Energy storage type emergency backup power supply voltage stabilizing method and system with voltage stabilizing and frequency stabilizing functions

Abstract

The application discloses a kind of energy storage type emergency backup power supply voltage stabilizing method and system with voltage regulating and frequency stabilizing function, belong to power electronics technical field. Including: constructing virtual synchronous generator frequency regulation model, based on frequency variation rate and frequency deviation adaptive adjustment virtual moment of inertia coefficient and damping coefficient, output phase information;Establish the discrete-time state space model of three-phase converter and LC filter, adopt two-step prediction strategy to compensate digital control delay;With state estimation value as initial state, traverse candidate voltage vector, select optimal voltage vector through target function containing voltage tracking deviation and control increment;The optimal voltage vector is converted into pulse width modulation signal, and the output voltage is synthesized by driving power switch tube. The application realizes high-precision, fast dynamic adjustment of energy storage type emergency backup power supply output voltage and frequency, significantly improves power quality and system stability under complex conditions such as load mutation and off-grid switching.

Description

Technical Field

[0001] This application belongs to the field of power electronics technology, specifically relating to a voltage regulation method and system for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization functions. Background Technology

[0002] As modern power systems evolve towards intelligence and distributed systems, energy storage-based emergency backup power supplies, as core components ensuring the power supply security of critical infrastructure, are playing an increasingly prominent role in responding to grid anomalies, ensuring power quality, and achieving peak shaving and valley filling. These systems utilize the collaborative operation of energy storage battery packs and power electronic conversion devices to quickly switch to independent power supply mode in the event of external power failures, providing continuous support for power quality-sensitive fields such as healthcare, data centers, and precision manufacturing. Especially in microgrid architectures, energy storage systems not only need to perform energy buffering functions but also act as the main control unit to maintain bus stability in off-grid operation.

[0003] Voltage regulation and frequency stabilization technology is a core control component of energy storage-type emergency backup power supplies. It aims to ensure that the system's output voltage amplitude and frequency remain within allowable ranges during periods of severe load fluctuation or operational condition switching. This technology typically employs a dual closed-loop control strategy of voltage and current to adjust the pulse width modulation signal of the bidirectional power converter in real time, thereby achieving refined compensation of the output waveform. Its basic principle lies in constructing a responsive feedback loop to counteract the negative interference of internal resistance voltage drop and load changes on the power output characteristics, thus maintaining the quality of power supply.

[0004] However, existing technologies still have significant shortcomings in handling dynamic voltage stabilization tasks under complex power grid conditions. Traditional control models, when faced with the instantaneous connection of large-capacity loads or switching between grid connection and disconnection modes, suffer from large overshoot and long recovery times due to the difficulty in dynamically optimizing control parameters, making it difficult to achieve millisecond-level transient response. Simultaneously, existing frequency stabilization algorithms often ignore the nonlinear disturbances generated by the connection of multiple heterogeneous loads, resulting in limited frequency tracking accuracy and making the system highly susceptible to oscillations in weak power grid environments. Therefore, a voltage stabilization solution for energy storage-type emergency backup power supplies with voltage regulation and frequency stabilization functions is desired to improve its dynamic response speed and power quality stability under complex conditions such as load changes and grid connection / disconnection switching. Summary of the Invention

[0005] The purpose of this invention is to provide a voltage regulation method and system for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization functions, which can effectively solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A voltage stabilization method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization functions includes the following specific steps: The virtual mechanical power command and the measured value of the actual active power output by the system are obtained as virtual electromagnetic power to construct a virtual synchronous generator frequency regulation model; the frequency regulation model includes the following virtual rotor motion equations: ; in, For virtual mechanical power, For virtual electromagnetic power, The virtual moment of inertia coefficient, For virtual angular velocity, This is the derivative of the virtual angular velocity with respect to time, i.e., the rate of change of frequency. The damping coefficient is... Rated angular velocity; The virtual angular velocity is obtained by solving the motion equations of the virtual rotor. And based on the frequency change rate Adaptive adjustment of the virtual rotational inertia coefficient and based on frequency deviation Dynamically adjust the damping coefficient For the virtual angular velocity The integrated output is used for voltage regulation phase information; A discrete-time state-space model including a three-phase converter and an LC filter is established. In each control cycle, a voltage reference trajectory is synthesized based on the phase information. Based on the discrete-time state-space model, a two-step prediction strategy is adopted to compensate for the computational delay of the digital control system. The two-step prediction strategy includes applying the voltage vector selected in the previous control cycle to perform a one-step forward prediction of the system state in the current control cycle to obtain the state estimate at the start of the next control cycle, and using the state estimate as the initial state for prediction. Using the state estimate as the initial state, the voltage prediction trajectory corresponding to all candidate voltage vectors is calculated in the prediction time domain. Each candidate voltage vector is evaluated using an objective function that includes voltage tracking deviation and control increment. The voltage vector that minimizes the objective function value is selected as the optimal voltage vector. The optimal voltage vector is converted into a pulse width modulation signal to drive the power switching transistors of the three-phase converter to synthesize the output voltage waveform.

[0007] Furthermore, the virtual rotational inertia coefficient is adaptively adjusted based on the frequency change rate, including: when the absolute value of the frequency change rate is detected to be greater than a first preset change rate threshold, the virtual rotational inertia coefficient is gradually increased by a first step size; when the absolute value of the frequency change rate falls back to below a second preset change rate threshold, the virtual rotational inertia coefficient is gradually decreased by a second step size.

[0008] Furthermore, the dynamic adjustment of the damping coefficient based on the frequency deviation includes: when the absolute value of the angular velocity deviation exceeds a preset deviation threshold, the damping coefficient is increased linearly according to the magnitude of the angular velocity deviation or according to a piecewise function curve.

[0009] Furthermore, it also includes: during the transient process of switching from grid-connected mode to off-grid mode or from off-grid mode to grid-connected mode, increasing the damping coefficient to a preset maximum damping value, and restoring the damping coefficient to the dynamic adjustment value after the transient process ends.

[0010] Furthermore, the length of the prediction time domain To meet Integers; the objective function Represented as: ; in, This is the weighting factor for voltage tracking deviation. To control the weighting coefficient of the increment, For the first Voltage reference trajectory at time t, For the corresponding voltage prediction trajectory, For the first The change in voltage vector at time t.

