Constraint Programming (CP) is a popular paradigm to deal with hard combinatorial problems in Artificial Intelligence (AI). One of the key successes of CP is propagation. Propagation helps to remove values that do not lead to a solution. The main objective of this research is to develop propagation algorithms particularly for two constraints: the table constraint and the regular constraint. These constraints are two of the most popular ones in CP. Table constraints are defined by listing explicitly the allowed (or disallowed) combinations of values for their involved variables. Theoretically, they can encode any constraint. Three algorithms concerning table constraints have been developed. Firstly, we propose a GAC propagator for table constraints. This algorithm could be considered as an improvement of STR2. Secondly, we introduce a compression algorithm which converts (ordinary) table constraints into smart table constraints. Smart table constraints allow basic arithmetic constraints to be included in their table and can represent table constraints compactly. Generally, a more compact form of table can lead to a more robust propagation algorithm. Thirdly, we show how to reify a table constraint. In addition, we introduce how to reify a subset of constraints. This reification eases the modeling step. A regular constraint is defined on a sequence of variables and forces that the sequence of values taken by these variables is recognized by a finite automaton. A soft-regular constraint allows violation of this sequence of values with a penalty. Existing violation measures for soft-regular constraint are not adapted for time-oriented problems. Thus, we propose a novel violation measure to solve such problems and introduce a soft version of regular constraints based on this measure. (FSA - Sciences de l'ingénieur) -- UCL, 2019