A park water-light-storage station full-black start and seamless grid-connected-off-grid switching method based on network construction type energy storage

Abstract

This invention discloses a method for full-dark start-up and seamless grid-connected / off-grid switching of a hydro-solar-storage station in a park based on grid-type energy storage. The method includes: passively identifying the five-dimensional characteristic parameters of the hydro-generator unit using the high-frequency modulation signal output by the grid-type energy storage; during full-dark start-up, constructing a time-reversal mapping curve of the hydraulic torque based on the characteristic parameters, controlling the energy storage to output power according to the curve to start the hydro-generator unit, and performing pre-activation processing and sequential photovoltaic grid connection; during grid-connected operation, real-time acquisition of energy storage electrical quantities, grid electrical quantities, and turbine mechanical shaft angular velocity is performed, decoupling to obtain the angular velocity deviation and grid impedance, and substituting these into the interlayer boundary mapping control law to predict the switching time and phase correction amount; after the switching command is issued, the converter switches to a grid-type damped injection hybrid mode, and performs phase correction circuit breaker operation after closing verification at the predicted switching time. This invention achieves smooth power matching during full-dark start-up and oscillation-free stable operation after seamless grid-connected / off-grid switching.

Description

Technical Field

[0001] This invention relates to the field of intelligent decision-making technology in the supply chain, and in particular to a method for the full-dark start-up and seamless grid-connected / off-grid switching of a park-based hydro-solar-storage station based on grid-connected energy storage. Background Technology

[0002] Grid-based energy storage systems, due to their ability to establish their own voltage and frequency, have become a core supporting technology for black start and islanded operation of new energy power plants. In recent years, black start schemes based on grid-based energy storage have been verified in several megawatt-level new energy bases, realizing the function of energy storage as the main power source to gradually restore photovoltaic, wind power, and part of the load. In terms of seamless switching between grid and off-grid, existing technologies mainly use methods such as pre-synchronization adjustment, adaptive impedance control, or virtual resistance pre-synchronization to enable the grid-based converter to perform closing or opening operations when it detects that the voltage, frequency, and phase meet preset thresholds, controlling the voltage surge during the switching transient process to within 5% of the rated value.

[0003] However, the aforementioned technologies are mainly designed for single-grid energy storage systems or wind-solar-storage systems. When the system also includes small hydropower units, existing methods reveal two fundamental problems that have not yet been effectively solved. First, small hydropower units experience a spatiotemporal inversion effect during startup: energy storage establishes voltage and outputs power with a millisecond-level response speed, while the hydro-turbine generator's hydraulic mechanical system has a large inertial response on the order of seconds. The energy storage output curve and the hydraulic torque growth curve are misaligned or even canceled on the time axis, leading to power matching failure during startup and repeated frequency oscillations. Although existing black-start strategies mention combining the turbine output characteristics, they essentially still treat hydropower as a passive follower after energy storage startup, failing to fundamentally solve the time-scale mismatch problem between the two power sources. Secondly, there is a coupled oscillation between impedance discontinuity and structural reconfiguration lag during the grid connection and disconnection process: the electrical control switching of energy storage can be completed in milliseconds, while the mechanical adjustment of the hydropower unit speed governor takes hundreds of milliseconds to several seconds. The speed difference between the two constitutes a control blind zone that cannot be covered by existing pre-synchronization methods, resulting in continuous power oscillation within a few seconds after the switch.

[0004] To address the aforementioned issues, existing research has attempted to mitigate oscillations by improving VSG control parameters or increasing virtual impedance. However, none of these studies have incorporated the dynamic response of the turbine's mechanical shaft into the observation and control system, nor have they proposed a method for designing the energy storage output waveform that can actively match the inertia of the hydraulic machinery. Therefore, there is an urgent need for a method that can eliminate spatiotemporal phase reversal effects and compensate for control blind spots during full-black start-up and seamless grid-connected / off-grid switching. Summary of the Invention

[0005] The purpose of this invention is to provide a method for the full-dark start-up and seamless grid-connected / off-grid switching of a park-based hydro-solar energy storage station based on grid-connected energy storage, in order to solve at least some of the problems existing in the prior art.

[0006] The technical solution, a method for seamless grid-connected / off-grid switching of a park-based hydro-solar-storage station in complete darkness, includes the following steps:

[0007] Based on the high-frequency modulation signal output by the grid-type energy storage converter, the parallel small hydropower units are passively identified to obtain their hydraulic and electrical characteristic parameters.

[0008] After the start command in complete darkness is triggered, a time-reversal mapping curve of hydraulic torque is constructed based on the characteristic parameters, and the energy storage is controlled to output power according to the curve to start the hydropower unit.