[0011] Furthermore, it also includes: using an extended Kalman filter algorithm to estimate the state of charge of the energy storage battery in real time, and dynamically calculating the power output boundary and constructing the output envelope based on the state of charge, the junction temperature of the power converter, the saturation current limit of the filter inductor, and the DC bus voltage support capability; when the external power adjustment command exceeds the boundary of the output envelope, clamping the actual output power command within the boundary of the output envelope; the virtual mechanical power is limited by the actual output power command.

[0012] Furthermore, the step of converting the optimal voltage vector into a pulse width modulation signal includes: using a space vector pulse width modulation algorithm to calculate the conduction time of each bridge arm switching device, and allowing only one of the three bridge arms to operate during any switching state transition, in order to suppress common-mode voltage fluctuations.

[0013] Furthermore, it also includes: real-time monitoring of the polarity direction of the output current of each phase during each pulse width modulation cycle, and adjusting the on-time of the upper transistor and the on-time of the lower transistor of that phase according to the monitored current polarity direction, so as to compensate for the distortion of the output voltage waveform caused by the dead time.

[0014] A voltage regulation and frequency stabilization system for an energy storage-type emergency backup power supply includes: The frequency regulation model construction module is used to construct a virtual synchronous generator frequency regulation model. The frequency regulation model includes a virtual rotor motion equation. The virtual angular velocity is obtained by solving the virtual rotor motion equation. The virtual moment of inertia coefficient in the virtual rotor motion equation is adaptively adjusted based on the frequency change rate. The damping coefficient in the virtual rotor motion equation is dynamically adjusted based on the frequency deviation to output phase information for voltage regulation. The model predictive control module is used to establish a discrete-time state-space model including a three-phase converter and an LC filter. In each control cycle, it synthesizes a voltage reference trajectory based on phase information. Based on the discrete-time state-space model, it adopts a two-step prediction strategy to compensate for the computational delay of the digital control system. The two-step prediction strategy includes applying the voltage vector selected in the previous control cycle to perform a one-step forward prediction of the system state in the current control cycle, obtaining the state estimate at the start of the next control cycle, and using the state estimate as the initial state for prediction. The model predictive control module is also used to calculate the voltage prediction trajectory corresponding to all candidate voltage vectors in the prediction time domain with the state estimate as the initial state, and to evaluate each candidate voltage vector using an objective function that includes voltage tracking deviation and control increment, and select the voltage vector that minimizes the objective function value as the optimal voltage vector. The pulse width modulation module is used to convert the optimal voltage vector into a pulse width modulation signal, which is used to drive the power switching transistors of the three-phase converter to synthesize the output voltage waveform.

[0015] In summary, this application includes at least one of the following beneficial technical effects: 1. This invention achieves refined regulation of the output characteristics of emergency backup power supplies by introducing virtual synchronous generator technology and model predictive control algorithms. Compared with traditional dual closed-loop voltage and current control, this invention significantly reduces the voltage transient drop depth when facing sudden load surges and greatly shortens the time for voltage recovery to steady state. Frequency tracking accuracy is greatly improved, effectively solving the system oscillation problem in weak grid environments and significantly enhancing the stability of power supply.

[0016] 2. This invention can automatically optimize the control logic based on the remaining energy. Under complex operating conditions such as high load fluctuations or nonlinear load connections, the quasi-proportional resonant control and adaptive inertial compensation of this invention can work together to significantly reduce the total harmonic distortion rate of the output voltage, ensuring the safe operation of precision manufacturing equipment and sensitive electronic loads.

[0017] 3. This invention achieves a balance between internal energy management and external grid regulation demands through decoupling control strategies and power sharing mechanisms. While ensuring emergency power supply quality, the system can improve the utilization rate of energy storage batteries through intelligent peak shaving and valley filling, maintaining a high overall energy conversion efficiency. Simultaneously, due to its comprehensive load shedding and health status protection functions, the system's maintenance-free operation cycle is extended, reducing operation and maintenance costs.

[0018] 4. This invention reduces the transient response time during grid-to-offline switching to an extremely short time, achieving truly seamless switching. This responsiveness is crucial for critical infrastructure such as medical operating rooms and large data centers, effectively avoiding the risk of equipment restarts or data loss due to transient fluctuations during power switching. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall technical solution for a voltage stabilization method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization functions; Figure 2 This is a schematic diagram illustrating the core principle of coordinated regulation based on virtual synchronous generators and model predictive control. Figure 3 It is a logical flowchart for assessing the state of charge of energy storage batteries and constructing the dynamic output envelope; Figure 4 It is a schematic diagram of the multi-level interaction relationship and data flow between multi-channel sampling, state-space model prediction and pulse width modulation command output; Figure 5 This is a schematic diagram comparing the system's frequency fluctuation suppression characteristics with existing technologies. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the following description is provided in conjunction with the appendix. Figures 1-5 The present invention will be further described in detail below with reference to specific embodiments.

[0021] The voltage regulation method and system for energy storage emergency backup power supply with voltage regulation and frequency stabilization functions provided in this application embodiment can be applied to the control system of energy storage emergency backup power supply. The control system includes a multi-channel high-frequency sampling unit, a core controller, an energy storage battery pack, a bidirectional power converter, an LC filter unit, and a grid connection point switchgear.

[0022] The core controller can use industrial-grade power electronic control chips such as DSP and FPGA, and can be integrated into the energy storage converter (PCS). It can seamlessly switch between grid-connected and off-grid modes, and its application scenarios cover critical infrastructure sensitive to power quality, such as medical operating rooms, large data centers, and precision manufacturing production lines, as well as microgrid distributed energy storage systems.

[0023] In other possible implementations, this method can also be applied to various energy storage power systems with emergency backup functions, such as residential energy storage and industrial and commercial energy storage.

[0024] In a first aspect, in an exemplary embodiment, a voltage regulation method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization function is provided. Taking the application of this method to the aforementioned core controller as an example, the method includes the following steps: S100, through real-time perception of the internal core components of the energy storage emergency backup power supply and the physical quantities of the external power grid connection point, completes signal acquisition and preprocessing, providing accurate, low-noise and highly synchronized raw data input for subsequent control algorithms.

[0025] S200 establishes a precise nonlinear mapping between the internal physical state of the battery and its external electrical output capability, and evaluates the state of charge of the energy storage battery and the current allowable power output boundary in real time and accurately, providing reasonable and feasible constraints for the generation of subsequent power commands.

[0026] S300: Construct a frequency regulation model based on virtual synchronous generator technology, enabling the power electronic converter to simulate the rotor kinematics of a traditional synchronous generator, providing inertial and damping support for the system, and realizing active regulation and stable control of the grid frequency.

