Estimation × EV Systems
Adaptive Battery Digital Twin for SOC Estimation & Fault Diagnosis
A 4S1P lithium-ion digital twin that estimates state of charge and internal resistance while separating sensor faults from weak-cell behavior.
0.0222 V
Healthy residual RMS
100 s
Sensor fault detected
176 s
Weak-cell isolated
4S1P
Pack
Problem
Battery-management decisions depend on hidden states such as SOC and internal resistance, while sensor failures and cell degradation can produce superficially similar voltage behavior.
Architecture
The model separates the physical-cell and fault-injection layer, an adaptive digital-twin estimator, expected-voltage generation, residual calculation and a diagnostic layer that classifies healthy, voltage-sensor-fault and weak-cell cases.
Diagnosis
The diagnostic logic combines voltage residuals, estimated internal resistance and thermal response. This gives the system multiple signals for fault isolation instead of relying on a single threshold.
Result
The healthy voltage residual remained low, while the injected voltage-sensor fault was detected from 100 s and the weak-cell case was isolated at 176 s as resistance and temperature behavior diverged.
Evidence