Secondary Battery Life Diagnosis via Electrolyte Diffusion Estimation

The patent changes the diagnostic parameters from conventional discharge capacity and internal resistance to electrolyte diffusion coefficient and its difference from threshold value (ΔD). This parameter transformation enables accurate remaining life evaluation while maintaining diagnostic simplicity, as the trained model automatically processes input data (voltage drops, temperature rises, discharge capacity) to output these predictive parameters without requiring complex simulation operations during actual diagnosis.

The patent performs preliminary action by training the diagnostic model in advance using simulation data and experimental data. The trained model stores the relationship between battery characteristics and remaining life in its parameters. During actual diagnosis, only simple measurements are needed, and the model automatically outputs accurate predictions without requiring users to perform complex simulations or have specialized knowledge, thus resolving the contradiction between accuracy and complexity.

The patent creates a copied version of the complex simulation process by training an artificial neural network model to replicate the simulation results. Instead of performing actual simulations during diagnosis, the pre-trained model copies the simulation outcomes based on input measurements. This copying approach maintains high accuracy while eliminating the need for users to operate complex simulation software, thus resolving the contradiction between accuracy and ease of operation.

The patent substitutes the mechanical simulation system with an artificial neural network model. The trained model replaces the need for complex simulation calculations during actual diagnosis, transforming a computationally intensive process into a simple data input-output operation. This substitution maintains estimation accuracy while dramatically improving ease of operation and reducing time consumption.