The integration of renewable energy sources presents a significant challenge to grid stability due to their inherent intermittency and variability. To mitigate this, the “renewable energy plus storage” paradigm has become essential. Among various storage technologies, battery energy storage systems (BESS) are widely deployed due to their stable voltage output, reliability, and technological maturity. For large-scale applications, cascaded H-bridge (CHB) multilevel inverters offer a superior topology, enabling direct medium-voltage connection and enhancing system modularity and scalability. However, ensuring the long-term safety, reliability, and performance of these massive energy storage battery packs remains a critical operational challenge.
Traditional fault detection methods for energy storage battery units often rely on offline testing or simple online monitoring of voltage and current thresholds. These approaches frequently fail to provide early warning for energy storage battery degradation or incipient faults, often identifying problematic units only after a complete failure has occurred. This reactive strategy jeopardizes system stability and increases maintenance costs. Therefore, developing an online, proactive condition monitoring technique that can assess the health of individual energy storage battery units within a cascaded system during normal operation is paramount.
This article presents a comprehensive online intermittent discharge detection scheme integrated into the control strategy of a cascaded H-bridge energy storage battery grid-connected inverter. The proposed methodology builds upon the Hybrid Pulse Power Characterization (HPPC) test, adapting it for continuous, non-invasive operation within a working energy storage power station. By leveraging a modified two-stage power conversion topology and a dedicated high-low frequency hybrid modulation control strategy, the scheme enables periodic pulsed discharge testing of selected energy storage battery packs. This process yields dynamic and static characteristic data—such as open-circuit voltage (OCV) and internal resistance as functions of depth-of-discharge (DOD)—without interrupting the overall power transfer to the grid. The acquired data provides a robust foundation for state-of-health (SOH) analysis, facilitating early fault detection, preventive maintenance, and enhanced operational safety for the entire cascaded energy storage battery system.
System Architecture and Operational Principles
The foundation of the proposed online detection system is a two-stage cascaded H-bridge inverter topology, as shown in the conceptual diagram below. This structure is key to enabling independent control and testing of each energy storage battery unit.
The topology comprises ‘n’ identical power modules connected in series on their AC output sides. Each module is responsible for one energy storage battery pack and consists of three key stages:
- Battery Pack: The primary energy storage battery unit, typically composed of series-parallel connected cells to achieve the required voltage and capacity.
- DC/DC Converter: A bidirectional converter that interfaces the energy storage battery with the DC-link. Its primary functions are:
- Regulating the charge/discharge current of the energy storage battery during normal operation and testing.
- Providing galvanic isolation if required.
- Disconnecting the energy storage battery from the DC-link during its designated rest period in the test sequence, allowing the DC-link capacitor to temporarily sustain the H-bridge output.
- H-Bridge Inverter: A single-phase full-bridge converter that generates the AC voltage waveform. The AC outputs of all ‘n’ H-bridges are connected in series to synthesize a high-quality multilevel voltage waveform for grid connection.
The total AC output voltage of the inverter is the sum of the individual H-bridge output voltages:
$$ v_{AB} = \sum_{i=1}^{n} v_{H_i} $$
where \( v_{H_i} \) is the output voltage of the i-th H-bridge module. An LCL or L filter is typically used at the grid connection point to attenuate switching harmonics.
The salient feature of this topology is the independent DC/DC stage for each energy storage battery. This independence is the enabler for the online detection scheme, as it allows the control system to manipulate the power flow from a single energy storage battery pack without affecting the others. During a test cycle, the DC/DC converter of the unit under test (UUT) can be commanded to deliver a precise pulsed current profile, while the converters of other units maintain normal operation.
| Feature | Benefit for Online Detection |
|---|---|
| Independent Battery Control | Enables isolated pulsed discharge/rest cycles for any single energy storage battery pack. |
| DC-Link Voltage Decoupling | The DC-link capacitor buffers power, allowing the H-bridge to operate normally even when its connected energy storage battery is in a rest phase. |
| Modularity and Scalability | Facilitates easy integration of the detection scheme for any number of cascaded modules. |
| Fault Isolation | A faulty energy storage battery pack can be disconnected by its DC/DC converter, preventing cascading failures. |
Intermittent Discharge Detection Methodology
The core of the condition monitoring strategy is the online intermittent discharge test, an adaptation of the standardized HPPC test. The objective is to characterize the dynamic and static parameters of a energy storage battery at various states of charge during normal system operation.
