According to the lack of method for highly efficiently spawning coefficients for multi-parent recombination in real-encoded genetic algorithm, an efficient genetic algorithm based on adaptive boundary constraint (ABC) is proposed. This method quickly generates coefficient vectors by adaptively scaling the boundary of the subsequent coefficient based on the value of the previous one, allowing for efficient reconstitution under any number of parent recombination scenarios. Experiment results on CEC2017 benchmarks demenstrate that proposed algorithm outperforms EDBF (empirical distribution based framework) a lot in 29 optimization problems.