[0009] After the hydropower units are connected to the grid at the same time, the energy storage power will be gradually transferred to the hydropower units, and the photovoltaic sub-arrays will be connected to the photovoltaic arrays one by one according to their electrical and environmental conditions.

[0010] During grid-connected operation, the energy storage electrical quantity, the grid electrical quantity and the turbine mechanical shaft angular velocity are collected in real time. After decoupling, the angular velocity deviation and grid impedance are obtained, and then substituted into the layer boundary mapping control law to predict the switching time and phase correction.

[0011] According to the switching command, the converter is switched to a hybrid mode of grid-type damped injection. After closing verification at the predicted switching time, the circuit breaker operation with phase correction is performed to complete the seamless switching.

[0012] Beneficial effects: By introducing the hydraulic mechanical response as an active constraint into the energy storage output waveform design and switching control, this invention eliminates the spatiotemporal phase reversal effect and control blind zone between energy storage and hydropower in the hydro-solar-storage system, and achieves smooth power matching during the dark start-up process and oscillation-free stable operation after grid connection and disconnection. Attached Figure Description

[0013] Figure 1 This is a flowchart of the overall solution of the present invention. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0015] like Figure 1 As shown, this embodiment describes in detail the data processing flow of a method for the full-dark start-up and seamless grid-connection / off-grid switching of a park-based hydro-solar-storage station based on grid-connected energy storage, specifically including:

[0016] S1: Based on the high-frequency modulation signal output by the grid-type energy storage converter, the parallel small hydropower units are passively identified to obtain their hydraulic and electrical characteristic parameters.

[0017] In this application, the passive identification of characteristic parameters of small hydropower units is performed first. This step is carried out continuously when the hydro-solar-storage station in the park is in normal grid-connected operation. The purpose is to obtain the core dynamic parameters of the hydropower units in reverse by utilizing the electrical output capability of the grid-type energy storage converter itself without adding additional physical sensors.

[0018] Data Acquisition Phase: The grid-type energy storage converter actively superimposes a low-amplitude, high-frequency pseudo-random sequence modulation signal onto its rated voltage reference. For example, the amplitude of this modulation signal is limited to within one percent of the rated voltage, the frequency range is set to 200 Hz to 2 kHz, and the pseudo-random sequence uses an m-sequence to ensure spectral flatness. Simultaneously, the three-phase instantaneous voltage and current values ​​at the converter output are synchronously acquired at a sampling frequency of 20 kHz, labeled as Ua, Ub, Uc and Ia, Ib, Ic respectively, forming the original voltage and current sequences.

[0019] Signal preprocessing stage: The acquired raw voltage and current sequences are input into a second-order high-pass digital filter with a cutoff frequency of 150 Hz. This filter is designed to remove the 50 Hz fundamental frequency and its low-order harmonic components, retaining only the high-frequency response components generated by the pseudo-random modulation signal. After filtering, the high-frequency voltage and current components are obtained.

[0020] Time-frequency analysis stage: Short-time Fourier transform is performed on the high-frequency response voltage and current components. Specifically, a Hanning window is selected as the analysis window function, with a window length of 256 sampling points and an overlap rate of 50% between adjacent windows. The transform result generates a two-dimensional time-frequency spectrum, which simultaneously contains the energy distribution information of the signal on both the time and frequency axes.

[0021] Feature extraction stage: Two types of key features are extracted from the time-frequency spectrum. The first type of feature is located in the high-frequency band from 200 Hz to 2 kHz: for each frequency point in this band, the amplitude decay curve over time is extracted, and the decay time constant of each curve is calculated. The second type of feature is located in the low-frequency band from 0.1 Hz to 10 Hz: the low-frequency oscillation amplitude envelope in this band is extracted to obtain the oscillation frequency and damping ratio. These two types of features together constitute the high-frequency decay characteristics and low-frequency oscillation mode data.

[0022] In the parameter inversion stage, the extracted high-frequency attenuation feature vector and low-frequency oscillation mode data are input into a pre-trained five-dimensional parameter inversion neural network model. The model adopts a five-input, five-output network structure, and its internal weight coefficients have been determined through extensive offline simulation and experimental data training. After forward computation, the model outputs estimated values ​​of five feature parameters: mechanical inertia time constant (in seconds, representing the kinetic energy of the generator rotor), water flow inertia time constant (in seconds, representing the inertial response of the water intake system to changes in water flow), governor integral coefficient (dimensionless, reflecting the speed of the governor's integral link), excitation voltage build-up time constant (in seconds, representing the speed at which the excitation system builds up voltage), and head sensitivity coefficient (in percentage per meter, reflecting the degree of influence of head fluctuations on the turbine's output torque). These five parameters are combined into a five-dimensional feature parameter vector.

