A gas-energy storage dynamic decoupling predictive control system and method

Abstract

The application discloses a kind of gas-energy storage dynamic decoupling predictive control system and method, belong to power system operation and control technical field.The application real-time acquisition three-phase voltage signal at grid-connected point PCC, extract frequency variation rate df / dt and frequency deviation Δf as feedforward starting signal;Through prediction control unit, the future short-time frequency evolution state is estimated based on unscented Kalman filter UKF, and the predicted frequency trajectory is output;Model predictive control decision unit takes the predicted trajectory as time-varying reference quantity, and combines energy storage SOC, charge-discharge power limit and ramp rate constraint to solve energy storage compensation power sequence;Energy storage converter outputs compensation power before gas turbine governor action, fills in transient power vacuum period, and gas turbine subsequently undertakes medium-low frequency and steady-state power demand, so as to realize the dynamic decoupling of both in response time scale.The application aims at solving the problems of control lag and lack of prediction in the prior art of gas-energy storage hybrid power generation system.

Description

Technical Field

[0001] This invention belongs to the field of power system operation and control technology, specifically relating to a dynamic decoupling predictive control system and method for gas-energy storage. Background Technology

[0002] Currently, the mainstream coordinated control strategy for hybrid power generation systems incorporating gas turbines and energy storage is based on a responsive feedback control architecture. The control system tracks the grid phase via a phase-locked loop (PLL) and monitors in real time the frequency deviation Δf or power deviation ΔP at the local grid connection point or load side. When these electrical quantities deviate from their rated values ​​and exceed set thresholds, the energy storage converter (PCS) activates, adjusting power according to the droop control curve to smooth system fluctuations. The gas turbine primarily undertakes the task of steady-state base load or low-to-medium frequency power regulation.

[0003] Current technologies rely on frequency deviation Δf as the trigger signal. According to the fundamental physical laws of power systems, the evolution sequence of power imbalance events is: the appearance of a power deficit, a sudden change in the rate of frequency change df / dt, and the accumulation of frequency deviation Δf. Existing solutions must wait for a noticeable and significant frequency shift, typically taking 50ms-200ms, before taking action, which is a reactive compensation. This lag time, combined with the 200ms-500ms mechanical response time of the gas turbine, results in a power vacuum period during the transient process, leading to a lower minimum frequency and potentially triggering low-frequency load shedding in severe cases.

[0004] Traditional solutions typically measure data at the local end of the equipment, the load side, or the energy storage access point. The electrical quantities at these locations may be superimposed with local load switching impacts, converter switching harmonics, and local line disturbances. If the energy storage system operates directly based on these local signals, it is prone to unnecessary charge and discharge adjustments, increasing the battery cycle life and reducing the accuracy of identifying upstream power imbalance trends.

[0005] Existing control strategies primarily focus on frequency deviation suppression or steady-state power allocation, failing to fully incorporate factors such as the energy storage's state of charge (SOC), battery charge / discharge power limits, power change rate limits, and energy storage lifetime protection into a unified optimization framework. Therefore, achieving a coordinated balance between frequency support and energy storage health management is difficult, easily leading to overcharging, over-discharging, or current surges in energy storage.

[0006] Existing methods can only adjust based on the current state and cannot predict the future trend of grid frequency evolution within a few hundred milliseconds. Therefore, they cannot achieve optimal compensation and are prone to overcharging or over-discharging of energy storage, affecting equipment lifespan and system stability. Summary of the Invention

[0007] This invention provides a dynamic decoupling predictive control system and method for gas-energy storage, aiming to solve the comprehensive problems in the prior art, such as power vacuum periods in transient processes caused by control logic lag, susceptibility of local control signals to interference leading to malfunctions in energy storage, single control objective lacking global optimization and coordination, and strong control blindness due to lack of situation prediction capability.

