A multimodel fire detection and alarm system for automobiles using internet of things and machine learning

Abstract

Fire accidents in vehicles lead to the loss of human lives. The fire detection and alarm systems are often error-prone and respond to nonactual indications of fire presence, known as false alarms. The proposed model detects the fire at the smoldering stage and buzzes an alarm if an actual fire or smoke is detected. This system can achieve this real alarm using multiple internet of things-based sensors, namely smoke/gas, flame, temperature, and a visualization camera. The visualization camera continuously captures images of the vehicle to check the existence of fire. Machine learning algorithms are executed on the sensor and image dataset to reduce false alarms and achieve high accuracy of results by using various performance metrics. K E Y W O R D Sfire detection and alarm system, flame sensor, gas sensor, internet of things, temperature sensor, visualization camera INTRODUCTIONLoss of human lives and economic losses are observed every year due to fire accidents. Many types of research were carried out to study the cause and explosion of fire and the efficient ways to control or prevent the fire outbreak in automobiles. Research findings state that most of the fire accident takes place during summer due to overheating of engine and other components. Use of old vehicle and low-maintenance vehicles is highly prone to get fire. Identified reasons for fire outbreaks are gas Leakage, electrical short circuit, electric car battery, AC problems, etc. The leakage of toxic gases due to car burning is hazardous.In recent days, the probability of fire accidents in vehicles has been higher, and it could be even more in electric cars. In fire detection, sensors are used to detect the fire; a single sensor cannot sense the fire effectively. It is high time to detect fire beforehand. Smoke is a significant sign for detecting early fire conditions. A number of fire detection systems were proposed in the past few years. Most approaches use different visual elements, including flame colors, motion, and geometrical contour. Integrated sensors work efficiently in Fire environments.These Integrated alarm systems allow fast identification, alarm alerts, and even fire extinguishing. This can be achieved by using integrated sensors.In this article, a fire detection and alarm system is proposed to detect fire in earlier stages. The proposed model uses integratedsensors and a camera for fire detection. Therefore, the proposed model avoids the rising of false alarm. The data is collected from the sensors such as smoke, temperature, and flame and are used to train the machine using the machine learning algorithms. Similarly, the image patterns of smoke and flame are also collected and used to train the machine. Therefore, when the fire is detected, the alarm will be buzzed.The remainder of the article is organized as follows: Section 2 discusses the related work. Section 3 presents the proposed system. Section 4 elucidates the components of the proposed architecture. Section 6 confers the efficiency of the proposed system and discusses the ...