The invention discloses an earthquake motion peak acceleration prediction method based on a second-order neuron deep neural network, belongs to the field of earthquake engineering, and aims to solve the technical problem of low precision of earthquake motion prediction. The earthquake motion peak acceleration prediction method comprises the following steps: 1, selecting earthquake magnitudes, projection distances, shear wave velocities, regions, cover layer thicknesses, fault types and periods as input parameters in a data set, and taking corresponding earthquake motion peak accelerations as target parameters; 2, establishing a deep neural network comprising three hidden layers, wherein neurons are second-order elements, a hyperbolic tangent function is adopted as an activation function, amean square error function and an Adam self-adaptive optimization function are adopted for back propagation, and an average absolute error function is adopted as an evaluation function; 3, training adeep neural network model; and 4, peak acceleration prediction. According to the method, a multi-input structure and a second-order neural network are adopted so that the precision of predicting theearthquake motion peak acceleration can be improved, and the applicability of the deep neural network model can be ensured.