Abstract: In view of the unclear main controlling factors of tight oil reservoir productivity and the large difference of single well production,the dynamic time warping algorithm and LightGBM algorithm are used to establish the main controlling factors analysis and production prediction method of tight oil productivity.This method can comprehensively take into account geological,engineering and development factors,rapidly identify the main controlling factors of tight oil well productivity and the production change mode,and then predict oil well productivity,which possesses strong universality.The research results are applied to the tight oil block in Qian′an Area.The results show that there are four kinds of oil well production change modes in this block,which are slow decline type,sharp decline type,fluctuation decline type and stable production type.The main controlling factors are the effective length of horizontal section,the thickness of producing oil layer,the control reserve multiplied by permeability divided by viscosity,oil protential,the density of fracturing section,fracture density,fracturing fluid intensity,sand to liquid ratio,sand strength,dynamic liquid interface and shut-in time.By using this method to predict the productivity of tight oil wells,the average accuracy is over 90%,and the boundaries of geological and engineering main control factors required for the cumulative oil production of single wells within 4 a to reach 0.5×10⁴ t and 1.0×10⁴ t respectively can be calculated.The research results have certain guiding significance for the optimization of tight oil development plan.

Key words: tight oil, main controlling factors of production capacity, similarity measure, dynamic time warping, decision tree, Qian′an Area