基于深度学习机制的火警协议智能分析模型

doi:10.16180/j.cnki.issn1007-7820.2019.01.0017

摘要/Abstract

摘要：

火灾自动报警器品牌和型号众多,不同品牌或型号的数据协议均有所不同,这为智慧消防的信息集成增加了难度。为解决这一问题,文中提出一种将深度学习技术与火警协议分析模型相结合的SSAESMFA模型。该模型基于栈式稀疏自动编码器和Softmax回归,能够通过对已知消防设备的协议数据进行深度学习,实现对未知协议数据的识别。经过训练,SSAESMFA对协议数据的识别准确率能达到99.83%,实验表明,该模型有较高的对消防设备协议数据的特征提取分类能力,可有效提高智慧消防的信息集成效率。

关键词: 火灾自动报警器, 智能分析, 深度学习, 自动编码器, Softmax回归, 特征提取

Abstract:

The automatic fire alarm had many brands and models, The data protocols of different brands or models were not the same, which made it difficult for information integration of smart fire.To solve the problem,this paper presented a SSAESMFA model that combined deep learning technology with Fire protocol intelligent analysis model.The model based on stack sparse auto encoder and softmax regression,provided the identification of unknown protocol data through deep learning of known protocol data of fire equipment.After training, The accuracy of SSAESMFA in identifying fire alarm protocol was 99.83%, Experiments showed that the model had a high feature extraction and classification ability of fire equipment protocol data,effectively improved the efficiency of information integration of smart fire.

Key words: automatic fire alarm, intelligent analysis, deep learning, auto encoder, Softmax regression, feature extraction

中图分类号:

TP183

史册,裘炅. 基于深度学习机制的火警协议智能分析模型[J]. 电子科技, 2019, 32(1): 80-85.

SHI Ce,QIU Jiong. The Fire Protocol Intelligent Analysis Model Based on Deep Learning[J]. Electronic Science and Technology, 2019, 32(1): 80-85.

图/表 8

图1

图2

图3

图4

图5

表1

图6

表2

参考文献 16

[1]	王寅良 . 基于协议智能分析的消控联网平台研究和实践[D]. 杭州:杭州电子科技大学, 2015.
	Wang Yinliang . Based on protocol intelligent analysis of the research and practice of fire alarm network platform[D]. Hangzhou:Hangzhou Dianzi University, 2016.
[2]	张绍龙 . 基于BP神经网络算法的火灾报警智能分析的研究与实践[D]. 杭州:杭州电子科技大学, 2016.
	Zhang Shaolong . The research and practice of fire intelligent analysis based on BP neural network[D]. Hangzhou:Hangzhou Dianzi University, 2016.
[3]	赵丽辉 . 基于BP神经网络的火警误报优化相关问题的研究与实践[D]. 杭州:杭州电子科技大学, 2017.
	Zhao Lihui . Research and practice on fire alarm misinformation optimization based on BP neural network[D]. Hangzhou:Hangzhou Dianzi University, 2017.
[4]	Kolen J F, Kremer S C. Gradient flow in recurrent nets: the difficulty of learning LongTerm dependencies[J]. Wiley-IEEE Press, 2001,28(2):237-243 doi: 10.1109/9780470544037.ch14
[5]	Yoshua Bengio . Foundations and Trend? in machine learning[M]. NY,USA:Now Publishers, 2014.
[6]	徐凤荣 . 基于模糊神经网络的智能火灾探测报警系统的研究[D]. 秦皇岛:燕山大学, 2006.
	Xu Fengrong . Research on intelligent fire detection and alarm system based on fuzzy neural network[D]. Qinhuangdao:Yanshan University, 2006.
[7]	傅天驹 . 基于深度学习的林火图像识别算法及实现[D]. 北京:北京林业大学, 2016.
	Fu Tianju . Forest fire image recognition algorithm and realization based on deep learning[D]. Beijing: Beijing Forestry University, 2016.
[8]	Rose-Pehrsson S L, Shaffer R E, Hart S J , et al. Multi-criteria fire detection systems using a probabilistic neural network[J]. Sensors & Actuators B Chemical, 2000,69(3):325-335. doi: 10.1016/S0925-4005(00)00481-0
[9]	Bahrepour M, Meratnia N, Havinga P J M .Use of AI techniques for residential fire detection in wireless sensor networks[J]. Ceurws, 2009,35(4):311-321.
[10]	Bengio Y, Lamblin P, Popovici D, et al. Greedy layer-wise training of deep networks [C].Rio:International Conference on Neural Information Processing Systems, 2007.
[11]	Salimans T, Kingma D P. Weight normalization: a simple reparameterization to accelerate training of deep neural networks [C].Boston:30 ^th Conference on Neural Information Processing Systems , 2016.
[12]	Vincent P, Larochelle H, Lajoie I , et al. Stacked denoising autoencoders: learning useful representations in a deep network with a local denoising criterion[J]. Journal of Machine Learning Research, 2010,11(12):3371-3408. doi: 10.1016/j.mechatronics.2010.09.004
[13]	徐德荣, 陈秀宏, 田进 . 稀疏自编码和Softmax回归的快速高效特征学习[J]. 传感器与微系统, 2017,36(5):55-58. doi: 10.13873/J.1000-9787(2017)05-0055-04
	Xu Derong, Chen Xiuhong, Tian Jin . Fast and efficient feature learning algorithm based on sparse autoencoder and Softmax regression[J]. Transducer and Microsystem Technologies, 2017,36(5):55-58. doi: 10.13873/J.1000-9787(2017)05-0055-04
[14]	毛勇华, 桂小林, 李前 , 等. 深度学习应用技术研究[J]. 计算机应用研究, 2016,33(11):3201-3205.
	Mao Yonghua, Gui Xiaolin, Li Qian , et al. Study on application technology of deep learning[J]. Application Research of Computers, 2016,33(11):3201-3205.
[15]	Friedman J, Hastie T, Tibshirani R . Regularization paths for generalized linear models via coordinate descent[J]. Journal of Statistical Software, 2010,33(1):1-8.
[16]	Abadi M, Agarwal A, Barham P , et al. TensorFlow: large-scale machine learning on heterogeneous distributed systems[J]. Arxiv, 2016,16(57):1-19.

训练次数	SSAESM决策准确率/%	BP神经网络决策准确率/%
20	95.44	87.65
40	98.95	88.10
60	99.53	88.94
80	99.65	89.48
100	99.81	90.34
120	99.83	91.03

训练次数	SSAESMFA训练误差
20	0.062 5
40	0.019 2
60	0.013 4
80	0.004 8
100	0.004 0
120	0.001 6