are 3 processes, namely the process of determining the dataset, forming a decision tree and forming rules. This system has 3 attributes that are used to detect condition status, namely temperature, fire light intensity and smoke content. From the results of several tests carried out, it is known that the percentage error in reading the TMP102 temperature sensor is 2.1% and the MQ-2 gas sensor can read indoor gas levels well, where the sensor reading value is directly proportional to the output voltage produced, namely the higher the smoke content. that is detected, the higher the sensor output voltage value. From the results of the flame sensor test, you can read the ADC detected by the sensor on the fire source based on the distance of the sensor to the fire source. In system testing using the Decision Tree method, a success percentage of 100% was obtained. This shows that the system can work well in determining room conditions based on input from sensors.]]>
Copyrights © 2023
