面向多传感器信息融合移动焊接机器人PEMFC/Li-ion电池系统能量分配优化控制研究

针对长焊缝难以实现连续高精度焊接的难点，开展面向多传感器信息融合的焊接机器人PEMFC/Li-ion电池系统能源分配与优化控制研究。系统功率需求是能源分配的基础，为实时获取功率需求并确定系统运行状态，采用双目视觉传感器和激光位移传感器信息融合获取焊缝跟踪信息。针对当前燃料电池汽车能量优化方法不能满足焊接机器人对系统供能实时性、鲁棒性和稳定性要求的现状，本项目创新性地采用F3RBM-SVM进行模式识别、基于CNN-DS证据理论实现多传感器信息融合；基于模拟退火改进增量电导法和分布估计法分别进行燃料电池和锂电池管理；在此基础上，将基于规则优化的混合能量管理和基于云模型与组合赋权法的焊接机器人综合性能评价相结合，提高混合动力系统的实时性、鲁棒性和稳定性，使焊接机器人平稳、高精度的运行。为更好地推广燃料电池混合动力系统在焊接机器人中的应用奠定基础，对改善混合动力系统动态特性的科学性具有重要意义。

Aiming at the difficulty of continuous and high-precision welding of long weld, it is necessary to study the energy distribution and optimal control of PEMFC/Li-ion battery system of welding robot based on multi-sensor information fusion. Because the power demand of the system is the basis of energy matching, in order to obtain the power demand in real time and determine the operation status of the system, binocular vision sensor and laser displacement sensor information fusion is used to obtain the seam tracking information. In view of the current situation that the energy optimization method of fuel cell vehicle can’t meet the real-time, robust and stability requirements of welding robot for system energy supply, this project innovatively adopts F3RBM-SVM for pattern recognition, realizes multi-sensor information fusion based on CNN-DS evidence theory. Based on simulated annealing improved incremental conductance method and distribution estimation algorithm, proton exchange membrane fuel cell (PEMFC) and lithium battery (Li-ion battery) are managed respectively. On this basis, the hybrid energy management based on rule optimization and the comprehensive performance evaluation of welding robot based on cloud model and combined weighting method are combined to improve the real-time, robustness and stability of hybrid power system, so that the welding robot can run smoothly and with high precision. The above research will lay a foundation for the better promotion of the application of fuel cell hybrid system in welding robot, and it is of great significance to improve the science of dynamic characteristics of hybrid system.