Generalized arc consistency(GAC) is the most widely studied consistency for solving constraint satisfaction problem(CSP). MDDc, STR2 and STR3 are widely used algorithms for enforcing GAC on table constraints, where MDDc represents constraints in a compression method which converts a table constraint into a multi-valued diagram(MDD). When a MDD has many identical parts, the compression ratio is high and MDDc outperforms others. STR3, similar to STR2, dynamically reduces the table size by maintaining a table of only valid tuples. When valid tuples decrease slowly, STR3 outperforms STR2. Considering that finding valid outgoing edges of MDD nodes and checking and deleting invalid tuples in dual table are the most time-consuming operations in MDDc and STR3, this study proposes AdaptiveMDDc and AdaptiveSTR respectively for MDDc and STR3 to perform the above two operations in an adaptive way so that the lowest cost strategy in different backtrack search stages is always employed. Benefiting from lower cost of strategy and significant performance difference of each strategy during different backtrack search stages, AdaptiveMDDc and AdaptiveSTR have better performance compared to the original methods, and ApaptiveSTR is over three times faster than STR3 in some instances.