To solve the problem of scattering and dose-rate setting for on-site calibration of fixed X, gamma radiation dosimeter, and to ensure the safe and stable operation of nuclear facilities, a portable X-ray irradiation device and self-shielded X-ray irradiation device were established based on X-ray machine. Based on Monte Carlo simulation, machine learning algorithms and experimental methods, the scattering radiation correction of X-ray radiation field was completed and the radiation field dose rate setting technology was studied. Firstly, MCNP was used to simulate the radiation characteristics of portable X-ray reference radiation field and self-shielded X-ray reference radiation field, such as uniformity, scattered radiation and energy spectrum distribution. The obtained results were compared with experimental data to validate the effectiveness of the numerical simulation method and the feasibility of applying the device to field calibration. Secondly, an ambient scattered radiation correction system for on-site calibration was built based on machine learning algorithm, and the mean absolute error (MAE) was used to evaluate the performance of the machine learning model. Finally, on-site calibration experiments were carried out by combining the irradiation device and secondary ionization chamber with the ambient scattered radiation correction system built. The results show that both the portable X-ray irradiation device and self-shielded X-ray irradiation device can provide reference radiation field meeting the requirements of GB/T 12162.1―2000. The relative error of calibration factor and laboratory calibration factor of fixed X, gamma radiation dosimeter calibration based on environmental scattered radiation correction system is less than 6.2%, meeting the requirements of on-site calibration work.