Development of an AI Construction Timeline Forecasting System
Construction delays are the norm, not exception: 70-80% of construction projects exceed planned timelines. An AI system forecasts final delivery dates accounting for current progress, weather, supplies, and historical delay patterns, enabling preventive measures.
Data Sources
BIM model (Building Information Modeling):
- Planned timelines for each WBS element (Work Breakdown Structure)
- Task interdependencies
- Resource allocation: crews, equipment, materials
Operational data:
- Construction site control: % completion per task (weekly/daily)
- Material log: supply deliveries, shortages
- Labor reports: number of workers on site
- PIMS (Project Information Management System): Primavera P6, MS Project
IoT and technical data:
- Site cameras + CV analysis: automatic progress measurement
- Equipment sensors: operating hours, productivity
- GPS tracking: personnel and equipment movement
External factors:
- Weather forecast: days unsuitable for concrete work (< +5°C), high work (wind > 10 m/s)
- Holidays and quarantines
- Supply logistics: key material order status
Forecast Model
Earned Value Analysis (EVA) + ML:
EVA—project management standard:
# Earned Value metrics
SPI = EV / PV # Schedule Performance Index (< 1 = behind schedule)
CPI = EV / AC # Cost Performance Index
# Traditional forecast (EAC):
EAC_schedule = BAC_duration / SPI # if current pace continues
# Problem: SPI doesn't account for work type, weather, dependencies
ML improvement:
features = {
'current_spi': earned_value / planned_value,
'spi_trend_4w': spi_now - spi_4weeks_ago,
'critical_path_float': total_float_critical_path,
'weather_bad_days_upcoming': forecast_bad_days_next_30,
'material_delivery_risk': pending_critical_deliveries_score,
'labor_availability': actual_workers / planned_workers,
'subcontractor_delay_history': mean_delay_by_subcontractor,
'site_area': construction_area_sqm,
'project_complexity': wbs_depth * subcontractor_count,
'season': month # winter affects pace
}
delay_prediction = lgbm_model.predict(features)
# delay_prediction = expected delay in working days
Delay Risk Detector
Critical Path Monitoring: Delays on critical path = project delay:
def critical_path_risk(project_schedule, current_progress, forecast):
critical_tasks = project_schedule.get_critical_path()
risks = []
for task in critical_tasks:
delay_risk = estimate_task_delay(task, current_progress, forecast)
if delay_risk.probability > 0.3:
risks.append({
'task': task,
'expected_delay_days': delay_risk.expected_days,
'probability': delay_risk.probability,
'impact': task.successor_chain_length
})
return sorted(risks, key=lambda x: x['impact'] * x['probability'], reverse=True)
Automatic warnings:
- 3 consecutive weeks of SPI < 0.9 → delay risk > 30 days
- Critical material supplier didn't confirm delivery 14 days before date
- Weather forecast: 5+ days of bad conditions in a row at critical phase
Computer Vision for Progress Monitoring
Automatic progress measurement from site cameras:
- 360° panoramic cameras (Theta, Insta360)—daily photos
- YOLOv8: construction object detection (walls, floors, roofing)
- BIM comparison: % completion by structural elements
3D scanning integration:
- LiDAR scan (Leica BLK360, Faro Focus) → point cloud
- BIM comparison: color-coded visualization of lag
- As-built vs. as-designed: automatic discrepancy detection
PMIS Integration
- Primavera P6: API for reading/writing activities and progress
- Autodesk BIM 360: Cloud API for BIM data
- MS Project Server: REST API
- Russian systems: 1C:Construction, ISUP
System metrics:
- Completion forecast accuracy: MAPE < 10% 60 days before delivery
- Early warning: flag delays 3+ weeks before actual slippage
- Coverage: % of projects under active monitoring
Timeline: basic EVA system with ML delay forecast—5-6 weeks. Full system with BIM integration, CV monitoring and automatic alerts—4-5 months.