With the rapid development of renewable energy, the deployment of solar inverters has grown exponentially. As the proportion of renewable energy installations increases, higher demands are placed on grid stability. Many aging solar inverters fail to meet the latest grid requirements in terms of grid-connected performance, including maximum power point tracking (MPPT), automatic generation control (AGC) response rates, primary frequency response, and control accuracy. In this study, I focus on improving the internal response speed of solar inverters, optimizing power distribution execution, and reasonably adjusting MPPT step sizes. By refining key algorithms, I aim to enhance the response speed and power execution accuracy of older solar inverters, address their grid adaptability issues, and increase the power generation of aging photovoltaic (PV) plants.
The grid-connected performance requirements for renewable energy plants, such as wind farms and PV power stations, involve using appropriate active power control systems or standalone control devices to achieve active power-frequency droop control. This enables these plants to participate in rapid grid frequency adjustment at the connection point. The fast frequency response is characterized by an active power-frequency droop特性, implemented through a predefined frequency-active power piecewise linear function. The relationship can be expressed as:
$$ P = P_0 – \frac{P_N}{R} \cdot \frac{f – f_N}{f_N} $$
where \( f_d \) represents the frequency response dead band, typically ranging from 49.94 Hz to 50.06 Hz for PV power stations. \( f_N \) is the system rated frequency, \( P_N \) is the rated power, \( P_0 \) is the initial active power value, and \( R \) is the frequency response droop rate, set at 3% for PV power stations. The active power-frequency droop characteristic for PV power stations is illustrated schematically below, showing how active power adjusts based on frequency deviations.
Key technical requirements for fast frequency response include: first, the regulation target change must be no less than 10% of the rated output under frequency step disturbances; second, the frequency control deviation should be within ±1% of the rated output; third, the frequency measurement resolution must not exceed 0.003 Hz; fourth, the frequency sampling period should be no more than 100 ms; and fifth, the active power control cycle for renewable energy fast frequency response must not exceed 1 second. A schematic of the step disturbance regulation for renewable energy fast frequency response is provided, demonstrating the dynamic adjustment process.
To address the grid-connected performance issues of aging solar inverters, I propose several enhancement schemes. One major problem is the prolonged internal response delay in conventional solar inverters. Typically, these solar inverters use communication modules to receive external commands, which are then forwarded via serial ports to power execution modules. This reception and transmission process introduces a delay of approximately 2 seconds. Upon receiving commands, the power execution unit often employs MPPT power ramp algorithms to track and control power, requiring an additional 1 to 3 seconds to reach the target power. This cumulative delay hinders the ability of solar inverters to meet the stringent 2-second response time required for fast frequency response.
The solution to reducing internal response delays in solar inverters involves upgrading the communication and processing modules to cut down the reception and transmission latency to around 1 second. While the power execution module’s response time generally meets the technical requirement of not exceeding 5 seconds, no further modifications are needed in this aspect. By optimizing the delay time, solar inverters can achieve faster response, as depicted in the improved timing diagram.
Another critical aspect is the optimization of active power control in solar inverters. PV generation units are commonly modeled as controlled current sources in international electromechanical transient simulations. In this model, the active and reactive power of solar inverters are decoupled for control. The reactive power control section can receive dynamic reactive power control commands from the PV plant’s reactive power control system, calculating the reactive reference current \( I_{qcmd} \) based on the reactive control strategy. The implementation involves variables such as \( Q_{gen} \) (measured reactive power), \( V_{max} \) and \( V_{min} \) (maximum and minimum terminal voltages of the PV unit), and the reactive power reference value \( Q_{cmd} \), which is dynamically assigned by the PV plant’s reactive power control system according to the operating state and reactive regulation margin of each solar inverter. Additionally, the thresholds for the reactive reference current \( I_{qcmd} \), denoted as \( I_{q}^{max} \) and \( I_{q}^{min} \), are output by the converter current limiting calculation module, with \( K_{Qi} \) and \( K_{Vi} \) serving as reactive control parameters.