[0027] The S400 employs a model predictive control strategy, which solves for the optimal voltage vector sequence in the discrete time domain to achieve fast and accurate tracking of the inverter output voltage, effectively suppressing voltage fluctuations caused by load changes.

[0028] The S500 converts the optimal voltage vector command into the turn-on and turn-off time signals of the three-phase inverter bridge arm power switches, drives the switches to operate through the hardware pulse width modulation module, synthesizes the desired sinusoidal voltage waveform, and completes the final voltage regulation and frequency stabilization execution.

[0029] In an exemplary embodiment, based on the above-described S100, the detailed execution process of signal acquisition and preprocessing is explained, and the above-described S100 includes: S101. Simultaneously acquire analog signals through a multi-channel high-frequency sampling unit. In this embodiment, a multi-channel high-frequency sampling unit is set as the physical front end for data acquisition. The sampling unit integrates a high-precision synchronous sampling analog-to-digital converter. The number of channels can be selected as 8 or 16 channels, and the quantization bit width is set to 16 bits or 24 bits. This hardware configuration can keep the quantization error generated during the sampling process below 0.01%.

[0030] In practice, the system sets the sampling frequency within a preset range of 10kHz to 50kHz and transmits the acquired raw sampling data stream to the controller for processing in real time and without interruption through a high-speed serial peripheral interface.

[0031] For different electrical physical quantities inside the energy storage system and at the grid connection interface, the range and accuracy settings of this sampling unit are as follows: For the three-phase AC voltage on the output side of the energy storage emergency backup power supply, the sampling range is set to 0 to 1000V to be compatible with the standard operating condition with an effective AC value of 380V and some medium and high voltage application scenarios.

[0032] For three-phase AC current, the sampling range is designed to match the rated power of the system, covering a range of 0 to 2000A, ensuring that the current waveform can be captured completely and without distortion even under brief overload impacts.

[0033] For the DC bus voltage, the sampling range is set to 0 to 1200V to accurately monitor the DC side energy buffer status between the battery pack and the inverter.

[0034] For the terminal voltage of the energy storage battery pack itself, the sampling accuracy is required to reach the millivolt level in order to sensitively identify minute potential fluctuations caused by changes in the internal chemical reaction process; for the charging and discharging current of the battery pack, the signal is first converted by a Hall sensor or a high-precision shunt, and then sent to a multi-channel high-frequency sampling unit to complete the analog-to-digital conversion.

[0035] S102. Perform hardware anti-aliasing filtering on the acquired raw signals. In this embodiment, before the analog signal enters the analog-to-digital conversion stage, in order to ensure the authenticity and effectiveness of subsequent digital domain signal analysis, all sampled signals must pass through a hardware anti-aliasing low-pass filter.

[0036] The cutoff frequency of this filter is set to a preset noise frequency threshold, typically 0.4 to 0.5 times the current sampling frequency. This hardware-level preprocessing effectively filters out high-frequency harmonic components introduced by the switching action of power semiconductor devices, as well as electromagnetic interference noise in the space, preventing spectral aliasing from the source.

[0037] S103. Perform a composite digital filtering algorithm on the digitized signal to enhance the signal-to-noise ratio. In this embodiment, the digital signal quantized by the analog-to-digital converter first enters a circular data buffer opened in memory; at the software execution level, the system then starts a composite digital filtering process. The first step of the process is to use a median filtering algorithm to remove pulse-type spike noise caused by occasional external environmental events such as lightning strikes or contactor contact switching.

[0038] Following closely behind is the moving average filtering algorithm, which is used to smooth the signal waveform. The window width used in the moving average operation can be dynamically adjusted between 16 and 64 sampling points according to the time constant of the control loop.

[0039] In particular, the aforementioned composite filtering algorithm can work synergistically to address the glitches and interference signals that current sensors are easily coupled to when high-power power electronic devices are switching on and off nearby, significantly improving the signal-to-noise ratio and laying a solid foundation for accurate mathematical calculations.

[0040] S104. Perform Clarke transformation on the filtered three-phase voltage and current signals. In this embodiment, after completing the aforementioned filtering preprocessing steps, the obtained three-phase voltage and three-phase current signals will enter the coordinate transformation stage. The system uses Clarke transformation to convert these electrical quantities from the three-phase stationary coordinate system to the two-phase stationary coordinate system.

[0041] During the transformation calculation process, depending on the specific requirements of the subsequent control algorithm, the system can choose to follow the principle of constant amplitude or constant power, and generate voltage vector components and current vector components accordingly.

[0042] In this embodiment, S1 constitutes a complete and rigorous signal acquisition and conditioning link. Through this series of operations, the original electrical signal, which originally contained noise and was in a three-phase coupled state, is transformed into a low-noise, high-fidelity two-phase quadrature vector signal. These vector components are not only the direct input of the subsequent vector control strategy, but will also be used to calculate the instantaneous active power and reactive power, thereby providing accurate numerical basis for the energy flow direction judgment of the entire system.

[0043] Through high-precision, multi-channel synchronous sampling hardware design, coupled with a conditioning mechanism that combines hardware anti-aliasing and software composite digital filtering, and finally coordinate transformation processing, the system successfully captures the key electrical states of the energy storage system in real time and accurately, providing an indispensable perception basis for the subsequent decision-making and execution of the entire control system.

[0044] In an exemplary embodiment, based on the above-described S200, the detailed execution process of energy storage battery state assessment and power boundary determination is explained, wherein S200 includes: S201. Establish a second-order Thevenin equivalent circuit model of the energy storage battery. In this embodiment, the system describes the dynamic behavior of the energy storage battery based on a high-fidelity second-order Thevenin equivalent circuit. This equivalent circuit model consists of the following components connected in series: an ideal voltage source characterizing the battery's open-circuit voltage, a resistive element characterizing the ohmic internal resistance, and two sets of parallel RC networks composed of parallel polarization resistors and polarization capacitors, respectively. The first set of parallel RC networks is used to simulate the short-time constant dynamic characteristics during the electrochemical reaction process inside the battery, and the second set of parallel RC networks is used to simulate the long-time constant dynamic characteristics during the diffusion process.