Test Principle and Sequence
The test involves subjecting the selected energy storage battery unit to a series of controlled discharge pulses, each followed by a prolonged rest period. A complete test sequence spans a significant portion of the battery’s discharge curve. During normal system operation, one energy storage battery pack at a time is designated as the UUT and undergoes this sequence.
The typical profile for a single test point consists of:
1. Rest Period (Pre-pulse): The energy storage battery is disconnected from the DC-link by its DC/DC converter. Its terminal voltage stabilizes towards its true open-circuit voltage.
2. Discharge Pulse: The DC/DC converter is controlled to inject a constant current, \( I_{dis} \), from the energy storage battery into the DC-link for a short duration, \( t_{pulse} \) (e.g., 10-30 seconds). This simulates a high-power demand event.
3. Rest Period (Post-pulse): The energy storage battery is again disconnected, allowing its internal chemistry to relax. The terminal voltage recovery is monitored.
This pulse-rest cycle is repeated at predetermined intervals of depth-of-discharge (DOD). The entire process for one UUT takes place over many minutes or hours, while the overall inverter system continues to feed power to the grid using the other energy storage battery modules.
Data Analysis and Parameter Extraction
The voltage and current waveforms captured during the pulse and rest periods provide the data needed for health assessment.
1. Open-Circuit Voltage (OCV) vs. DOD:
The voltage at the end of the extended rest period, when the current is zero and the battery polarization has diminished, is taken as the OCV. By plotting OCV against the corresponding DOD for multiple test points, a characteristic OCV-DOD curve is obtained. This curve is a fundamental fingerprint of the energy storage battery chemistry and degrades with aging. Deviations from a baseline or reference curve indicate changes in the energy storage battery state. The OCV at a specific DOD, \( V_{OC}(DOD) \), is a critical parameter for state-of-charge (SOC) estimation algorithms.
2. Internal Resistance Calculation:
The dynamic response during the current pulse reveals the internal resistance. Using the instantaneous voltage step at the beginning and end of the discharge pulse, the discharge resistance \( R_{dis} \) can be calculated. A similar method applies if a charge pulse is used. The figure below illustrates the calculation points.
$$ R_{discharge} = \frac{V_{t_2} – V_{t_3}}{I_{t_2} – I_{t_3}} $$
where:
- \( V_{t_2}, I_{t_2} \): Voltage and current just before the end of the discharge pulse.
- \( V_{t_3}, I_{t_3} \): Voltage and current just after the discharge pulse ends (start of recovery).
The internal resistance is a direct indicator of energy storage battery health. An increasing trend in \( R_{dis} \) over time signifies aging, increased losses, and reduced power capability. Monitoring this parameter allows for the proactive replacement of degrading units before they cause performance issues or fail.
| Parameter | Symbol | Extraction Method | Significance for Health Monitoring |
|---|---|---|---|
| Open-Circuit Voltage | \( V_{OC}(DOD) \) | Voltage after long rest period | Indicates electrochemical state, SOC estimation, detects major degradation. |
| Polarization Resistance | \( R_{pol} \) | Analysis of voltage recovery time constant | Reflects kinetic limitations and diffusion processes within the energy storage battery. |
| Ohmic/Discharge Resistance | \( R_{dis} \) | Instantaneous voltage change vs. current change | Measures conductive losses, cell connection integrity, and overall power capability. Primary aging indicator. |
| Pulse Power Capability | \( P_{pulse} \) | \( V_{pulse} \times I_{dis} \) | Direct measure of the energy storage battery‘s ability to deliver high power. |
High-Low Frequency Hybrid Modulation and Control Strategy
Implementing the online test requires a sophisticated modulation strategy for the cascaded H-bridge inverter. The goal is to seamlessly manage power flow when the UUT transitions between supplying power (during its pulse) and not supplying power (during its rest). This is achieved through a High-Low Frequency Hybrid Modulation scheme with a modified sorting algorithm.
Operating Modes of H-Bridge Modules
In this strategy, each H-bridge module can operate in one of two fundamental modes:
- Low-Frequency (LF) Module (Square-Wave Operation): These modules output fundamental frequency square waves with levels of +1, 0, or -1 (corresponding to \( +V_{dc}, 0, -V_{dc} \)). They are the primary workhorses, synthesizing the bulk of the output voltage staircase. A module connected to a energy storage battery that is actively discharging operates in this mode.
