Modular multiplication simplified homomorphic encryption neural network linear layer hardware acceleration method and system

The invention discloses a modular multiplication simplified homomorphic encryption neural network linear layer hardware acceleration method and system, and the method comprises the steps: receiving input weight data of a neural network model through a user side, carrying out the quantification through employing an addition quadratic power method, and then carrying out the rearrangement and transposition; the user side receives input source data of the neural network model, rearranges the source data in an img2col mode, transposes the source data, and encodes the source data into a plaintext polynomial; the user side encrypts the coded plaintext polynomial into a ciphertext corresponding to the homomorphic encryption scheme, and sends the ciphertext and the processed weight data to the hardware accelerator; and the hardware accelerator starts neural network linear layer operation on the ciphertext and the weight data under the homomorphic encryption condition, and feeds back an operation result to the user sid