[0045] Battery terminal voltage The relationship between the internal state variables and the observation equations is described as follows: In this equation, This refers to the directly measurable battery terminal voltage. Let be the nonlinear function of the battery open-circuit voltage with respect to the state of charge (SOC). This represents the real-time current flowing through the battery; a positive value is used during charging, and a negative value is used during discharging. The ohmic internal resistance of the battery. and The polarization voltages established at both ends of the first and second order polarization branches are respectively; the evolution of the polarization voltage is determined by the time constant of each branch, and its dynamic equation can be conventionally derived by those skilled in the art based on the parallel RC network.

[0046] S202. Obtain the time-varying parameters of the equivalent circuit model. In this embodiment, the parameters included in the above-mentioned second-order Thevenin equivalent circuit are the ohmic internal resistance. First-order polarization resistance and polarization capacitance, second-order polarization resistance and polarization capacitance, and open-circuit voltage function. These vary with the state of charge (SOC), battery temperature, charge / discharge current rate, and degree of battery aging.

[0047] The system obtains these time-varying parameters through one of the following two methods: Method 1, online parameter identification: The system adopts a recursive least squares algorithm with a forgetting factor, using the current and several historical current and terminal voltage sample values ​​as input data, and recursively calculates the estimated values ​​of each model parameter at the current moment; the forgetting factor can be set to 0.98, which is used to exponentially weight the historical data to decay, so that the identification results can track the slow changes in battery parameters.

[0048] Approach 2, Multidimensional Lookup Table: The system pre-calibrates a multidimensional lookup table through offline experiments. The dimensions of the lookup table cover SOC, battery temperature, charge / discharge current rate, and aging degree. During system operation, the controller indexes the corresponding model parameter values ​​from the lookup table using a multidimensional interpolation algorithm based on the real-time SOC estimate, temperature sensor measurement, current current rate, and cumulative charge / discharge cycle count.

[0049] The offline calibration experiment procedure is as follows: pulse charge and discharge tests are performed on the battery at different SOC points and different temperatures, the voltage response curve is recorded, and the model parameters under each operating condition are identified using curve fitting tools and stored in non-volatile memory.

[0050] S203. The Extended Kalman Filter (EKF) algorithm is used to estimate the State of Charge (SOC) in real time. In this embodiment, after obtaining the battery model and its current parameters, the system uses the EKF algorithm to estimate the SOC in real time and in a closed loop. The EKF is a state estimation method suitable for nonlinear systems, and its specific application in this application includes the following key configurations: the state vector consists of SOC, the first-order polarization voltage, and the first-order polarization voltage. and the second-order polarization voltage These three elements together constitute the whole.

[0051] The state transition equation is established based on the ampere-hour integral method and the discretized dynamic equation of the polarization voltage. The discretized form of the ampere-hour integral part is as follows: In this formula, The current state of charge. This represents the state of charge at the previous moment. The Coulomb efficiency coefficient ranges from 0.98 to 1.00. The sampling period is in seconds. This refers to the battery's nominal capacity, measured in ampere-seconds. This is the current measurement value from the previous moment. The observation equation is the same as the terminal voltage equation in step S201. Due to the open-circuit voltage function Typically, the function is nonlinear. The extended Kalman filter performs local linearization on this nonlinear function by calculating the Jacobian matrix in each iteration period.

[0052] The filtering algorithm executes a standard prediction-update loop in each sampling period: first, it calculates the state prior estimate and its error covariance based on the state transition equation; then, it calculates the Kalman gain using the actual measured terminal voltage value; finally, it corrects the state prior estimate using the voltage measurement residual to obtain the posterior estimate of the SOC. The specific iterative formulas and matrix operations are conventional methods used by those skilled in the art when implementing extended Kalman filtering, and will not be elaborated here.

[0053] When the system is first started or powered on again after a long period of inactivity, the initial value of the state of charge is determined by the inactivity voltage lookup table method. Specifically, if the battery has been inactivity for more than 2 hours, the system measures the current open-circuit voltage of the battery and looks up the corresponding SOC value according to the pre-stored open-circuit voltage-SOC characteristic curve, which is used as the initial state input of the extended Kalman filter.

[0054] S204. Execute graded power control logic based on real-time SOC. In this embodiment, the system executes a graded power control strategy based on the real-time SOC estimate output in step S203 to coordinate the different behavior modes of the energy storage system under conditions of energy surplus and energy shortage.

[0055] When the SOC is greater than the first preset power threshold, for example, 80%, the system is determined to be in a state of sufficient power. At this time, the control priority is set to frequency downsampling compensation. When the grid frequency is too high due to excess power generation, the system absorbs the extra active energy in the system by increasing the charging power absorbed from the grid or decreasing the discharge power released to the grid.

[0056] When the SOC is less than the second preset power threshold, for example, 20%, the system is determined to be in a power supply edge state. At this time, the control priority is switched to frequency increase support. When the grid frequency is low due to insufficient power generation, the system prioritizes ensuring emergency discharge capability by increasing the discharge power or decreasing the charging power to inject active power into the grid to support frequency recovery.

[0057] When the State of Charge (SOC) is between the first preset power threshold and the second preset power threshold, the system can freely participate in bidirectional power regulation according to the grid demand, without being subject to priority constraints caused by the SOC.

[0058] S205. Dynamically calculate the power output boundary and construct the output envelope. In this embodiment, the dynamic calculation of the power output boundary not only depends on the SOC, but also must take into account the physical limits of the power converter itself; the system monitors the following key constraints in real time: the junction temperature of the power semiconductor devices in the power converter, the saturation current limit of the AC side filter inductor, and the current support capability of the DC bus voltage.

[0059] The system constructs an output envelope based on the current operating point, which specifies the maximum allowable charging and discharging power at different duration scales. Specifically, the system sets the short-time overload multiple to 1.5 times the rated power, with a corresponding predetermined duration of 30 seconds.

[0060] When the demand value of the external power regulation command exceeds the boundary of the output envelope, the limiting function in the controller automatically intervenes to clamp the actual output power command within the boundary of the envelope, thereby preventing the energy storage battery from being overcharged or over-discharged or the power converter from being damaged due to overheating or overcurrent.

[0061] In this embodiment, by establishing a second-order Thevenin equivalent circuit model and utilizing the extended Kalman filter algorithm, accurate closed-loop estimation of the battery's internal state of charge—a key variable that cannot be directly measured—is achieved. Simultaneously, by combining the SOC hierarchical control logic and the output envelope calculation under multiple physical constraints, the upper and lower limits of the energy storage system's allowable power output and control priorities at the current moment are clarified. This provides reasonable constraints for the solution space of the power command of the subsequent frequency regulation model and the control quantity of the model's predicted voltage control, forming a closed-loop logical dependency relationship, and jointly ensuring the coordination and safety of the system during the dynamic voltage regulation and frequency stabilization process.

