Design Optimisation/Simulation TRIZ Cases
- Neurosymbolic Gas Turbine Component Design Under Constraints
Hybrid neural and rule-based design explores compliant gas turbine cooling structures.
- Underground Cave Rock Stress Mapping With Numerical Inverse Analysis
Microseismometer and stress-gauge data correct point predictions into volumetric stress maps for underground cave monitoring.
- Subsurface Rock Property Models Retrieved by Seismic Stratigraphy
Match rock property models to new seismic data using stratigraphic classification.
- Complex-Valued Neural Networks for Compact Wind Speed Prediction
Group Lasso regularization produces a compact network for accurate wind speed prediction.
- Low-Rank Regression for Faster Physical Parameter Optimization
Historical iteration data builds a low-rank approximation that speeds physical parameter updates without stepwise inverse solutions.
- Bit Response Superposition Eye Simulation Without Noise Overcounting
This case adds characterized noise once after edge-response superposition, producing accurate eye diagrams with less computation.
- Laser-Bonded Composite Strength Prediction Through Layered Modeling
Separate metal, non-metal, and layer models simulate tensile behavior to predict laser-bond strength more accurately.
- Thermodynamic chip architecture of a hybrid thermodynamic-classical computing system
This case uses superconducting flux elements and Langevin dynamics to offload statistical sampling from classical computing devices.
- Debris Flow Abrasion Testing Assembly for Removable Concrete Blocks
A low-speed debris flow abrasion tester uses segmented block mounting to improve wear testing and link indoor data to field assessments.
- Coating Color Matching with Radiative-Transfer-Constrained ML
Radiative-transfer constraints guide ML from spectrophotometer curves to accurate coating component mixtures.
- Robotic Weld Path Simulation for Collision-Prone Seams
Evaluate seam waypoints, adjust constraints, and preview feasible weld paths before a robot encounters collisions or inaccessible regions.
- Analog Mixed-Signal Circuit Layout Synthesis with Machine Learning
This case uses deep learning, templates, feedback, and rule checks to reduce manual effort and layout iteration cycles.
- Substrate Treatment Recipe Validation for Uniform Coverage
Simulation checks treatment time across substrate unit regions before processing, helping prevent untreated areas and improve uniformity.
- Blockchain-Triggered Simulation Model Recalibration
A blockchain stores simulation data and triggers recalibration when thresholds are exceeded, keeping models current without manual updates.