In the context of solving-problem, the Case-Based Reasoning (CBR) methodology is one of the most useful. The retrieval phase relies on a similarity measure allowing to identify the most similar cases in the case base with respect to the new target case. Each case stored in the case base is compared to the target case by calculating the distance between these two cases to determine their degree of similarity. All cases stored in the case base are represented by a set of observed features, which allows for solving the problem and Fault Identification (FI). Generally, the features are characterized by their heterogeneous nature, numerical and nominal. The nature of the features is therefore taken into account when choosing the distance calculation method. However, the major problem is that the distance calculation is not accurate when performed in the feature representation space, i.e., separately from the decision to be made. To classify the target case and identify the fault, it is necessary to place the features and the distance calculation in the context of the decision to obtain accurate and precise distance calculation results. This paper proposes a possibilistic distance calculation approach where the representations of the features are transformed into a homogeneous feature space (common space) of the decision. This is done by calculating the possibility distribution of the fault with respect to each feature. This approach improves the precision of the classification of the target case and the identification. The proposed approach transforms the heterogeneous features into a homogeneous space and integrates the distance calculation into the context of the decision. This aims to overcome the limitations of decision-making when the features are of a heterogeneous nature. The expected results include better precision in case classification and more reliable fault identification.