[0062] In an exemplary embodiment, based on the above-described S300, the detailed execution process of frequency regulation based on virtual synchronous generator technology is explained, wherein S300 includes: S301. Establish a virtual rotor motion equation as the core algorithm for frequency regulation. In this embodiment, the system solves a differential equation simulating the rotor motion of a synchronous generator in real time through the controller, namely the virtual rotor motion equation. The mathematical expression of this equation is: In this equation, This is virtual mechanical power, and its value corresponds to the system power command issued by the upper-level energy management system. This is the virtual electromagnetic power, whose value corresponds to the measured value of the actual active power output by the system. This is the virtual moment of inertia coefficient, whose physical meaning corresponds to the moment of inertia of a traditional generator rotor, and the unit is kilograms per square meter. This refers to the virtual angular velocity, which is the angular frequency variable calculated and generated internally by the control system. The rated angular velocity corresponds to the angular velocity value converted from the standard power grid frequency of 50Hz, i.e. radians per second; is the damping coefficient, used to simulate the damping winding of a synchronous generator's suppression of power oscillations.

[0063] The power difference on the left side of the equation The virtual angular velocity represents the acceleration or deceleration power acting on the virtual rotor; when the virtual mechanical power is greater than the virtual electromagnetic power, the power difference is positive. An increase; conversely, a decrease in virtual angular velocity; the second term on the right side of the equation. It provides a damping torque proportional to the frequency deviation to suppress continuous frequency oscillations.

[0064] S302. Adaptive adjustment of the virtual moment of inertia coefficient J. In this embodiment, to ensure that the system exhibits a suitable inertial response when subjected to disturbances of varying degrees of intensity at different frequencies, the virtual moment of inertia coefficient J is adjusted. It is not a fixed value, but rather adopts an adaptive adjustment mechanism.

[0065] The system calculates the rate of change of the power grid frequency in real time, i.e., the derivative of frequency with respect to time, through a phase-locked loop or frequency measurement circuit during each control cycle. The unit is Hz per second. When the absolute value of the detected frequency change rate exceeds a first preset change rate threshold, for example, greater than 0.5 Hz per second, it indicates that the power grid is experiencing a severe frequency drop or surge. At this time, the controller gradually increases the value of the virtual rotational inertia coefficient J by a step size to simulate the effect of increased rotor mass of the synchronous generator. The increased J value will slow down the rate of change of the virtual angular velocity ω, thereby suppressing further frequency deterioration. The first preset change rate threshold is greater than the second preset change rate threshold, and the difference between the two constitutes a hysteresis interval, which is used to avoid repeated switching of the virtual rotational inertia coefficient J when the frequency change rate fluctuates around a single threshold.

[0066] When the grid frequency gradually stabilizes and the absolute value of the frequency change rate falls below the second preset change rate threshold, for example, below 0.1 Hz per second, the controller gradually reduces the value of the virtual rotational inertia coefficient J in a second step size, releasing the previously stored virtual kinetic energy so that the system frequency can converge to near the rated value more quickly.

[0067] The values ​​of the first step length and the second step length are set according to the rated power of the system and the desired dynamic response speed: the value of the first step length is 0.1 to 0.5 times the base value of increasing the virtual moment of inertia coefficient J per second, and the value of the second step length is 0.05 to 0.2 times the base value of decreasing J per second.

[0068] The adaptive adjustment range of the virtual moment of inertia coefficient J is limited to a preset minimum value. With the maximum value Between. As a specific example, for an energy storage-type emergency backup power system with a rated power of 100kW, the reference value of the virtual rotational inertia coefficient J is... Set to 0.5 kg / m 2 The minimum value Set to 0.2 times the baseline value, i.e., 0.1 kg / m³. 2 The maximum value Set to 4 times the baseline value, i.e., 2.0 kg / m 2 When the system frequency changes drastically, J gradually increases until it does not exceed 2.0 kg / m. 2 Once the frequency stabilizes, J is gradually reduced to no less than 0.1 kg / m. 2 .

[0069] The range of values ​​for the virtual moment of inertia coefficient J is determined based on the following factors: the system's rated power, the DC bus voltage level, and the desired virtual inertia time constant. Inertial time constant The relationship with the moment of inertia coefficient satisfies ,in This is the system's rated capacity. Typically, the virtual inertial time constant... The value range is from 0.5 seconds to 5 seconds, and the feasible range of J is calculated in reverse.

[0070] S303. Implement variable parameter adjustment of the damping coefficient D based on frequency deviation. In this embodiment, the damping coefficient... The value is also dynamically adjusted, rather than being a constant; the system calculates the deviation between the virtual angular velocity and the rated angular velocity in real time. .

[0071] When frequency deviation When the absolute value is small, the damping coefficient Maintain a low baseline value to reduce active power loss during steady-state operation of the system.

[0072] When frequency deviation When the absolute value exceeds a preset deviation threshold, the damping coefficient... The damping coefficient can be increased linearly or according to a piecewise function curve, depending on the magnitude of the deviation. Increasing the damping coefficient can generate a stronger damping torque, accelerate the decay of frequency oscillations, and prevent the system from oscillating continuously in a weak power grid environment due to underdamping.

[0073] Damping coefficient The adjustment range is also limited to a preset minimum damping value and a maximum damping value to avoid the oscillation being difficult to converge due to insufficient damping or the frequency recovery being too slow due to excessive damping.

[0074] S304. Enhanced damping to suppress oscillations during the transient process of switching between grid-connected and off-grid modes. In this embodiment, during the transient process of the energy storage system switching from grid-connected mode to off-grid mode or from off-grid mode to grid-connected mode, the system automatically adjusts the damping coefficient. It is raised to near its preset maximum damping value; this operation can effectively absorb the oscillation energy caused by voltage phase change and power difference due to mode switching, and suppress reactive power circulation and power overshoot that may occur at the moment of grid connection switch action.

[0075] After the transient process ends and the system enters a new steady-state operation, the damping coefficient Then restore to the normal adjustment value determined by step S303.

[0076] In this embodiment, by introducing a virtual rotor motion equation and implementing adaptive and variable parameter adjustments to the virtual moment of inertia coefficient and damping coefficient, the power electronic converter acquires the inertial response and damping characteristics of a traditional synchronous generator; the virtual angular velocity output by this frequency regulation model... Its phase information provides a rotation angle reference for the reference coordinate system of subsequent model prediction voltage control.

