In radiation dosimetry and medical dosimetry, optical fiber dosimeters have great advantages in application due to their small size and long transmission distance. However, in some low-dose radiation environments, the water inequivalence error and angle error of optical fiber dosimeters can have significant impacts that cannot be ignored. In this paper, the effects of ray energy, ray solid angle and optical fiber probe size on water inequivalence error and angle error are analyzed by using Monte Carlo simulation. Due to the complexity of the influencing factors of the angle error, which is difficult to be analyzed precisely and quantitatively, a correction method based on BP neural network is proposed. The final validation results demonstrate that the correction scheme can reduce the total angle error of the optical fiber dosimeter to 2.72%.