Forecasting The Crude Oil Output With Improved BP Neural Network

Crude oil output is the basis to analysis petroleum reservoir performance and make oil field development planning. In order to obtain the connection weights of minimum error energy function between the neurons, based on the theory of artificial neural network, modifying zero residual with Gauss-Newton method to the symmetric positive definite matrix of the second derivative of error energy function, to determine the search direction and achieve the most rapid decline. Owing to the improvement, the trained neural network has sufficient ability to learn and generalize. Isometric and non-isometric real time estimate is both suitable for large-scale crude output forecast and small-scale block output forecast. The improvement overcomes the computational complexity of "simulated annealing algorithm", can find the global optimal solution in a short period. Used in the crude output forecast, and compared with the actual error, it shows that the error is less than 1%, so the effect is remarkable.