摘要
针对传统危化品仓库固定式监控器中监控范围小、报警准确率低的特点,研究了一款危化品仓库巡逻机器人,采用以拉依达准则改善BP神经网络融合性能的多传感器数据融合方法,通过采集泄露危化品浓度、仓库内环境温度和湿度等数据,在对数据进行拉依达去噪、归一化后利用BP神经网络进行融合输出。样机试验结果表明,该方法可有效提升危化品仓库巡逻机器人对空间环境的把握度,大幅度提高报警的准确性和可靠性,同时具备良好的传感器扩展性。
According to the small monitoring range and much false alarm of traditional surveillance equipment in hazardous chemicals warehouse,a data fusion method base on BP neural network used in patrol robot is proposed,this method firstly collected the original data of temperature,humidity and the concentration of leaked hazardous chemicals in surroundings,and filtered the noise data by pauta criterion,then normalized all data before fuse operation,and finally fused all the data from each sensor by BP neural network. The experimental result indicates that the method greatly improved the patrol robot＇s detectability of surroundings,and increased the dependability and accuracy in alarm,also it can be expanded additional sensors easily.
出处
《传感技术学报》
CAS
CSCD
北大核心
2016年第12期1936-1940,共5页
Chinese Journal of Sensors and Actuators
基金
广东省重大平台和科研立项项目(2015KQNCX241)
广东省大学生科技创新培育专项资金项目(pdjh2016b0998)
2016年广州大学华软软件学院院级科研项目(ky201601)