[0023] Parameter Update Phase: According to a further improvement in this embodiment, the aforementioned five-dimensional feature parameter vector is updated every ten seconds during system grid-connected operation and stored in the state memory. The contents of this memory are always the latest identification results, for use in subsequent black boot and seamless handover steps.

[0024] S2: After the start command in complete darkness is triggered, the time-reverse mapping curve of hydraulic torque is constructed based on the characteristic parameters, and the energy storage is controlled to output power according to the curve to start the hydropower unit.

[0025] When the all-black start command is triggered, this scheme does not directly allow the energy storage to establish voltage and power in the traditional ramp manner. Instead, it first performs pre-activation processing, then constructs a time inversion mapping curve that precisely matches the response characteristics of the hydraulic machinery, and controls the output power of the energy storage accordingly.

[0026] Pre-activation processing stage: At the instant the all-black start command arrives, the shutdown status data of all hydropower units is collected, specifically including: rotor speed value output by the speed sensor (in revolutions per minute), guide vane opening feedback value (expressed as a percentage of rated opening), and excitation current feedback value (in amperes). The speed of each unit is compared with a threshold of 5% of the rated speed, and the excitation current is compared with a threshold of 2% of the rated excitation current. When the speed of a unit is lower than the above speed threshold and the excitation current is lower than the above current threshold, the logic judgment module outputs a pre-magnetization enable signal for that unit. In a preferred embodiment, for a unit that obtains the pre-magnetization enable signal, the control system immediately closes the solid-state relay between its excitation circuit and the auxiliary battery. Subsequently, the actual values ​​of the battery terminal voltage and excitation current are collected, and closed-loop regulation is performed using 5% of the rated excitation current as the target reference value. The error between the excitation current and the target reference value is continuously monitored. When the absolute value of the error is less than 1% and the state is maintained for more than two seconds, a pre-magnetization completion flag for that unit is generated. Only after all units have reached the logical truth value for the pre-magnetization completion flag can the next step of time inversion mapping curve generation be allowed.

[0027] Hydraulic torque response function construction stage: Read the latest five-dimensional characteristic parameter vector from the state memory, and extract the mechanical inertia time constant (denoted as T) from it. m ), the inertial time constant of the water flow (denoted as T) w ), governor integral coefficient (denoted as K) i ) and head sensitivity coefficient (denoted as S) H Based on these parameters, a response function for the hydraulic torque as a function of time is constructed. For example, this function is: T hyd (t)=T base ·(1-e -(t-tdelay) / τ )·(1+S H ·ΔH)·u(tt delay ); where: T hyd (t) represents the hydraulic torque at time t (unit: Newton-meter or per unit); T base The base torque is typically taken as the steady-state output torque of the turbine at rated head and rated guide vane opening; τ is the comprehensive time constant, defined as τ=0.5T. w +T m ;t delay Let t be the delay time. delay =0.1K i ΔH is the deviation of the current water head from the rated water head (unit: meters), initially set to zero; u(tt) delay ) is a unit step function, when t <t delay The time value is 0 when t ≥ t delayThe time value is 1, used to simulate the pure delay of water flow impacting the runner after the guide vanes open. This function describes the complete dynamic process of water flow driving the runner to generate mechanical torque after the turbine guide vane opening command is issued.

[0028] Discretization and Inversion Mapping Stage: The hydraulic torque response function is discretized over the time interval required to build up from zero to rated power. In this embodiment, the discretization step size is 1 millisecond, resulting in a discrete torque sequence, where each element represents the hydraulic torque value at the corresponding moment. Further, a time inversion operation is performed on this discrete torque sequence: the order of the elements in the sequence is completely reversed, i.e., the first element is swapped with the last element, the second element with the second-to-last element, and so on, resulting in an inversion sequence. The physical significance of this inversion sequence lies in its description of the power waveform that the energy storage should output to ensure precise time matching between the energy storage output and the hydraulic torque: when the hydraulic torque reaches its peak in the later stages, the energy storage should output the corresponding power earlier, thus forming a temporal "mirror image" relationship.

[0029] In the power template curve generation stage: the inversion sequence is normalized so that its maximum value becomes 1. Then, this normalized sequence is multiplied by a scaling factor equal to the rated power of the energy storage divided by the maximum value of the inversion sequence. The resulting sequence is the energy storage power output template curve. This curve is characterized by: initially, the energy storage outputs relatively low power to establish voltage and excitation; as the hydraulic torque begins to increase, the energy storage output power rises synchronously; when the hydraulic torque approaches its peak, the energy storage output power decreases, thus keeping the sum of the energy storage output and the hydraulic output smooth throughout the startup process.

