In this paper, a multi-attention residual spiking neural network (MAR-SNN)-based grounding grid fault diagnosis method is proposed to deal with the existing single and unintelligent problems in the diagnosis of grounding grid. Firstly, creating the grounding grid dataset for training, using the electrical impedance tomography (EIT) after re-meshing to improve imaging speed and enhancing image features between different fault levels by using the local adaptive contrast enhancement method; Secondly, the MAR-SNN model is built by a new multi-attention spiking residual block is proposed to realize the intelligent fault diagnosis of grounding grid. The residual block performs identity mapping after two spiking neurons, adopts PLIF spiking neurons and BN layer, and introduces multi-attention mechanism to improve the accuracy of model recognition separately; Finally, using EIT and the trained MAR-SNN model to construct the intelligent fault diagnosis model of grounding grid. The comparative analysis of the models shows that the performance of MAR-SNN is superior to the existing advanced models, and in the test set the accuracy is 96.31%. Among them, the accuracy of mild and medium corrosion degree can reach 100% and 97.20% respectively. At the same time, the experimental results show that the proposed method can realize the intelligent fault diagnosis of grounding grid including fault detection and level identification, so verify the effectiveness and feasibility of the method.