For MPPT in aging solar inverters, early engineering applications often utilized the hill-climbing method due to its simplicity and convenience. However, with advancements in PV technology and the enhanced performance of digital signal processing (DSP), more sophisticated techniques have become feasible for engineering use. The variable-step perturbation and observation MPPT algorithm introduces a variable-step coefficient \( N \). As the system approaches the maximum power point, the derivative of output power with respect to PV array voltage \( \frac{dP}{dU} \) decreases. To minimize voltage oscillations near the maximum power point—thereby improving the stability and energy efficiency of the grid-connected PV system—the voltage reference value in the variable-step perturbation and observation algorithm can be calculated as follows:
$$ U_{ref_{i+1}} = U_{ref_i} + N \left( \frac{dP}{dU} \right) $$
where \( U_{ref_{i+1}} \) and \( U_{ref_i} \) represent the PV array bus voltage references after the \( i+1 \)-th and \( i \)-th perturbations, respectively. The perturbation step size is proportional to the power derivative with respect to voltage, and the step coefficient \( N \) determines the performance of the MPPT algorithm, primarily influencing the system’s rapidity and precision. A diagram of the MPPT algorithm illustrates this process, highlighting the optimization steps for solar inverters.
To validate the proposed schemes, I conducted active power control tests on solar inverters. The tests involved both load reduction and load increase scenarios to evaluate response times, regulation times, steady-state control deviations, and overshoot. For instance, in load reduction tests, the solar inverter’s active power was initially at 1400.2 kW, and the system commanded a reduction to 1200 kW. The solar inverter achieved a maximum active power of 1198.7 kW, completing the first stage. Similar steps were followed for further reductions to 1000 kW, 800 kW, 600 kW, and 400 kW, with the solar inverter consistently meeting targets. In load increase tests, starting from 384.5 kW, commands were given to raise power to 600 kW, 800 kW, 1000 kW, 1200 kW, and 1400 kW, with the solar inverter achieving the desired levels each time. The results, summarized in the table below, confirm that the response time, regulation time, maximum steady-state control deviation, and maximum overshoot all comply with requirements, demonstrating the effectiveness of the enhancements for solar inverters.
| Test No. | Initial Load (kW) | Target Load (kW) | Final Load (kW) | Response Time (s) | Regulation Time (s) | Max Steady-State Control Deviation (%) | Max Overshoot (kW) |
|---|---|---|---|---|---|---|---|
| 1 | 1400.2 | 1200 | 1198.7 | 0.74 | 0.89 | 0.85 | 17.2 |
| 2 | 1193.1 | 1000 | 986.9 | 0.88 | 0.89 | 0.82 | 21.3 |
| 3 | 995.8 | 800 | 785.6 | 0.85 | 0.95 | 0.89 | 39.6 |
| 4 | 783.5 | 600 | 586.7 | 0.86 | 1.67 | 0.93 | 47.4 |
| 5 | 585.7 | 400 | 387.6 | 0.80 | 1.53 | 0.95 | 55.3 |
| 6 | 384.5 | 600 | 584.8 | 0.75 | 1.54 | 0.92 | 26.7 |
| 7 | 586.4 | 800 | 787.4 | 0.74 | 1.53 | 0.88 | 20.4 |
| 8 | 782.2 | 1000 | 990.3 | 0.79 | 1.17 | 0.92 | 17.5 |
| 9 | 984.2 | 1200 | 1193.5 | 0.69 | 0.91 | 0.77 | 14.5 |
| 10 | 1187.5 | 1400 | 1397.6 | 0.98 | 1.21 | 0.43 | 8.5 |
Additionally, by upgrading the software programs and algorithms of solar inverters and optimizing the MPPT strategies, I observed an increase in power generation of approximately 2% to 5% per solar inverter unit. This improvement underscores the potential of algorithm refinements in boosting the efficiency of aging solar inverters.
In conclusion, this study addresses the performance gaps in aging solar inverters within older PV plants, focusing on the impact of regulatory requirements and the inherent limitations of these solar inverters. By identifying key factors that fail to meet grid-connected standards and prioritizing optimizations in response speed and MPPT, I have developed effective solutions to enhance performance. The implementation, centered on control algorithms and response times, has successfully improved the grid adaptability of aging solar inverters and increased the overall power generation of PV plants. This approach demonstrates the importance of continuous innovation in solar inverter technology to support grid stability and renewable energy integration.