Crystal oscillator based on artificial neural network

The invention belongs to the field of integrated circuits, and in particular relates to a crystal oscillator based on an artificial neural network. The crystal oscillator comprises a control voltage generation module and a crystal oscillation module; and the control voltage generation module is composed of a temperature sensor, a frequency meter, a data selector, an input data processing unit, the artificial neural network and an output data processing unit. According to the crystal oscillator disclosed by the invention, control voltage required by two ends of a variable-capacitance diode is generated through the artificial neural network; the network can approximate a nonlinear function at high precision by utilizing the intrinsic nonlinear mapping capability of the artificial neural network, so that control voltage having the nonlinear function to temperature is generated; the voltage precisely compensates deviation of frequencies of TCXO and VCTCXO due to temperature change by controlling the capacitance v