中文摘要：

传统BP神经网络解耦算法可实现六维力传感器解耦,但在训练过程中容易出现振荡,收敛速度慢,陷入局部极值等问题,应用效果不理想。文中提出一种基于改进粒子群BP神经网络算法,应用到六维力传感器解耦,提高了线性解耦算法精度,同时利用粒子群算法（PSO）对BP神经网络进行优化,利用适应度函数对训练过程中得到的解进行评价,追随当前最优解搜寻全局最优。实验仿真结果表明：在六维力传感器解耦应用中,该算法实现容易,收敛速度快,解耦精度高,达到了预期的应用效果。

英文摘要：

Traditional BP neural network took advantages of approximating nonlinear systems with arbitrary precision to decouple.But in the training process,there were some problems such as concussion,slow convergence speed and local extreme. The application was not ideal.This paper proposed an algorithm based on modified particle swarm BP neural network algorithm and it was applied to study six-axis force sensor decoupling.The algorithm avoided the lower accuracy of traditional linear decoupling algorithm,used PSO to optimize BP neural network,evaluated the solution resulted by training through fitness function,and followed the current optimal solution to search global optimum.Simulation results show that the algorithm is easy to be implemented as well as fast convergence and high decoupling precision which achieves the desired application effect.