[0030] Energy storage output phase: The energy storage power output template curve is used as the active power command and is sent point-by-point to the power inner loop of the energy storage converter with a control cycle of 1 millisecond. The converter executes space vector pulse width modulation (SVPWM) according to the current command value, and the output voltage rises point-by-point from zero according to the power value specified by the template curve. At the same time, the DC side voltage and current output power of the energy storage converter are collected in real time and compared with the expected value in the template curve. When the relative error between the actual output power and the template value exceeds 5%, the power point tracking correction stage is triggered, and the deviation is eliminated by adjusting the modulation ratio.

[0031] S3: After the hydropower units are connected to the grid simultaneously, the energy storage power will be gradually transferred to the hydropower units, and the photovoltaic subarrays will be connected to the photovoltaic arrays one by one according to their electrical and environmental conditions.

[0032] After the hydropower unit reaches its rated speed and rated voltage with the energy storage activated, it performs synchronous grid connection. Once grid connection is complete, this scheme executes two parallel processing flows: power transfer and photovoltaic unit-by-unit connection.

[0033] Synchronization and grid connection condition determination: Under the reference bus voltage established by the energy storage, the three-phase voltage, frequency, and phase of the hydropower unit outlet side are collected, and the corresponding electrical quantities on the bus side are also collected. The voltage difference, frequency difference, and phase difference are calculated. When the voltage difference is less than 5% of the rated voltage, the frequency difference is less than 0.1 Hz, and the phase difference is less than 2 degrees, and these three conditions are simultaneously met and sustained for more than 100 milliseconds, the synchronization check device outputs a closing permission signal. The control system then closes the generator outlet circuit breaker accordingly to complete the grid connection.

[0034] Power transfer phase: After grid connection, the active power output of the generator (denoted as P) is simultaneously collected at a frequency of 100 Hz. gen ) and the active power output of energy storage (denoted as P) ess The calculated power allocation ratio is: P gen ÷(P gen +P ess The power transfer ramp time is set to 10 seconds. Within each control cycle, the active power command of the energy storage unit is gradually decreased from its current value using a linear ramp function, with a final target value of 10% of the rated power. Simultaneously, the governor command of the hydropower unit is gradually increased from its current value using the exact same ramp function, causing the active power of the hydropower unit to gradually rise to over 90% of its rated power. This ramp transfer process ensures the smoothness of power exchange and avoids frequency fluctuations caused by sudden power changes. The power allocation ratio is continuously monitored; when the ratio is greater than 0.9, the power transfer is considered complete. At this point, the energy storage converter switches from main power output mode to voltage source standby mode, retaining only voltage support and frequency regulation capabilities, and no longer undertaking base load power output.

[0035] Photovoltaic array integration phase: Integration of photovoltaic subarrays begins sequentially, either in parallel with or after power transfer. The DC-side voltage, ground insulation resistance, and irradiance (in watts per square meter) of each photovoltaic subarray are collected at a frequency of 10 Hz. The following thresholds are set for startup conditions: DC-side voltage greater than 80% of the minimum maximum power point tracking (MPPT) voltage; ground insulation resistance greater than 1 megohm; and irradiance greater than 100 watts per square meter. Subarrays that simultaneously meet all three conditions are marked as "start-up allowed" and sorted in descending order of irradiance value, with the subarray having the best irradiance condition being prioritized for startup.

[0036] In one alternative implementation, for the first-order subarray, its DC-side circuit breaker is closed first, followed by the AC-side grid-connection contactor. At the moment of connection, the output frequency of the energy storage converter and the grid-connected current of the photovoltaic subarray are monitored at a sampling frequency of 2 kHz. If the energy storage output frequency deviates from the rated 50 Hz by more than ±0.2 Hz, or the total harmonic distortion rate of the grid-connected current exceeds 5%, the connection is deemed to have caused an unacceptable disturbance to the system. In this case, the startup operation of the next subarray is temporarily blocked, and the system is allowed to stabilize for 3 seconds before starting the next subarray. This process continues until all photovoltaic subarrays marked "start-up permitted" are successfully connected. After all connections are completed, the bus voltage and frequency are collected, and it is confirmed that both operate stably within ±2% of their rated values ​​for more than 30 seconds. At this point, the black-out startup process is considered complete.

