Outside temperature variation and fiber strain can raise center wavelengh excursion of fiber Bragg grating (FBG),which will cause temperature and strain crossing susceptivity when measuring with FBG. A FBG system of BP neural network with artificial neural network theory is setup by using Matlab neural network toolbox,the network in gradient descent algorithm with momentum term is trained,and the results show convergence is faster. Temperature error of the network after training is less 2%,and strain error is less 5%,which prove the realization of simultaneous measurement of temperature and strain.