- High-Frequency (HF) Module (PWM Operation): One module at a time operates in this mode. It outputs a high-frequency PWM waveform to finely regulate the total output voltage and compensate for harmonics from the LF modules. A module whose energy storage battery is in the rest phase (disconnected by its DC/DC) operates as the HF module. Its DC-link capacitor provides the necessary energy buffer.
The total modulation reference voltage \( v_{AB}^* \) is synthesized as follows:
$$ v_{AB}^* = v_{LF}^* + v_{HF}^* $$
where \( v_{LF}^* \) is the sum of the outputs from the (n-1) LF modules, and \( v_{HF}^* \) is the PWM output from the single HF module.
Control System Structure
A dual-loop control system manages the overall operation:
1. Main Controller (Voltage and Current Control):
This generates the total reference voltage \( v_{AB}^* \). It typically uses an outer voltage loop to regulate the sum of all DC-link voltages and an inner current loop to control the grid current. A Proportional-Resonant (PR) controller is often used in the current loop for accurate sinusoidal tracking.
$$ v_{AB}^* = G_{PR}(s) \cdot (i_{grid}^* – i_{grid}) + v_{grid} $$
where \( G_{PR}(s) \) is the PR controller transfer function, and \( v_{grid} \) is a feedforward term for grid voltage.
2. High-Frequency Module Controller:
A dedicated PI controller regulates the DC-link voltage of the HF module. The output of this PI controller, multiplied by a unit sinusoidal signal in phase with the grid, generates the HF module’s modulation reference \( v_{HF}^* \).
$$ v_{HF}^* = \left( K_{p} + \frac{K_{i}}{s} \right) \cdot (V_{dc,ref}^{HF} – V_{dc}^{HF}) \cdot \sin(\omega t) $$
3. Modulation Wave Synthesis and Sorting:
The reference for the LF modules is obtained by subtracting the HF reference from the total reference:
$$ v_{LF}^* = v_{AB}^* – v_{HF}^* $$
A sorting algorithm then decides which specific LF modules output +1, -1, or 0 to best approximate \( v_{LF}^* \) while also balancing their individual DC-link voltages. The module designated as the UUT is given priority in the sorting algorithm to ensure it operates as an LF module (outputting +1 or -1) during its required discharge pulse.
When the UUT enters its rest phase, its operational mode is switched. Its DC/DC converter isolates the energy storage battery, and its H-bridge is designated as the new HF module. It now uses PWM to regulate the output using energy from its DC-link capacitor. Another module’s energy storage battery is then selected to begin its discharge pulse and becomes an LF module. This handover process is managed by the system’s central supervision controller.
| Test Phase | DC/DC Converter State | H-Bridge Mode | Power Source | Purpose |
|---|---|---|---|---|
| Discharge Pulse | Active: Controls constant discharge current \( I_{dis} \) | Low-Frequency (LF) | Energy storage battery pack | Delivers power to grid & provides pulse load for testing. |
| Rest Period | Inactive: Disconnects battery, DC-link floats | High-Frequency (HF) | DC-link capacitor | Maintains grid connection; allows battery voltage to stabilize for OCV measurement. |
Balancing and Stability Considerations
The control strategy inherently incorporates voltage balancing. The sorting algorithm balances voltages among the LF modules. The dedicated HF module controller balances the HF module’s DC-link voltage with the average voltage of the LF modules through the power flow adjustment described by the equation for \( v_{LF}^* \). If the HF module’s DC-link voltage drops (because its capacitor is supplying power), its controller increases \( v_{HF}^* \), which in turn reduces \( v_{LF}^* \), shifting power demand back to the LF modules and allowing the HF capacitor to recharge. This closed-loop interaction ensures stable operation across all modules, whether they are sourcing power from a energy storage battery or a capacitor.
System Performance and Simulation Analysis
To validate the proposed topology and control strategy, a detailed simulation model of a system with four cascaded H-bridge modules was developed. The key objective was to demonstrate that the intermittent discharge test could be performed on one energy storage battery unit without disrupting the quality of the grid-injected power.