[0077] Meanwhile, virtual mechanical power command Limited by the power output envelope calculated in the aforementioned steps, it ensures that the frequency regulation process will not endanger the safe operation of the energy storage system, forming a closed-loop logical dependency relationship with the preceding and following steps.

[0078] In an exemplary embodiment, based on the above-described S400, the detailed execution process of voltage regulation based on model predictive control is explained, wherein S400 includes: S401. Establish the state-space model of the three-phase converter and its LC filter. In this embodiment, the system first establishes an accurate mathematical model of the three-phase voltage source converter and its output LC filter in the energy storage emergency backup power supply; the state variables of the model include the three-phase current flowing through the filter inductor, the three-phase output voltage applied across the filter capacitor, and the three-phase load current flowing to the downstream load; this state-space model is a set of first-order differential equations in the continuous time domain, describing the relationship between the rate of change of inductor current and the input-output voltage difference, and the relationship between the rate of change of capacitor voltage and the difference between inductor current and load current.

[0079] To implement model predictive control in the controller, the system employs the Euler forward discretization method, with a sampling period of... Using a step size, the continuous-time state-space model is converted into a discrete-time difference equation. The discretized model can predict the evolution trend of the state variables at the next sampling time based on the current state variable values ​​and the candidate converter input voltage vector.

[0080] S402. Set the prediction step size and control time domain. In this embodiment, in order to balance the computational burden and control dynamic performance, the prediction step size, i.e. the prediction time domain length, is set to 2 to 5 sampling periods; this means that in each control period, the controller will recursively predict the system behavior for the next 2 to 5 steps based on the current state.

[0081] A longer prediction step size results in a smoother control effect, but the computational cost increases exponentially. Conversely, a too-short prediction step size may cause the closed-loop response to approach that of traditional deadbeat control, making it overly sensitive to model accuracy. The control time domain is typically set to be the same as or slightly smaller than the prediction time domain.

[0082] S403. Perform multi-step trajectory prediction based on initial state values ​​in each control cycle. In this embodiment, at the beginning of each control cycle, the system uses the current voltage and current sample values ​​provided by the aforementioned signal acquisition steps as the initial state.

[0083] Then, for each of the eight basic voltage vectors defined by the switching states of the six power switching devices of the three-phase converter, including six effective vectors and two zero vectors, it is calculated how the state trajectory of the system will evolve at each time step in the future if the voltage vector is applied.

[0084] The model parameters used in the prediction process are not static; the system integrates an online identification algorithm for the filter inductor parameters to monitor the saturation effect of the filter inductor as the current amplitude increases in real time; the identification algorithm is executed at a preset correction frequency, for example, updating the inductor parameter values ​​in the model every 100Hz, to ensure that the predicted trajectory is consistent with the behavior of the physical entity and to avoid the degradation of control performance due to parameter mismatch.

[0085] S404. Construct the objective function and solve for the optimal voltage vector online. In this embodiment, to select the optimal one from the eight candidate voltage vectors as the output of the current control cycle, the system constructs a comprehensive objective function, the mathematical expression of which is: In this objective function, The smaller the value of the objective function, the better the voltage vector control effect. The first preset weighting factor for voltage tracking deviation, The second preset weighting coefficient is used to control the increment; Indicates in The voltage reference trajectory value at time t is obtained by synthesizing the voltage amplitude command and phase information output from the aforementioned frequency adjustment step; Indicates in The predicted voltage trajectory value at time t; This represents the change in the applied voltage vector between two adjacent control cycles, i.e., the control increment; To predict the total number of steps, the value is between 2 and 5; summation sign. This indicates the sequence from step 1 to step 2. The prediction errors of each step are accumulated.

[0086] The first term of the objective function Penalizing voltage tracking error forces the predicted output voltage to approximate the reference voltage trajectory as closely as possible; Item 2 Abrupt changes in the control input are penalized to suppress excessive fluctuations in the switching frequency and reduce switching losses; this is achieved by adjusting the weighting coefficients. and The relative size allows for a trade-off between fast tracking response and smooth switching action.

[0087] The system calculates the objective function value corresponding to each of the eight basic voltage vectors by iterating through them. Select to make The voltage vector that reaches its minimum value is taken as the optimal control solution for the current control cycle.

[0088] S405. A two-step prediction strategy is adopted to compensate for the calculation delay. In this embodiment, in actual digital control systems, there is an inherent calculation delay between the sampling time and the time of updating the pulse width modulation signal; if no compensation is made, the optimal vector calculated in the previous control cycle will be delayed until the second half of the current cycle before it can be executed, resulting in a decrease in control performance.

[0089] To eliminate the adverse effects of this delay, the controller employs a two-step prediction strategy; specifically, in... At any given time, the controller does not directly calculate the time used for Instead of determining the optimal vector at a given time, the system state is first predicted one step forward using the voltage vector selected in the previous cycle. State estimate at time 1.

[0090] Then with that The state at time step is the new initial state. The prediction and optimization processes described in steps S403 and S404 are then executed to finally determine the state at which the prediction should take place. The optimal voltage vector applied at any given time; the time consumed in this calculation process is compensated by including it in the step size of the model recursion, thereby achieving seamless compensation for computational delay.

[0091] In this embodiment, by establishing a state-space model including an LC filter, setting a reasonable prediction step size, constructing an objective function that balances tracking accuracy and control stability, and combining a two-step prediction delay compensation mechanism, high dynamic quality regulation of the output voltage of the energy storage emergency backup power supply is achieved, replacing the traditional proportional-integral dual closed-loop control. This effectively solves the problems of voltage overshoot and long recovery time during sudden load switching, and the optimal output voltage vector information will be directly transmitted to the subsequent pulse width modulation execution step.

[0092] In an exemplary embodiment, based on the above-described S500, the detailed execution process of pulse width modulation and waveform output is explained, wherein S500 includes: S501. The conduction time of each bridge arm switching device is calculated using a space vector pulse width modulation algorithm. In this embodiment, the system uses a space vector pulse width modulation algorithm to convert the target voltage vector output in the previous steps into the conduction time of each bridge arm switching transistor of the inverter.

