Production prediction at ultra-high water cut stage via Recurrent Neural Network

WANG Hongliang,MU Longxin,SHI Fugeng,DOU Hongen

Table 3 Predicted production of two other oilfields by LSTM model.

Time Monthly oil output of oilfield 1 Monthly oil output of oilfield 2 Actual output/t Predicted output/t Relative error/% Actual output/t Predicted output/t Relative error/% January 2018 359 499 348 874 2.96 7964 7961 0.04 February 2018 320 816 321 998 0.37 7137 7821 9.58 March 2018 345 256 344 530 0.21 7529 7893 4.83 April 2018 331 233 335 417 1.26 7462 7868 5.44 May 2018 342 588 339 255 0.97 8173 7965 2.54 June 2018 338 639 334 328 1.27 8020 7840 2.24 July 2018 348 521 340 717 2.24 8000 7876 1.55 August 2018 346 023 341 975 1.17 7968 7779 2.37 September 2018 330 353 334 579 1.28 7315 7795 6.56 October 2018 342 760 340 823 0.57 6988 7918 13.31 November 2018 329 982 333 913 1.19 6928 7712 11.32 December 2018 338 437 336 907 0.45 7481 7787 4.09