[0008] This invention is achieved through the following technical solution: A dynamic decoupling predictive control system for gas turbine-energy storage, comprising a gas turbine power generation unit, an energy storage unit, an energy storage converter PCS, an AC bus, and a grid-connected common coupling point PCC, and further including a state sensing module, a predictive control unit, a model predictive control decision unit, and a gas turbine control unit; The voltage sampling terminal of the state sensing module is connected to the AC bus at the grid common coupling point PCC, and is used to collect the three-phase voltage signal at the PCC point in real time, and extract the grid frequency f, frequency deviation Δf and frequency change rate df / dt based on the three-phase voltage signal. The prediction control unit is communicatively connected to the state perception module and is used to receive at least the frequency change rate df / dt and the current frequency deviation Δf, and to estimate the future short-term frequency evolution state based on the power grid frequency response model, and output the predicted frequency trajectory. The model predictive control decision unit is communicatively connected to the predictive control unit. It is used to receive the predicted frequency trajectory and, in combination with the energy storage state of charge (SOC), energy storage charging and discharging power limits, power change rate limits, and PCS operating constraints, to solve the energy storage compensation power sequence in the finite prediction time domain. The control signal input terminal of the energy storage converter is communicatively connected to the output terminal of the model predictive control decision unit, and is used to receive the first value of the energy storage compensation power command, and control the energy storage unit to output or absorb compensation power before the load fluctuation causes a significant shift in the electrical quantity at the PCC point. The gas turbine control unit is used to adjust the mechanical power output of the gas turbine according to the system steady-state power demand, low-frequency power deviation, or frequency deviation signal. In the early stages of transient disturbances, the energy storage unit undertakes high-frequency transient power compensation based on the df / dt feedforward prediction results. After the gas turbine governor and mechanical power gradually respond, the gas turbine undertakes the medium- and low-frequency and steady-state power demands, and the energy storage unit gradually reduces the compensation power and restores the SOC, thereby realizing the dynamic decoupling of the gas turbine and the energy storage system based on the response frequency band and time scale.

[0009] Furthermore, the state sensing module calculates the angular frequency using the phase angle output by the phase-locked loop (PLL) and obtains the frequency signal based on the angular frequency; it further obtains the frequency change rate df / dt through sliding time window fitting, differential calculation, or filtering differentiation; the sampling frequency of the state sensing module is higher than 1kHz, and the calculation time window for the frequency change rate df / dt is no greater than 5ms.

[0010] Furthermore, the prediction control unit incorporates an unscented Kalman filter (UKF) to filter out noise and local disturbances in the acquired signal and to estimate the state of the frequency evolution trajectory within a time window of 100ms to 500ms.

[0011] Furthermore, the objective function constructed by the model predictive control decision unit is:

[0012] in, N The total number of steps within the finite prediction time domain. f is the energy storage compensation power for the previous control step. pred (k) is the prediction frequency of the k-th prediction step, f ref The SOC(k) is the rated frequency or dynamic frequency reference value, and the SOC(k) is the energy storage state of charge. ref For energy storage target state of charge, P ess (k) represents the energy storage compensation power, q f q soc and q p These are the weighting coefficients corresponding to frequency deviation, SOC deviation, and power change rate, respectively. The MPC optimization process must satisfy at least the following constraints:

[0013]

[0014]

[0015]

[0016] in, For the k-th step battery current, SOC min and SOC max These are the lower and upper limits of energy storage SOC, respectively, P ess,min and P ess,max These represent the lower and upper limits of energy storage charging and discharging power, respectively. max I is the upper limit of the rate of change of energy storage power. bat,max This represents the upper limit of the battery's allowable current.

[0017] Furthermore, the state perception module, predictive control unit, MPC decision unit and energy storage converter communicate with each other via high-speed real-time industrial Ethernet or high-speed serial bus, with the communication cycle not exceeding the MPC control cycle, ensuring that the total delay from signal sampling to instruction execution is less than 30ms.