[0093] The basic process of the algorithm is as follows: First, based on the sector position of the target voltage vector in the complex plane, determine the two adjacent effective basic voltage vectors and the zero vector used to synthesize the vector; second, based on the volt-second balance principle, calculate the duration of each of the two effective vectors and the zero vector within one pulse width modulation cycle; finally, allocate the duration of each vector to the upper and lower switching transistors of the three-phase bridge arm to generate the corresponding comparison register values.

[0094] The action time allocation formula involved in the above volt-second balance calculation is a conventional method used by those skilled in the art when implementing space vector pulse width modulation, and will not be elaborated on here.

[0095] S502. An optimized switching vector sequence is used to suppress common-mode voltage fluctuations. In this embodiment, based on the determined action time of each vector, the arrangement sequence of the switching vectors is optimized. The traditional 7-segment space vector pulse width modulation starts and ends with a zero vector in each pulse width modulation cycle, and switches the zero vector type at the midpoint of the cycle. This sequence causes common-mode voltage fluctuations in the midpoint potential of the three-phase bridge arms. To suppress common-mode voltage fluctuations, this solution uses an optimized switching vector sequence. The specific principle is that during any switching state switch, only the switching transistor of one of the three-phase bridge arms is allowed to operate, while the switching states of the other two bridge arms remain unchanged.

[0096] This constraint is achieved by mis-configuring the update timing of the comparator register of the pulse width modulation module; the resulting common-mode voltage variation amplitude is limited to less than one-sixth of the DC bus voltage, effectively reducing the electromagnetic interference level of the inverter output.

[0097] S503. Zero-vector configuration strategy is adopted to reduce switching losses under high-power conditions. In this embodiment, in view of the high conversion efficiency requirements of high-power energy storage systems, this solution specifically configures the use strategy of zero vector; the system monitors the electrical angle of the output voltage vector in real time. When the electrical angle falls within a preset 60-degree sector range, the controller selects to keep the specific bridge arm switch corresponding to that sector in a non-operating state for several consecutive pulse width modulation cycles, that is, to continuously clamp the upper or lower switch of that bridge arm in the on or off state, and use another type of zero vector to achieve voltage synthesis.

[0098] This discontinuous pulse width modulation strategy ensures that each switching device does not perform switching action for one-third of the time in each fundamental frequency cycle, thereby reducing the total switching loss by about 30%. When the system is operating under light load conditions, the controller can automatically switch back to the conventional 7-segment continuous pulse width modulation mode to ensure the quality of the output waveform.

[0099] S504. Implement dead-time dynamic compensation based on current polarity monitoring. In this embodiment, during the actual operation of the inverter, in order to prevent the two upper and lower switches of the same bridge arm from being shot-through short-circuited, a dead time must be inserted between complementary switching signals, usually 2 microseconds to 5 microseconds. The introduction of dead time will cause amplitude reduction and waveform distortion between the actual waveform of the output voltage and the ideal command, especially near the zero crossing point of the output voltage.

[0100] This solution employs dead-zone dynamic compensation technology based on current polarity monitoring. The system monitors the polarity direction of the output current of each phase in real time during each pulse width modulation cycle. When the current of a certain phase is positive, the controller automatically increases the conduction time of the upper transistor of that phase by one dead time and decreases the conduction time of the lower transistor by one dead time. When the current is negative, the compensation direction is adjusted in the opposite direction accordingly. If the current amplitude is too small to reliably determine the polarity, the compensation operation is suspended to avoid voltage spikes caused by miscompensation.

[0101] S505. Perform closed-loop verification and threshold control on the output voltage harmonic quality. In this embodiment, the pulse width modulation command after dead-zone compensation is output to the drive circuit of the power switch through the enhanced pulse width modulation module of the controller; the system uses the output voltage data acquired in the aforementioned signal acquisition steps to calculate the total harmonic distortion rate of the output voltage in real time.

[0102] The specific calculation method is as follows: perform a fast Fourier transform on the output voltage waveform to extract the fundamental amplitude and the amplitude of each harmonic. The total harmonic distortion rate is defined as the square root of the ratio of the sum of the squares of the effective values ​​of each harmonic to the effective value of the fundamental wave.

[0103] The controller compares the calculated total harmonic distortion rate with a preset distortion threshold, which is set to three percent. When the total harmonic distortion rate exceeds this threshold, the system automatically fine-tunes the dead zone compensation or adjusts the switching frequency of the space vector pulse width modulation until the harmonic content of the output voltage returns to within the threshold. This mechanism ensures that even under nonlinear loads or parameter perturbation conditions, the output voltage waveform still meets the power supply quality requirements of sensitive precision loads.

[0104] In this embodiment, the precise turn-on time of each bridge arm switch is calculated using a space vector pulse width modulation algorithm. Combined with an optimized switching sequence, zero vector configuration strategy, and dead-zone dynamic compensation technology, the voltage vector command output by the model predictive control is accurately reproduced as the actual voltage waveform at the AC output of the inverter while suppressing common-mode interference, reducing switching losses, and correcting waveform distortion. This completes the final execution stage of the entire voltage regulation and frequency stabilization control process.

[0105] After executing steps S1 to S5 in sequence, the voltage regulation and frequency stabilization method for energy storage emergency backup power supplies disclosed in this application fully covers all aspects from data acquisition and sensing, energy storage status assessment, frequency inertia simulation, voltage model prediction and optimization to final pulse width modulation execution.

[0106] On the other hand, the energy storage-type emergency backup power supply voltage stabilization system with voltage regulation and frequency stabilization function disclosed in this application includes: The frequency regulation model construction module is used to construct a virtual synchronous generator frequency regulation model. The frequency regulation model includes a virtual rotor motion equation. The virtual angular velocity is obtained by solving the virtual rotor motion equation. The virtual moment of inertia coefficient in the virtual rotor motion equation is adaptively adjusted based on the frequency change rate. The damping coefficient in the virtual rotor motion equation is dynamically adjusted based on the frequency deviation to output phase information for voltage regulation. The model predictive control module is used to establish a discrete-time state-space model including a three-phase converter and an LC filter. In each control cycle, it synthesizes a voltage reference trajectory based on phase information. Based on the discrete-time state-space model, it adopts a two-step prediction strategy to compensate for the computational delay of the digital control system. The two-step prediction strategy includes applying the voltage vector selected in the previous control cycle to perform a one-step forward prediction of the system state in the current control cycle, obtaining the state estimate at the start of the next control cycle, and using the state estimate as the initial state for prediction. The model predictive control module is also used to calculate the voltage prediction trajectory corresponding to all candidate voltage vectors in the prediction time domain with the state estimate as the initial state, and to evaluate each candidate voltage vector using an objective function that includes voltage tracking deviation and control increment, and select the voltage vector that minimizes the objective function value as the optimal voltage vector. The pulse width modulation module is used to convert the optimal voltage vector into a pulse width modulation signal, which is used to drive the power switching transistors of the three-phase converter to synthesize the output voltage waveform.

