Convert computationally intensive engineering simulations into data-driven and/or physics-informed ML/AI surrogates that enable:
- Design Exploration & Optimization: Create accurate surrogate ML models that allow you to explore multiple designs and/or optimize designs.
- Product Validation: Create physics-informed ML models that can extrapolate computationally intensive simulation models. Integrate into these models, test and field data to have a unified predictive model.
- System Simulation & Controls: Integrate ML models with advanced control frameworks to create smart and autonomous machines that optimize performance and prevent maintenance events.

Thermomechanical Fatigue
Hyper-Fast Surrogate Model
Problem: Convert a multi-step, multi-physics CAE model of a heavy-duty diesel engine exhaust manifold to a hyper-fast PIML solver for rapid thermomechanical life (TMF) assessment of new designs, and to offset the high cost of traditional simulation.
Solution: PredictiveIQ incorporated a multi-stage physics-based model in a PIML framework to develop a CAD designer-oriented software that predicts TMF life.
Results: We have demonstrated a record 95% reduction in virtual product development cost.

Model Validation
Novel Algorithm for Test & Analysis Integration
Problem: Create a statistical model validation scheme to remove the subjectivity of telematics data alignment to maintain consistency of results across different models that are being validated against experiments.
Solution: Through a Fourier-based multivariate hypothesis testing, PredictiveIQ provides capabilities for validating models with experimental or field data.
Results: PredictiveIQ demonstrates statistically significant anomaly detection in nonstationary time series data.