What is the difference between Battery State of Charge (SoC) and State of Health (SoH)?

1) **State of Charge (SoC, 0% to 100%):** Represents the instantaneous available fuel gauge of the battery relative to its current maximum usable capacity; 2) **State of Health (SoH, 100% down to 70% or 0%):** Measures the permanent, irreversible chemical degradation of the battery compared to its brand-new nameplate capacity: SoH = fracC_currentC_nominal × 100%. In utility BESS assets, an SoH dropping below 70% to 80% marks the financial End-of-Life (EoL) threshold, requiring battery module replacement or augmentation.

Why is simple Coulomb Counting (current integration) inaccurate for LiFePO4 batteries?

Lithium Iron Phosphate (LiFePO4) exhibits an ultra-flat open-circuit voltage (OCV) discharge curve (remaining between 3.25V and 3.32V across 20% to 80% SoC). Pure Coulomb Counting integrates current sensor readings (SoC = SoC_0 - (int I dt / C_n)). Current sensor noise, thermal drift, and integration drift accumulate errors of **10% to 25% over time**, causing the BMS to mistakenly report wrong capacity.

How does the Extended Kalman Filter (EKF) achieve sub-1% state estimation accuracy?

The Extended Kalman Filter (EKF) is a recursive statistical estimation algorithm that fuses real-time current (I), terminal voltage (V), and temperature (T) measurements with a non-linear 2RC Equivalent Circuit Model of the battery. By dynamically balancing process uncertainty covariance (Q) against measurement noise covariance (R), the EKF continuously corrects state predictions in real time, achieving **sub-1.0% SoC precision and sub-2.0% SoH tracking accuracy**.

How does Recursive Least Squares (RLS) track internal resistance increase (R_int)?

As battery cells age, growth of the Solid Electrolyte Interphase (SEI) layer and active material cracking increase internal ohmic resistance (R_0) and polarization resistance (R_1, R_2). The RLS algorithm tracks micro-second voltage steps during current pulses, calculating the exact rise in internal resistance in real time as a secondary verification of cell health.

What is BESS Battery Augmentation and why is SoH modeling critical for project finance?

Instead of buying an oversized 150MWh battery on Day 1 to guarantee 100MWh delivered over 20 years, developers build a 100MWh system and perform scheduled **Battery Augmentations** (adding 10MWh of new cells at Year 5, Year 10, and Year 15). Accurate SoH Kalman filter telemetry allows asset managers to time battery additions precisely when needed, slashing upfront project Capex by **20% to 30%**.

BMS Algorithms & State Estimation

Battery State-of-Health (SoH) & Kalman Filter Guide 2026

By PSI Editorial · 18 min read · Updated September 2026

BMS battery management system controller running Extended Kalman Filter EKF algorithms to calculate real time State of Health SoH of lithium energy storage cells — PakSolarInsights

⚡ Battery SoH Estimation Fast Facts (2026)

Sub-1% Accuracy: Extended Kalman Filters (EKF) achieve < 1.0% state estimation precision.
2RC Modeling: Captures ohmic resistance + charge transfer + diffusion dynamics.
Degradation Tracking: Recursive Least Squares (RLS) continuously tracks internal resistance growth.
Augmentation Optimization: Accurately times multi-million dollar BESS capacity expansions.

Atomic Summary: In multi-megawatt battery energy storage systems (BESS), knowing the exact remaining capacity and health of millions of electrochemical cells is critical for safety, dispatch revenues, and warranty claims. Because battery degradation cannot be measured directly with physical sensors, modern Battery Management Systems (BMS) deploy Extended Kalman Filters (EKF) and Recursive Least Squares (RLS) algorithms to estimate State of Health (SoH) in real time. Master 2RC equivalent circuit modeling, covariance filtering, and capacity augmentation planning.

Complete Comparison: Battery State Estimation Algorithms

Estimation Algorithm	Operating Principle	SoC Estimation Error	Computational Overhead	LFP Flat Curve Tolerance
Coulomb Counting (Ah-Integration)	Direct mathematical integration of current (int I dt)	> 10% (Accumulates drift)	Minimal (Simple addition)	POOR (Severe drift on flat curves)
Extended Kalman Filter (EKF)	Non-linear Taylor expansion + 2RC state estimation	< 1.0% Error	Moderate (Standard 32-bit MCU)	EXCELLENT (Self-correcting)
Unscented Kalman Filter (UKF)	Deterministic sigma-point statistical sampling	< 0.8% Error	Higher (Matrix inversions)	MAXIMUM (Best accuracy)

Frequently Asked Questions

What causes 'Knee-Point' acceleration in battery degradation?

During initial cycling (100% down to 80% SoH), degradation is linear as the SEI layer thickens slowly. Once active lithium inventory drops past a critical threshold, lithium plating begins on the anode, triggering a sharp non-linear 'Knee-Point' where degradation accelerates 5x faster toward rapid capacity loss.

How does a BMS perform OCV calibration during rest periods?

When the battery sits idle with zero current for >30 minutes (allowing electrochemical diffusion overpotentials to relax), the BMS measures stable terminal voltage and references the calibrated OCV-SoC lookup table, resetting any accumulated Kalman filter drift back to exact 0.0% error.