煤矿救援蛇形机器人环境建模方法研究

煤矿救援机器人的研究对煤矿救灾工作的顺利开展有着重要的现实意义,简要分析了煤矿救援机器人在环境建模方面的研究现状,针对煤矿事故发生后,救援蛇形机器人如何在恶劣的井下进行环境识别和建模,提出了一种变结构模糊神经网络的多传感器数据融合算法。重点讨论了该算法的原理,结合煤矿井下的特殊环境和蛇形机器人自身结构特点,采用超声波传感器、红外测距传感器、激光雷达传感器组来获得障碍物的距离、位置信息及环境类型信息,然后,利用模糊神经网络对这些信息进行融合,采用改进的BP算法对网络进行学习,通过对结论网络权值的调整,择优选取模糊规则,从而自动的调节模糊神经网络的结构。以机器人在靠近障碍物时的八类典型环境标志为依据,通过模糊神经网络识别出障碍物的形状,完成环境的建模。利用实验获得的一组数据进行了仿真,结果表明该算法实现了在不同环境中模糊隶属函数的自动生成和模糊规则的择优提取,适用于复杂的非线性系统,对于煤矿救援蛇形机器人的环境识别和建模是一种行之有效的方法。

Research on environment modeling method of the coal mine rescue snake-like robot

Research of coal mine rescue robot has important practical significance to the smooth functioning of the coal mine rescue work. In this paper, the present research status of coal mine rescue robot in terms of environmental modeling is analyzed. In view of the coal mine after the accident, how the rescue snake-like robot to carry out environmental recognition and modeling in the poor underground, a kind of fuzzy neural network with changeable structure algorithm based on multi sensor data fusion is put forward. The principle of the algorithm is focused. Firstly, combined with the special environment of coal mine and the structure characteristics of the snake like robot, ultrasonic sensors, infrared sensors and laser radar sensor group are used to obtain the distance to the obstacle, the position information and the type of environment information. Then, these information are fused using fuzzy neural network and the network is trained using the improved BP algorithm. The weights of conclusion network are adjusted and the fuzzy rules are selected accordingly, the structure of fuzzy neural network is adjusted automatically.Finally, eight typical environment marks is analyzed, the shapes of obstacles is identified by the fuzzy neural network, and environmental modeling is completed. A group of experimental data is simulated, the results show that the automatic generation of fuzzy membership function and the preferential extraction of fuzzy rules are realized in different environments, and the algorithm is suitable for complex nonlinear system, this algorithm is an effective method for the coal mine rescue snake like robot to built environment identification and modeling.