关键词: 火灾探测, 加权支持向量回归, 参数优化, 神经网络

Abstract: The detection of fire is complex and significant in its early age.Aiming at the shortcomings of traditional fire detection method，a new intelligent fire detection system based on Weighted Support Vector Regression（WSVR） is presented.WSVR algorithm overcomes the disadvantages of neural network such as over learning etc.，and the problem of no considering the importance of each sample in the standard SVR.The experiment results show that the intelligent fire detection system is surperior to the detection systems based on neural network and the standard SVR，it has notable detection effect and well application foreground.

Key words: fire detection, Weighted Support Vector Regression（WSVR）, parameter optimization, neural network