Simulation Parameters and Setup
| Parameter | Value |
|---|---|
| Number of H-bridge Modules (n) | 4 |
| Nominal Battery Voltage per Module | 72 V |
| Grid Voltage (Phase) | 311 V (RMS) |
| Grid Frequency | 50 Hz |
| Filter Inductance (L1, L2) | 2 mH |
| DC-link Capacitance per Module | 2200 µF |
| Discharge Pulse Current (\( I_{dis} \)) | ~10 A (module dependent) |
| Pulse Duration / Rest Period | 10 s / 50 s (example ratio) |
Simulation Results and Discussion
The simulation results clearly illustrate the successful implementation of the online detection scheme.
1. Battery Current Profile:
The current of the UUT shows the distinct intermittent discharge profile. It features a series of constant-current discharge pulses, each lasting for several seconds, followed by extended periods of zero current (rest). This waveform confirms that the DC/DC converter is successfully executing the test sequence, drawing specific power pulses from the energy storage battery for characterization.
$$ I_{bat,UUT}(t) = \begin{cases} I_{dis} & \text{for } kT \leq t < kT + t_{pulse} \\ 0 & \text{for } kT + t_{pulse} \leq t < (k+1)T \end{cases} $$
where \( T \) is the total period of one pulse-rest cycle and \( k \) is the cycle index.
2. Grid Current Quality:
Despite the pulsed power draw from the UUT, the total grid current remains a high-quality, low-THD sinusoidal waveform. The high-low frequency hybrid modulation effectively compensates for the power variations introduced by the test. The current is tightly controlled to be in phase with the grid voltage, demonstrating that the overall system maintains excellent grid-tie performance and power factor. The disturbance from the UUT’s pulse is absorbed by the DC-link capacitors and managed by the control system’s redistribution of power among the other LF modules.
3. Inverter Output Voltage:
The synthesized multilevel voltage \( v_{AB} \) shows the characteristic stepped waveform. The number of voltage levels is \( 2n+1 \), which in this case is 9 levels. The waveform demonstrates that the modulation strategy is correctly implemented, with one module providing the PWM refinement while the others provide the fundamental square-wave steps. The transition of the UUT’s H-bridge between LF and HF modes is smooth and does not introduce significant transients into the output voltage.
4. Data for Health Analysis:
From the simulated voltage and current data of the UUT during a pulse, parameters like the instantaneous voltage drop \( \Delta V \) can be measured. Coupled with the known pulse current \( I_{dis} \), an approximate internal resistance \( R_{dis} = \Delta V / I_{dis} \) can be calculated for that DOD point. The stable voltage during the rest period provides the OCV. By running the simulation over a longer period representing multiple DOD points, a dataset mimicking a real HPPC test is generated, proving the feasibility of data collection for offline SOH analytics.
Conclusion and Future Perspectives
This article has presented a holistic and technically feasible solution for the online condition monitoring of cascaded battery energy storage systems. The proposed scheme integrates an intermittent discharge detection methodology directly into the power conversion and control layers of a two-stage cascaded H-bridge inverter. By leveraging a dedicated DC/DC converter per energy storage battery pack and a high-low frequency hybrid modulation strategy, the system can perform periodic characterization tests on individual energy storage battery units during normal grid-feeding operation.
The key achievements of this approach are:
- Non-Invasive Testing: The energy storage battery health assessment is conducted without taking the system offline, maximizing availability.
- Proactive Fault Detection: By tracking parameters like internal resistance and OCV-DOD characteristics over time, the system can identify degrading energy storage battery packs early, enabling predictive maintenance.
- Maintained Grid Performance: The control strategy ensures that the power quality and stability of the grid connection are uncompromised during the test cycles.
- Enhanced System Safety and Reliability: Early detection of weak or faulty units prevents catastrophic failures, improves overall system efficiency, and extends the operational life of the energy storage battery asset.
Future work can focus on several advanced areas:
- Intelligent Data Analytics and SOH Estimation: Integrating machine learning algorithms to automatically analyze the collected pulse data, estimate precise SOH and remaining useful life (RUL) for each energy storage battery pack, and generate actionable alerts.
- Adaptive Test Scheduling: Developing algorithms to optimize the timing and intensity of test pulses based on grid demands, energy storage battery usage history, and prior health indicators, minimizing any conceivable impact on system efficiency.
- Extended Topology Applications: Adapting the core principles of this online detection scheme to other multilevel topologies used in medium-voltage BESS, such as modular multilevel converters (MMC) with integrated energy storage battery cells.
In conclusion, the integration of online intermittent discharge detection represents a significant step towards smarter, safer, and more reliable large-scale energy storage battery power stations, forming a critical enabler for a resilient and renewable-dominant power grid.