<![CDATA[Penelitian ini dikembangkan untuk menginvestigasi model optimasi adaptif Time–Cost Trade-Off (TCTO) yang mengintegrasikan Genetic Algorithm (GA) untuk merepresentasikan non-linearitas dan ketidakpastian hubungan waktu–biaya pada proyek jalan. Metode kuantitatif diterapkan pada enam proyek jalan nyata di Indonesia dengan 1.842 catatan aktivitas; data durasi, biaya, sumber daya, dan variabel eksternal dikumpulkan, diverifikasi, dinormalisasi, dan dianalisis dengan model matematis non-linier serta simulasi GA yang dijalankan menggunakan MATLAB/Python/R. Hasil menunjukkan bahwa model non-linier TCTO–GA memuat kecocokan data yang jauh lebih baik (median R² = 0,89 vs 0,62 untuk model linier), menurunkan error prediksi (MAPE 6,2% vs 13,7%; RMSE 0,081 vs 0,174; paired t-test p < 0,01), serta menghasilkan pengurangan durasi rata-rata 15,4% (±2,1) dengan kenaikan biaya percepatan hanya +2,6% (±0,9), dibandingkan crashing linier (durasi −7,8%; biaya +5,9%). Model mempertahankan stabilitas antar replikasi (SD fitness = 0,9%) dan ketahanan pada skenario gangguan (+20% hari hujan, +15% delay), serta validasi menunjukkan MAPE pasca-optimasi 7,1% (4/6 proyek ±5%) dan penilaian pakar median 4,2. Penelitian ini menunjukkan bahwa integrasi TCTO–GA secara simultan menutup celah representasi non-linearitas dan ketidakpastian, sehingga memberikan kontribusi teoretis pada optimasi stokastik dan alat bantu praktis bagi manajer proyek untuk perencanaan yang lebih adaptif dan tahan gangguan. This study was developed to investigate an adaptive Time–Cost Trade-Off (TCTO) optimization model that integrates a Genetic Algorithm (GA) to represent non-linearity and uncertainty in explicit time–cost relationships for road construction projects. A quantitative approach was applied to six real road projects in Indonesia, comprising 1,842 activity records. Duration, cost, resource, and external variability data were collected, verified, normalized, and analyzed via nonlinear mathematical modeling and GA simulations implemented in MATLAB/Python/R. Results showed that the nonlinear TCTO–GA achieved substantially better data fit (median R² = 0.89 vs 0.62 for linear models), reduced predictive error (MAPE 6.2% vs 13.7%; RMSE 0.081 vs 0.174; paired t-test p < 0.01), and produced an average duration reduction of 15.4% (±2.1) with only a +2.6% (±0.9) acceleration cost—relative to crashing linear approaches which yielded −7.8% duration and +5.9% cost. The model demonstrated replication stability (SD fitness = 0.9%), resilience under disturbance scenarios (+20% rainy days, +15% material delay), post-optimization MAPE = 7.1% (4/6 projects within ±5% of realized outcomes), and a median expert usability score of 4.2. The study indicates that TCTO–GA fills the methodological gap by jointly addressing non-linearity and uncertainty, contributing to optimization theory and providing a practical, adaptive decision-support tool for more resilient road project planning.]]>
Copyrights © 2025