[0018] Furthermore, the system automatically switches its operating mode based on the electrical status of the PCC point: When |df / dt|≤ the set threshold, the power grid is judged to be in steady state, the energy storage is in standby or SOC maintenance state, and the gas turbine undertakes the base load; When |df / dt|> the set threshold, it is determined that a power imbalance has occurred in the power grid, and the energy storage is immediately triggered to output compensation power in a transient support mode with a millisecond-level response. When the energy storage SOC exceeds the preset safe range, the weight of the SOC recovery target is increased in the MPC optimization function to limit the energy storage from continuing deep charging and discharging, and enter the SOC over-limit protection mode.

[0019] A gas-energy storage dynamic decoupling predictive control method, using the gas-energy storage dynamic decoupling predictive control system as described above, the method includes the following steps: Step 1: The state sensing module detects the three-phase voltage in real time at the PCC point at a sampling frequency higher than 1kHz, and calculates and outputs the frequency change rate df / dt signal without waiting for a complete power frequency cycle. Step 2: The prediction control unit uses only the df / dt and the current Δf as observation values, inputs them into the unscented Kalman filter (UKF), performs optimal estimation of the system state, and outputs the predicted frequency trajectory; Step 3: The Model Predictive Control Decision Unit directly and uniquely uses the predicted frequency trajectory as the time-varying reference trajectory in its cost function. In each control cycle, it solves the optimal energy storage compensation power sequence, and the control cycle of the MPC is less than the effective variation cycle of the df / dt signal. Step 4: The energy storage converter outputs compensation power before the gas turbine governor operates, according to the issued command, to fill the power vacuum period and achieve decoupling of the response time scale between energy storage and gas turbine.

[0020] Furthermore, step 1 automatically switches the operating mode based on the electrical status of the PCC point: When |df / dt|≤ the set threshold, the power grid is judged to be in steady state, the energy storage is in standby or SOC maintenance state, and the gas turbine undertakes the base load; When |df / dt|> the set threshold, it is determined that a power imbalance has occurred in the power grid, and the energy storage is immediately triggered to output compensation power in a transient support mode with a millisecond-level response. When the energy storage SOC exceeds the preset safe range, the weight of the SOC recovery target is increased in the MPC optimization function to limit the energy storage from continuing deep charging and discharging, and enter the SOC over-limit protection mode.

[0021] A dynamic decoupling predictive control system for gas-energy storage, as described above, is applied to a microgrid in an industrial park.

[0022] The beneficial effects of this invention are: This invention shifts the control start signal from the result quantity Δf to the trend quantity df / dt, eliminating feedback lag and achieving advance compensation for energy storage before a significant frequency drop, thus advancing the effective response time of energy storage by more than 50ms. Combined with MPC prediction, it can fill the power vacuum period before the gas turbine operates, raising the lowest frequency point by more than 0.05Hz.

[0023] This invention combines PCC point signal acquisition with state estimation algorithms to improve the control signal's ability to reflect the overall power imbalance trend of the system and reduce the impact of local disturbances on energy storage control.

[0024] This invention introduces Model Predictive Control (MPC) to construct a multi-objective optimization framework that includes frequency support, energy storage SOC management, power constraints, and ramp rate constraints. This avoids overcharging and over-discharging of energy storage and current surges, extending the lifespan of energy storage by approximately 10%-15% while ensuring grid safety, and achieving rolling optimization of energy storage compensation power.

[0025] This invention achieves power splitting naturally by controlling the physical time scale of the signal, with energy storage handling transient high-frequency components and the gas turbine responsible for steady-state base load. This effectively solves the problems of delayed response and strong power coupling in traditional combined power supply systems, and improves the grid-friendliness of hybrid power generation systems. Attached Figure Description

[0026] Figure 1 This is the overall architecture and signal flow diagram of the present invention.

[0027] Figure 2 This is a schematic diagram of the PCC point signal acquisition circuit wiring of the present invention.