[0037] S4: During grid-connected operation, real-time data are collected on energy storage electrical quantities, grid electrical quantities, and turbine mechanical shaft angular velocity. After decoupling, the angular velocity deviation and grid impedance are obtained and substituted into the interlayer boundary mapping control law to predict the switching time and phase correction.

[0038] When the hydro-solar storage station in the park is in normal grid-connected operation, this solution continuously performs boundary observation and predictive calculations to prepare the necessary control parameters in advance for possible seamless switching between grid connection and off-grid operation.

[0039] Multi-source data synchronous acquisition phase: Three types of data are continuously acquired at a sampling frequency of 20 kHz. The first type is the three-phase voltage and three-phase current at the output of the energy storage converter; the second type is the three-phase voltage and three-phase current at the grid connection point on the grid side; the third type is the mechanical angular velocity and angular acceleration values ​​output by the incremental encoder installed at the turbine shaft. All acquired data are accompanied by high-precision timestamps and stored in a circular buffer with a depth of 500 milliseconds for subsequent time alignment and extrapolation prediction.

[0040] Electrical angular velocity calculation and deviation decoupling stage: The fundamental positive-sequence component is extracted from the three-phase voltage and current at the output of the energy storage converter. The extraction of the fundamental positive-sequence component can employ a phase-locked loop based on a second-order generalized integrator or a positive-sequence separation algorithm based on a notch filter. Based on the phase angle of the extracted fundamental positive-sequence voltage, the electrical angular velocity value is obtained by differentiating it with respect to time. Simultaneously, the latest turbine mechanical angular velocity value is read from the circular buffer. Subtracting the electrical angular velocity from the mechanical angular velocity yields the angular velocity deviation sequence. This deviation sequence reflects the degree of speed asynchrony between the electrical and mechanical sides. A moving average filter is applied to this deviation sequence with a window length of 20 milliseconds to eliminate high-frequency noise interference, resulting in a filtered angular velocity deviation sequence.

[0041] Online grid impedance identification stage: In a preferred embodiment, to obtain the real-time value of the grid-side equivalent impedance, the grid-connected energy storage converter injects a pulse voltage disturbance with an amplitude of 0.5% of the rated voltage into the grid side. This pulse disturbance has an extremely short duration (e.g., 1 millisecond) and will not have a noticeable impact on normal power quality. The amplitude and phase changes of the grid-side response current are measured, and the grid equivalent resistance and equivalent reactance are solved online using a least-squares recursive algorithm to finally obtain the grid impedance identification value, expressed as a complex number of resistance and reactance values. This identification value is updated every 10 milliseconds to track rapid changes in grid impedance.

[0042] The boundary mapping control law calculation stage involves inputting the filtered angular velocity deviation sequence and the grid impedance identification value into the boundary mapping control law. This control law comprises two core calculation branches. The first branch calculates the dynamic damping coefficient: based on the absolute value of the angular velocity deviation, the damping coefficient is calculated using an exponential decay function. This means that the larger the angular velocity deviation, the larger the damping coefficient, thus applying stronger electrical damping during switching to suppress mechanical oscillations. The second branch calculates the phase correction time: based on the rate of change of grid impedance (i.e., the change in impedance per unit time) and the rate of change of angular velocity deviation (i.e., angular acceleration), a time correction is calculated through a linear combination. This correction is used to compensate for the inherent delay between electrical control and mechanical response.

[0043] The switching time and phase correction prediction stage: A linear extrapolation is performed on the angular velocity deviation sequence and grid impedance identification value over the past 100 milliseconds to predict the trajectory of change over the next 50 milliseconds. Simultaneously, based on the voltage amplitude difference, frequency difference, and phase difference between the current energy storage output terminal and the grid side, the changing trends of these differences on the future time axis are extrapolated. When the extrapolation results show that the voltage difference, frequency difference, and phase difference will simultaneously meet the pre-synchronization conditions at a future moment (for example, the voltage difference is less than one percent, the frequency difference is less than 0.05 Hz, and the phase difference is less than 1 degree), this moment is recorded as the "pre-synchronization fulfillment moment." Subsequently, the phase correction time calculated in the previous step is added to this pre-synchronization fulfillment moment to obtain the final predicted switching time. At the same time, the phase correction time is multiplied by the rated angular frequency (314 radians per second) to convert it into a phase correction angle value. The phase correction angle indicates that after the traditional pre-synchronization conditions are met, an additional short waiting time is required so that the switching action falls precisely on the phase point with the minimum impedance fluctuation.

[0044] S5: According to the switching command, the converter is switched to a hybrid mode of grid-type damped injection. After closing verification at the predicted switching time, the circuit breaker operation with phase correction is performed to complete the seamless switching.

