Design Optimisation/Simulation TRIZ Cases

A GPU-based physics simulator executes linear system equations using shader programs to solve mesh motion directly on the graphics processor.

A simulation system develops surrogate models to select significant input parameters from initial sets.

A workflow simulation method uses provenance data similarity and sequence alignment to map intermediate states between anchor points in execution traces.

A computational process selects neural network design options using precomputed gradient functions and loss values to guide architecture search.

Information processing device modifies sensor design parameters to optimize feature quantities for machine learning models.

A broker constructs an interface specification to synchronize design and test models through atomic transactions.

An automated system estimates mechanical earth model parameters using ultrasonic image logs and borehole geometry analysis.

A modified Gauss-Seidel preconditioner omits higher-order terms to enable parallel solution of large system matrices.

A virtual reality system uses smart objects and touch-sensitive displays to simulate realistic shopping environments for consumer research.

Finite element analysis of a double fracture model predicts rock fracture initiation under varying horizontal stress differences.

A thermal coupling algorithm discretizes heat flow across non-conformal mesh interfaces using overlapping areas to enhance simulation accuracy.

A combinatorial process evaluation framework integrates simulation models with parallel statistical analyses to generate global prediction models.

A response-surface mobility model constructed from ensemble Monte Carlo data accelerates drift-diffusion simulations.

Pre-calculating stress distribution eliminates labor-intensive FEM re-meshing, enabling accurate crack growth prediction in high-temperature equipment.

An information loss determination engine simulates content experience for disabled users to compute quantified data loss metrics.

A physics-informed multimodal autoencoder encodes heterogeneous data into a shared latent space using Gaussian mixture distributions.

A risk assessment system computes attack risk by propagating threats across heterogeneous network nodes.

A 1D CFD model adds a force term to the gas momentum equation to predict Taylor bubble velocity.

An intermediary layer standardizes heterogeneous runtime data via aspect models, reducing system complexity while improving adaptability.

A continuous wave lidar simulation method generates ray sets with emission timing to compute signal contributions and mix them into output signals.

Calculates failure rates for individual wire segments using thermal maps and reliability data to identify disproportionate contributors.

A two-dimensional porous seepage microscopic model with a camera component enables direct visual observation of fluid flow behavior.

Local maximum limits and spatial change constraints optimize mesh quality while reducing computing time.

Positioning identifiers decouple property data from mesh granularity, eliminating iterative recreation and accelerating aircraft design workflows.

Electric heating elements in well tubing heat biomass slurry, eliminating complex heat exchangers and reducing operational costs.

Evaluates arbitrary hardware and software state expressions at discrete time points to construct system memory views.

A lattice Boltzmann model derives weight coefficients using a univariate polynomial equation to ensure numerical stability.

Segmented minimization avoids saddle points and reduces computational burden by switching between conjugate gradient and quasi-Newton methods during relaxation.

Simulation defines fatigue zones in segmented die casting inserts, enabling material optimization that doubles lifespan while reducing production costs.

Fourth-order symplectic discretization reduces numerical dispersion errors in anisotropic magnetized plasma simulations.

An estimation device generates emphasis pattern images to improve far field accuracy.

Geomechanical modeling tool predicts rock failure using drillstring vibrational models and surface drilling parameters.

A simulation model compiler transforms acausal equations into causal machine instructions for efficient execution.

Koopman operator linearization replaces complex inverse model calculations with neural network evaluation to improve micro-displacement positioning precision.

Continuous conditional generative adversarial network generates realistic time series data conditioned on continuous inputs.

Segment state system operations into independent output and update types to eliminate extra storage elements required by conventional hardware translation.

Graph neural network predicts semiconductor mesh changes to reduce computational time while maintaining simulation accuracy.

A graph neural network processes mesh nodes via message passing to predict fluid velocity and pressure fields.

A rectilinear-block placement method uses a machine learning model to predict sub-block positions on a chip canvas.

Tunable standard inverters reduce clock skew and short-circuit power in hybrid tree-mesh networks without increasing design complexity.