Production performance forecasting method based on multivariate time series and vector autoregressive machine learning model for waterflooding reservoirs
ZHANG Rui,JIA Hu
Table 5 Predicted results of cumulative outputs of all producers in each well group.
Well
group No.
Actual cumulative
production/m3
Cumulative output from machine learning/m3 Cumulative output from
history matching/m3
Relative error of
machine learning
Relative error of
history matching
1 265.1 216.7 209.9 0.183 0.208
2 201.4 232.2 236.7 0.153 0.175
3 173.3 194.1 200.6 0.120 0.158
5 155.6 181.0 185.3 0.163 0.191
6 117.0 104.5 134.5 0.107 0.150
7 113.6 129.8 94.7 0.143 0.166
8 118.9 132.5 101.5 0.114 0.146
9 89.6 71.3 109.8 0.204 0.225