[0045] When the dispatching layer or local operation panel issues a switching command from grid-connected to off-grid (or off-grid to grid-connected), this solution performs mode switching, timing alignment and final closing operations to achieve a truly "seamless" switching.

[0046] Hybrid control mode switching phase: Within the first control cycle (1 millisecond) after the switching command arrives, the internal control structure of the grid-type converter is immediately modified. Specifically, a damping injection term is added to the original active-frequency droop control loop. The magnitude of this damping injection term is proportional to the product of the dynamic damping coefficient and the mechanical angular acceleration. Simultaneously, a transient damping term is added to the reactive-voltage control loop, the magnitude of which is proportional to the product of the transient damping coefficient and the voltage change rate. After the above modifications, the converter's control mode changes from pure grid-type (GFM) to a hybrid mode of "grid-type plus damping injection" (GFM+DI). This hybrid mode can actively suppress oscillations caused by mechanical inertia and electrical control asynchrony during the switching transient process.

[0047] Switching Time Alignment and Synchronization Acquisition Phase: Using the Global Positioning System (GPS) or IEEE 1588 Precision Time Protocol (PTP) synchronization signal as a high-precision clock reference, the current time is compared with the predicted switching time in real time. When the absolute value of the difference between the two is less than 1 millisecond, a burst synchronization acquisition operation is triggered. Burst acquisition records the instantaneous waveforms of the circuit breaker grid-side voltage and energy storage-side voltage at an extremely high sampling rate of 100 kHz, while simultaneously calculating the instantaneous frequency and instantaneous phase on both sides. Based on these instantaneous data, the instantaneous voltage difference (the absolute value of the difference between two instantaneous voltage values) and the instantaneous phase difference (the absolute value of the difference between two phase angles) are calculated.

[0048] Closing condition verification and phase correction execution phase: The instantaneous voltage difference and instantaneous phase difference mentioned above are judged. If the instantaneous voltage difference is less than 0.5% of the rated voltage and the instantaneous phase difference is less than 0.5 degrees, the closing condition is determined to be met, and the circuit breaker closing pulse is immediately issued. Conversely, if the above conditions are not yet met, but it is predicted that the conditions will be met soon within a 2-millisecond time window before and after the predicted switching time, a phase correction strategy is adopted: the issuance time of the closing pulse is delayed by a duration equal to the phase correction angle calculated in the previous step divided by the rated angular frequency. Through this delay, the circuit breaker completes closing at the most favorable phase point in the electromagnetic transient state, thereby minimizing the closing inrush current.

[0049] Switching Completion Confirmation and Mode Recovery Phase: After the circuit breaker operates, feedback signals from its auxiliary contacts are collected to confirm that the switching is complete. After the switching is completed, the converter continues to operate in a hybrid "grid-type plus damping injection" mode for 200 milliseconds (i.e., 10 power frequency cycles) to fully absorb any residual oscillations that may occur after the switching. Subsequently, the coefficients of the damping injection term and the transient damping term are gradually reduced linearly to zero, allowing the converter to smoothly return to a single grid-type control mode (if the switch is in an off-grid islanded operation state) or a single grid-connected control mode (if the switch is in a grid-connected operation state). At this point, the entire seamless switching process between grid and off-grid is completed. Throughout the process, fluctuations in bus voltage and frequency are controlled within the allowable range, and the load equipment does not experience any power outages or shocks.

[0050] This invention has the following significant advantages: First, by introducing a power storage output waveform design method based on hydraulic torque time inversion mapping, the large inertial mechanical response of the hydropower unit is treated as an active constraint rather than a passive follower. This eliminates the spatiotemporal inversion effect between the millisecond-level output of energy storage and the second-level torque of hydraulic power in traditional black starts. This allows energy storage to accurately fill the power gap during the hydraulic torque ramp-up and automatically exit after the hydraulic output reaches its peak, achieving smooth power matching throughout the start-up process and avoiding repeated frequency oscillations. Second, by incorporating the turbine's mechanical shaft angular velocity and angular acceleration into the on-grid / off-grid switching observation and control system, a layer boundary mapping control law is proposed. Based on electrical pre-synchronization, dynamic damping injection and phase correction delay are introduced, filling the control blind zone between the millisecond-level switching of electrical control and the second-level response of mechanical adjustment. This reduces the voltage surge during the switching transient process to less than 0.5% of the rated value, and shortens the oscillation decay time after switching to less than 0.3 seconds. Third, by utilizing the low-amplitude, high-frequency modulated signal output by the grid-type energy storage system itself, the five-dimensional characteristic parameters of the hydropower unit are passively identified. This allows for adaptive tracking of slow time-varying characteristics such as head changes and equipment aging without the need for additional sensors, enabling the black-start and switching strategies to possess long-term self-optimization capabilities. Fourth, photovoltaic subarrays are connected group by group according to irradiance, and frequency and harmonic distortion rates are monitored in real time, effectively suppressing the impact of photovoltaic fluctuations on the islanded system. In summary, this invention overcomes the limitations of existing technologies that only focus on electrical quantity control, providing a highly robust, adaptive, and seamless black-start solution for park-level hydro-photovoltaic-storage stations containing small hydropower.

