<span lang="EN">This study investigates the effectiveness of Correlation Power Analysis (CPA) using the Hamming Weight model to extract AES encryption keys in a fully software-simulated environment. By leveraging Python programming, we emulate power traces not from hardware devices but through Hamming Weight calculations derived from byte-level operations during AES encryption. Simulated plaintexts are randomly generated, and key hypotheses are evaluated using Pearson correlation between expected bit-switching activity and simulated traces. The method achieved approximately 50% accuracy with just 10 plaintexts and up to 85% accuracy when using over 1,000 simulated inputs. Correlation coefficients above 0.90 were consistently observed for most key bytes. While the simulation avoids the complexity of real-world noise and hardware interference, it also lacks authentic electrical characteristics. This highlights both the novelty and the limitation of a software-only CPA framework. The findings underline the vulnerability of AES to side-channel attacks and suggest countermeasures like masking to reduce risk.</span>