中文摘要：

为了提高深海机器人多传感器系统的故障诊断能力，针对深海环境的特殊性，以及深海机器人的工作特点，提出了基于两级集成神经网络的故障诊断方法，建立了诊断子网络，从不同的方面对机器人故障进行诊断，并利用决策融合网络将诊断子网络的诊断结果进行融合会诊，得到系统最终的诊断结论；基于某型深海机器人海中试验数据进行计算机仿真实验的结果，验证了该方法的有效性和可行性。

英文摘要：

Research is undertaken to improve the fault diagnosis abilities of Deep-seabed Mining Robots＇ （DSMRs） multi-sensor system. For the specificity of the deep-seabed environment, and the trait of DSMRs work, a method based on an integrated neural network was proposed and a two-level neural networks model was constructed. From different aspects of robot fault diagnosis, and using the decision fusion network to fusion the results of consultation, find the final conclusions of the diagnosis system. Computer simulations using experimental data from a DSMR show that the proposed method is feasible.