[0051] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A method for full-dark start-up and seamless grid-connection / off-grid switching of a park-based hydro-solar-storage station based on grid-connected energy storage, characterized in that, include: Based on the high-frequency modulation signal output by the grid-type energy storage converter, the parallel small hydropower units are passively identified to obtain their hydraulic and electrical characteristic parameters. After the start command in complete darkness is triggered, a time-reversal mapping curve of hydraulic torque is constructed based on the characteristic parameters, and the energy storage is controlled to output power according to the curve to start the hydropower unit. After the hydropower units are connected to the grid simultaneously, the energy storage power will be gradually transferred to the hydropower units, and the photovoltaic sub-arrays will be connected to the photovoltaic arrays one by one according to their electrical and environmental conditions. During grid-connected operation, the energy storage electrical quantity, the grid electrical quantity and the turbine mechanical shaft angular velocity are collected in real time. After decoupling, the angular velocity deviation and grid impedance are obtained, and then substituted into the layer boundary mapping control law to predict the switching time and phase correction. According to the switching command, the converter is switched to a hybrid mode of grid-type damped injection. After closing verification at the predicted switching time, the circuit breaker operation with phase correction is performed to complete the seamless switching.

2. The method according to claim 1, characterized in that, Passive identification of parallel small hydropower units is performed to obtain hydraulic and electrical characteristic parameters, including: During normal grid-connected operation, a low-amplitude, high-frequency pseudo-random sequence modulation signal is superimposed on the rated voltage by the grid-type energy storage converter; The three-phase voltage and current at the output of the converter are collected, and the time-frequency spectrum is obtained by high-pass filtering and short-time Fourier transform. Features are extracted from the time-frequency spectrum: the amplitude decay time constant of the high-frequency components and the frequency and damping ratio of the low-frequency oscillations; The extracted features are input into a pre-trained five-dimensional inversion neural network. The network outputs the mechanical inertia time constant, the water flow inertia time constant, the governor integral coefficient, the excitation pressure build-up time constant, and the head sensitivity coefficient, forming a five-dimensional feature parameter vector.

3. The method according to claim 1, characterized in that, After the all-black startup command is triggered, and before the time-reversal mapping curve is generated, pre-activation processing is also included: Collect the speed, guide vane opening, and excitation current of all hydropower units; When the speed of a certain unit is lower than 5% of the rated speed and the excitation current is lower than 2% of the rated excitation current, a pre-magnetization enable signal for the unit is generated. Based on the pre-magnetization enable signal, the solid-state relay between the excitation circuit of the unit and the auxiliary battery is closed, and closed-loop regulation is performed with a target of five percent of the rated excitation current until the excitation current stabilizes near the target value for more than two seconds, generating a pre-magnetization completion flag.

4. The method according to claim 1, characterized in that, Constructing a time-reversal mapping curve for hydraulic torque, and controlling the energy storage to output power according to this curve to start the hydroelectric generator, includes: The mechanical inertia time constant, water flow inertia time constant, governor integral coefficient, and head sensitivity coefficient are read from the five-dimensional feature parameter vector to construct the response function of hydraulic torque as time increases. Discretize the response function over the time interval required to build up from zero to rated power to obtain a discrete torque sequence. The discrete torque sequence is reversed on the time axis to obtain the inverted sequence; the inverted sequence is normalized and multiplied by the scaling factor corresponding to the rated power of the energy storage to form the energy storage power output template curve. Using the template curve as the active power command, it is sent to the power inner loop of the energy storage converter every millisecond to perform pulse width modulation output.

5. The method according to claim 1, characterized in that, Gradually transfer energy storage power to hydropower units, including: After the hydropower units are connected to the grid simultaneously, the active power output of the generator and the active power output of the energy storage are collected at the same time, and the power distribution ratio between the two is calculated. Set a power transfer ramp time of ten seconds. In each control cycle, the active power command of the energy storage unit is reduced from the current value to 10% of the rated power by a linear ramp. At the same time, the speed governor command of the hydropower unit is increased by the same ramp, so that the active power of the hydropower unit rises to more than 90% of the rated power. When the power allocation ratio is greater than 0.9, the transfer is considered complete, and the energy storage converter switches to voltage source standby mode.

