<![CDATA[De-dieselization of isolated diesel power plants is not only a generation replacement problem but also an early-stage portfolio screening problem, because candidate locations differ in operating scale, generation cost, fuel efficiency, asset condition, reserve adequacy, operating intensity, and fuel logistics pressure. Existing prioritization approaches may identify ranked alternatives, but they often provide limited explanation of location heterogeneity and limited evidence on shortlist stability under expert-weight uncertainty. This study develops a cluster-informed robust prioritization model for early-stage de-dieselization screening by integrating PCA-K-Means, AHP-TOPSIS, and Monte Carlo robustness analysis. The model was applied to 190 isolated diesel power plant locations. PCA-K-Means identified five technical-economic typologies, while AHP-TOPSIS produced the baseline ranking using six prioritization criteria. Monte Carlo simulation evaluated ranking stability under ±20% AHP weight perturbation across 5,000 iterations. Fuel logistics cost, reserve margin, and generation cost became the most influential criteria. The global ranking was stable, with a median Spearman correlation of 0.997 and a P5-P95 range of 0.978-0.999. Eight locations formed a robust Top-10 core, eight formed a robust Top-20 extension, and several candidates were classified as weight-sensitive. The model helps planners distinguish immediate priority candidates, extended screening candidates, and locations requiring additional verification before detailed feasibility studies.]]>
