A hybrid bio-inspired self-organizing map neural network algorithm is proposed for path planning and task assignment for a multi-autonomous underwater vehicle (AUV) system within a mixed (dynamic and static) three-dimensional (3D) environment. A 3D hybrid bio-inspired neural network model is established to represent the underwater environment and the distribution of the neuron pheromone content gradually diffusing, centered on the source point of the neural wave. Through self-regulation of the neural wave diffusion, the targets can achieve self-adaptive capabilities. “Multiple Newton interpolation” is used to identify the real target among interference targets, and the multi-AUV system transitions from tracking the false target to tracking the real target. Based on the principle of AUV individual kinematics, a velocity vector synthesis algorithm is proposed to overcome the interference of ocean currents. Simulation studies performed in five different environments demonstrate that the proposed algorithm has high adaptability, and the potential for wide application because its neural waves can be updated in real time.

提出了一种混合生物启发自组织映射神经网络算法，用于混合（动态和静态）三维（3D）环境中的多水下机器人（AUV）系统的路径规划和任务分配。建立了一个3D混合生物启发式神经网络模型来代表水下环境，并以神经波的源点为中心逐渐分散神经元信息素含量的分布。通过神经波扩散的自我调节，目标可以实现自适应能力。“多重牛顿插值”用于识别干扰目标中的真实目标，并且多重AUV系统从跟踪虚假目标过渡到跟踪真实目标。根据AUV个人运动学原理，为了克服洋流的干扰，提出了一种速度矢量合成算法。在五个不同环境中进行的仿真研究表明，该算法具有很高的适应性，并且由于其神经波可以实时更新，因此具有广泛的应用潜力。