The invention discloses a battery capacity predication method and pre-warning system for an unmanned vehicle on a heavy rain road. According to the battery capacity predication method for the unmannedvehicle on the heavy rain road, by using a plurality of sensors to form a sensor network, multiple running environment factors of the unmanned vehicle on the heavy rain road can be comprehensively considered; rainfall resistance of a vehicle body is measured by using a force sensor; integrated data fusion is carried out by using a fusion coefficient; by using a genetic algorithm to carry out weighted coefficient optimization on collected data information, the influences of the different running environment factors on the vehicle battery capacity can be distinguished so that the obtained datastructure is more representative; the battery capacity of the unmanned vehicle under an extreme heavy rain environment is predicated in real time by using a two-layer neural network; and various quantitative and qualitative variable factors under a non-linear environment are sufficiently considered through the adoption of the the neural network. Compared with the common SOC battery capacity predication method, the battery capacity predication method for the unmanned vehicle on the heavy rain road has the advantages that the method is more intelligent, the obtained battery capacity predicationresult is more accurate, and a very good pre-warning function can also be realized.