Big data and artificial intelligence based early risk warning system of fire hazard for smart cities

Abstract

Driven by information technology, big data provides new development opportunities for city construction. People use multiple scientific advancements such as the Internet of Things (IoT) for data acquisition and Artificial Intelligence (AI) for big data analytics to enhance the integration and sharing of data and optimize the basic standards of smart cities. Past few years, the concept behind the Internet of Things has been a major research topic in the development of smart cities, education, industry, and commerce. Services and applications of IoT are the major factors for creating a sustainable urban life that is employed by smart cities. The stakeholders of smart cities become more aware, efficient, and interactive using Information and Communication Technology (ICT) in IoT. The applications of smart cities based on IoT have been increased in number which leads to production and increase in the amount of data and its processing. Moreover, the city stakeholders and governments take prior actions/precautions for processing the collected data from the IoT devices and predicting the future consequences for securing a sustainable environment. Artificial Intelligence is one of the key research techniques which several researchers have analysed and proved to be the best in improving the performance of detecting fire hazard in smart cities. In this research, a Deep Belief Network (DBN) with Recurrent LSTM Neural Network (R-LSTM-NN) is proposed for prediction of big data that are collected from smart cities based on IoT. Moreover, the proposed model mainly concentrates in predicting the fire hazard values that gathered from smart cities using IoT devices. The simulation results show that the proposed technique proves to be better when compared with other existing techniques in terms of accuracy, precision, recall, and F-1 score. The proposed model detects the fire outbreak with a 98.4% of accuracy that having 0.14% of minimal error rate. Furthermore, the proposed model can be used for various prediction problems that are faced by smart cities.