[0028] Figure 3 This is a comparison diagram of the response timing of the conventional frequency deviation control and the frequency change rate feedforward control of the present invention.

[0029] Figure 4 This is a flowchart of the three-layer control logic of perception, prediction, and decision-making in this invention. Detailed Implementation

[0030] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail.

[0031] It should be understood that, when used in this specification and the appended claims, the terms include indicating the presence of the described feature, integral, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0032] It should also be understood that the terminology used in this application specification is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "one," "an," and "that" are intended to include the plural forms.

[0033] The following is in conjunction with the appendix to this application specification. Figure 1-4 The technical solutions in the embodiments of this application are clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0034] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0035] Implementation Method 1 This embodiment provides a dynamic decoupling predictive control system for gas turbine-energy storage, which includes a gas turbine power generation unit, an energy storage unit, an energy storage converter PCS, an AC bus, and a common coupling point to the grid PCC. It also includes a state sensing module, a predictive control unit, a model predictive control decision unit, and a gas turbine control unit. The voltage sampling terminal of the state sensing module is connected to the AC bus at the grid common coupling point PCC, and is used to collect the three-phase voltage signal at the PCC point in real time, and extract the grid frequency f, frequency deviation Δf and frequency change rate df / dt based on the three-phase voltage signal.

[0036] The state sensing module calculates the angular frequency ω using the phase angle θ output by the phase-locked loop (PLL) and obtains the frequency signal based on the angular frequency; it further obtains the frequency change rate df / dt through sliding time window fitting, differential calculation, or filtering differentiation.

[0037] The sampling frequency of the state sensing module is higher than 1kHz, and the calculation time window for the frequency change rate df / dt is no greater than 5ms. The state sensing module can be built into the control circuit of the energy storage converter PCS, or it can be used as an independent control device communicating with the PCS.

[0038] The prediction control unit is communicatively connected to the state perception module and is used to receive at least the frequency change rate df / dt and the current frequency deviation Δf, and to estimate the future short-term frequency evolution state based on the power grid frequency response model, and output the predicted frequency trajectory.

[0039] The prediction control unit has a built-in unscented Kalman filter (UKF) to filter out noise and local disturbances in the acquired signal and to estimate the state of the frequency evolution trajectory within a time window of 100ms to 500ms.

[0040] In other embodiments, the state estimation algorithm may also employ extended Kalman filtering, sliding mode observer, particle filter, or other methods suitable for short-time state estimation of power grids. The Model Predictive Control (MPC) decision unit is communicatively connected to the predictive control unit. It receives the predicted frequency trajectory and, in conjunction with the energy storage SOC, energy storage charge / discharge power limits, power change rate limits, and PCS operating constraints, performs a rolling solution to the energy storage compensation power sequence within the finite prediction time domain. The MPC decision unit uses the predicted frequency trajectory as a time-varying reference quantity for the frequency support objective and constructs the following objective function:

[0041] in, N The total number of steps within the finite prediction time domain. f is the energy storage compensation power for the previous control step. pred (k) is the prediction frequency of the k-th prediction step, f ref The SOC(k) is the rated frequency or dynamic frequency reference value, and the SOC(k) is the energy storage state of charge. ref For energy storage target state of charge, P ess (k) represents the energy storage compensation power, q f q soc and q p These are the weighting coefficients corresponding to frequency deviation, SOC deviation, and power change rate, respectively.

[0042] The MPC optimization process must satisfy at least the following constraints:

[0043]

[0044]

[0045]

[0046] in, For the k-th step battery current, SOC min and SOC max These are the lower and upper limits of energy storage SOC, respectively, P ess,min and P ess,max These represent the lower and upper limits of energy storage charging and discharging power, respectively. max I is the upper limit of the rate of change of energy storage power. bat,max This represents the upper limit of the battery's allowable current.

