The invention discloses a rapid topological optimization method for a heat sink device cooling channel based on deep learning. The method comprises the steps that a typical heat sink device is adoptedas a template, a traditional all-directional orthogonal punishment material density method is adopted for performing integrated optimization iteration, a density distribution diagram and a density gradient distribution diagram of each iterative step high thermal conductivity material are output as input of a deep learning algorithm and a training set, and a deep neural network obtained after deeptraining is constructed; then, preliminary iteration is performed on the heat sink device to be optimized by adopting the traditional all-directional orthogonal punishment material density method, the preliminarily-optimized high thermal conductivity material density distribution diagram and density gradient distribution diagram are output as a training starting point of the deep learning algorithm and input into the deep neural network obtained after deep training in the last step, learning and training are carried out, and the finally-optimized high thermal conductivity material density distribution diagram is obtained and output, and the topological configuration structure of the cooling channel structure is constructed. By introducing the deep learning algorithm, the speed and efficiency of the optimally-designed heat sink device cooling channel structure topological configuration are improved.