6. The method according to claim 1, characterized in that, Photovoltaic arrays are connected to the photovoltaic system one by one according to their electrical and environmental conditions, including: The DC-side voltage, ground insulation impedance, and irradiance of each photovoltaic subarray are collected at a frequency of 10 Hz. The startup conditions are set as follows: DC side voltage is greater than 80% of the minimum voltage at maximum power point tracking, insulation resistance is greater than one megohm, and irradiance is greater than one hundred watts per square meter. For subarrays that meet all conditions, sort them from highest to lowest irradiance. The DC circuit breakers and AC contactors of each subarray are closed in sequence, and the output frequency of the energy storage converter and the photovoltaic grid-connected current are monitored at the moment of connection. If the frequency deviates from the rated value by more than 0.2 Hz or the grid-connected current harmonic distortion rate exceeds 5%, the startup of the next subarray will be paused, and the system will be allowed to stabilize for three seconds before continuing, until all qualified subarrays are connected.

7. The method according to claim 1, characterized in that, Real-time acquisition of energy storage electrical quantities, grid electrical quantities, and turbine mechanical shaft angular velocity; decoupling yields angular velocity deviation and grid impedance, including: The three-phase voltage and current at the output of the energy storage converter, the three-phase voltage and current at the grid connection point on the grid side, and the mechanical angular velocity and angular acceleration at the turbine shaft are continuously collected at a sampling frequency of 20 kHz. The fundamental positive sequence component is extracted from the voltage and current of the energy storage terminal, the electrical angular velocity is calculated, and the angular velocity deviation sequence is obtained by subtracting it from the mechanical angular velocity. The sequence is then filtered by moving average. A pulse voltage disturbance with an amplitude of 0.5% of the rated voltage is injected into the grid side, the response current is measured, and the equivalent resistance and reactance of the grid are identified online using the least squares method to obtain the grid impedance identification value.

8. The method according to claim 1, characterized in that, Substituting the boundary mapping control law into the prediction of switching times and phase corrections, including: The filtered angular velocity deviation sequence and the grid impedance identification value are input into the interlayer boundary mapping control law: the damping coefficient is dynamically calculated based on the absolute value of the angular velocity deviation, and the larger the deviation, the larger the damping coefficient. Calculate the phase correction time based on the rate of change of grid impedance and the rate of change of angular velocity deviation; By linearly extrapolating the angular velocity deviation and grid impedance over the past 100 milliseconds, the trajectory of change over the next 50 milliseconds is predicted. When the predicted voltage difference, frequency difference, and phase difference will simultaneously meet the pre-synchronization conditions, record the time when the prediction is met, add the phase correction time to this time as the prediction switching time, and convert the phase correction time into the corresponding phase correction angle.

9. The method according to claim 1, characterized in that, According to the switching command, the converter is switched to a hybrid mode of grid-type damped injection. At the predicted switching time, after closing verification, a phase-corrected circuit breaker operation is performed, including: In the first control cycle after the switching command arrives, a damping injection term is added to the active power-frequency droop control of the converter. This injection term is proportional to the product of the damping coefficient and the angular acceleration. At the same time, a transient damping term is added to the reactive power-voltage control, forming a hybrid control mode of grid-type damping injection. Using a high-precision synchronous clock as a reference, the current time is compared with the predicted switching time in real time. When the difference between the two is less than one millisecond, the voltage and frequency on both sides of the circuit breaker are collected at a burst sampling rate to calculate the instantaneous voltage difference and phase difference. If the instantaneous voltage difference is less than 0.5 percent of the rated voltage and the instantaneous phase difference is less than 0.5 degrees, a closing pulse will be issued immediately; if the conditions are not met but are expected to be met within a two-millisecond window before and after the predicted switching time, the closing pulse will be issued after a delay corresponding to the phase correction angle. After the circuit breaker operates, the auxiliary contact feedback is collected to confirm the switching is complete, and it continues to run in hybrid mode for 200 milliseconds. Then, the damping injection is gradually removed, and it returns to single control mode.

10. The method according to any one of claims 1 to 9, characterized in that, The five-dimensional feature parameter vector is updated every ten seconds and stored in the state memory during the system's grid-connected operation. The time inversion mapping curve during full-black startup and the boundary prediction during seamless switching are both calculated based on the latest stored five-dimensional feature parameter vector.

Images