[0107] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects.

[0108] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A voltage regulation method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization functions, characterized in that, Includes the following steps: The virtual mechanical power command and the measured value of the actual active power output by the system are obtained as virtual electromagnetic power to construct a virtual synchronous generator frequency regulation model; the frequency regulation model includes the following virtual rotor motion equations: in, For virtual mechanical power, For virtual electromagnetic power, The virtual moment of inertia coefficient, For virtual angular velocity, This is the derivative of the virtual angular velocity with respect to time, i.e., the rate of change of frequency. The damping coefficient is... Rated angular velocity; The virtual angular velocity is obtained by solving the motion equations of the virtual rotor. And based on the frequency change rate Adaptive adjustment of the virtual rotational inertia coefficient and based on frequency deviation Dynamically adjust the damping coefficient For the virtual angular velocity The integrated output is used for voltage regulation phase information; A discrete-time state-space model including a three-phase converter and an LC filter is established. In each control cycle, a voltage reference trajectory is synthesized based on the phase information. Based on the discrete-time state-space model, a two-step prediction strategy is adopted to compensate for the computational delay of the digital control system. The two-step prediction strategy includes applying the voltage vector selected in the previous control cycle to perform a one-step forward prediction of the system state in the current control cycle to obtain the state estimate at the start of the next control cycle, and using the state estimate as the initial state for prediction. Using the state estimate as the initial state, the voltage prediction trajectory corresponding to all candidate voltage vectors is calculated in the prediction time domain. Each candidate voltage vector is evaluated using an objective function that includes voltage tracking deviation and control increment. The voltage vector that minimizes the objective function value is selected as the optimal voltage vector. The optimal voltage vector is converted into a pulse width modulation signal to drive the power switching transistors of the three-phase converter to synthesize the output voltage waveform.

2. The voltage stabilization method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization function according to claim 1, characterized in that, The virtual rotational inertia coefficient is adaptively adjusted based on the frequency change rate, which includes: when the absolute value of the frequency change rate is detected to be greater than the first preset change rate threshold, the virtual rotational inertia coefficient is gradually increased by a step size. When the absolute value of the rate of change of frequency falls below the second preset rate of change threshold, the virtual rotational inertia coefficient is gradually reduced with a second step size.

3. The voltage stabilization method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization function according to claim 1, characterized in that, The damping coefficient is dynamically adjusted based on frequency deviation, including: when the absolute value of the frequency deviation exceeds a preset deviation threshold, the damping coefficient is increased linearly according to the magnitude of the frequency deviation or according to a piecewise function curve.

4. The voltage stabilization method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization function according to claim 1, characterized in that, It also includes: during the transient process of switching from grid-connected mode to off-grid mode or from off-grid mode to grid-connected mode, increasing the damping coefficient to a preset maximum damping value, and restoring the damping coefficient to the dynamic adjustment value after the transient process ends.

5. The voltage regulation method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization function according to claim 1, characterized in that, The length N of the prediction time domain satisfies The integer; the objective function is expressed as: in, This is the weighting factor for voltage tracking deviation. To control the weighting coefficient of the increment, For the first Voltage reference trajectory at time t, For the corresponding voltage prediction trajectory, For the first The change in voltage vector at time t.

6. The voltage stabilization method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization function according to claim 1, characterized in that, Also includes: The extended Kalman filter algorithm is used to estimate the state of charge of the energy storage battery in real time. Based on the state of charge, the junction temperature of the power converter, the saturation current limit of the filter inductor, and the DC bus voltage support capability, the power output boundary is dynamically calculated and the output envelope is constructed. When the external power adjustment command exceeds the boundary of the output envelope, the actual output power command will be clamped within the boundary of the output envelope. The virtual mechanical power is limited by the actual output power command.

7. The voltage stabilization method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization function according to claim 1, characterized in that, The steps for converting the optimal voltage vector into a pulse width modulation signal include: using a space vector pulse width modulation algorithm to calculate the conduction time of each bridge arm switching device, and allowing only one of the three bridge arms to operate during any switching state transition, in order to suppress common-mode voltage fluctuations.

8. The voltage regulation method for an energy storage-type emergency backup power supply with voltage regulation and frequency stabilization function according to claim 7, characterized in that, Also includes: The polarity direction of the output current of each phase is monitored in real time during each pulse width modulation cycle. Based on the monitored current polarity direction, the on-time of the upper transistor and the on-time of the lower transistor of that phase are adjusted accordingly to compensate for the distortion of the output voltage waveform caused by the dead time.

9. A voltage regulation and frequency stabilization system for an energy storage-type emergency backup power supply, applied to the method described in any one of claims 1-8, characterized in that, include: The frequency regulation model construction module is used to construct a virtual synchronous generator frequency regulation model. The frequency regulation model includes a virtual rotor motion equation. The virtual angular velocity is obtained by solving the virtual rotor motion equation. The virtual moment of inertia coefficient in the virtual rotor motion equation is adaptively adjusted based on the frequency change rate. The damping coefficient in the virtual rotor motion equation is dynamically adjusted based on the frequency deviation to output phase information for voltage regulation. The model predictive control module is used to establish a discrete-time state-space model including a three-phase converter and an LC filter. In each control cycle, it synthesizes a voltage reference trajectory based on the phase information. Based on the discrete-time state-space model, it adopts a two-step prediction strategy to compensate for the computational delay of the digital control system. The two-step prediction strategy includes applying the voltage vector selected in the previous control cycle to perform a one-step forward prediction of the system state in the current control cycle to obtain the state estimate at the start of the next control cycle, and using the state estimate as the initial state for prediction. The model prediction control module is also used to use the state estimate as the initial state, to traverse and calculate the voltage prediction trajectory corresponding to all candidate voltage vectors in the prediction time domain, and to evaluate each candidate voltage vector using an objective function that includes voltage tracking deviation and control increment, and select the voltage vector that minimizes the value of the objective function as the optimal voltage vector. The pulse width modulation module is used to convert the optimal voltage vector into a pulse width modulation signal to drive the power switching transistors of the three-phase converter to synthesize the output voltage waveform.

Images