[0047] The MPC decision unit solves the optimization problem in a rolling manner during each control cycle and sends the first control quantity in the energy storage compensation power sequence to the energy storage converter PCS for execution.

[0048] The PCS control signal input terminal of the energy storage converter is communicatively connected to the output terminal of the MPC decision unit. It is used to receive the first value of the energy storage compensation power command and control the energy storage unit to output or absorb compensation power before the electrical quantity at the PCC point is significantly deviated due to load fluctuations.

[0049] The gas turbine control unit is used to adjust the mechanical power output of the gas turbine according to the system steady-state power demand, low-frequency power deviation, or frequency deviation signal.

[0050] In the initial stage of transient disturbance, the energy storage unit undertakes high-frequency transient power compensation based on the df / dt feedforward prediction results. After the gas turbine governor and mechanical power gradually respond, the gas turbine undertakes the medium- and low-frequency and steady-state power demands, and the energy storage unit gradually reduces the compensation power and restores the SOC. This achieves dynamic decoupling between the gas turbine and the energy storage system based on the response frequency band and time scale.

[0051] The system automatically switches between grid-connected steady-state mode, transient support mode, and SOC over-limit protection mode based on the electrical status of the PCC point, and determines the operating condition: The grid-connected steady-state mode is as follows: When |df / dt|≤ the set threshold of 0.05Hz / s, the power grid is judged to be in steady state, the energy storage is in standby or SOC maintenance state, and the gas turbine undertakes the base load.

[0052] The transient support mode is as follows: When |df / dt|> the set threshold, it is determined that a power imbalance has occurred in the power grid, and steps S1-S4 are immediately triggered. The energy storage outputs compensation power in millisecond-level response.

[0053] The SOC over-limit protection mode is as follows: When the energy storage SOC exceeds the preset safety range, falling below 20% or exceeding 90%, the weight of the SOC constraint term in the MPC optimization function automatically increases, forcing the energy storage to prioritize the SOC recovery strategy to avoid overcharging and over-discharging.

[0054] The communication architecture of the system is as follows: The state perception module, predictive control unit, MPC decision unit and energy storage converter communicate with each other via high-speed real-time industrial Ethernet EtherCAT or high-speed serial bus. The communication cycle does not exceed the MPC control cycle ≤50ms, ensuring that the total delay from signal sampling to instruction execution is less than 30ms.

[0055] Implementation Method 2 This embodiment provides a dynamic decoupling predictive control method for gas-energy storage, using the dynamic decoupling predictive control system for gas-energy storage as described in Embodiment 1. The method includes the following steps: Step 1: Real-time extraction of the feedforward signal, specifically: The state sensing module detects the three-phase voltage in real time at the PCC point at a sampling frequency higher than 1kHz, and calculates and outputs the frequency change rate (df / dt) signal without waiting for a complete power frequency cycle.

[0056] Step 2: Perform situational awareness and optimal prediction, specifically as follows: The predictive control unit uses only the df / dt output from ① and the current Δf as observations, inputs them into the UKF, performs an optimal estimate of the system state, and outputs the predicted frequency trajectory. This step does not rely on any external load forecast data.

[0057] Step 3: Optimal decision and compensation, specifically: The MPC decision unit directly and uniquely uses the predicted frequency trajectory output from step 2 as the time-varying reference trajectory in its cost function. Within each control cycle (10ms-50ms), the optimal energy storage compensation power sequence is solved, and the control cycle of the MPC is less than the effective variation period of the df / dt signal to ensure the timeliness of the control.

[0058] Step 4: Perform advance compensation and dynamic decoupling, specifically as follows: According to the command issued in step 3, the energy storage converter outputs compensation power within 200ms after the power deficit occurs, before the gas turbine governor is activated, to fill the power vacuum period and achieve decoupling of the response time scale between energy storage (high-frequency transient power) and gas turbine (low-frequency steady-state base load).

[0059] The system automatically switches between grid-connected steady-state mode, transient support mode, and SOC over-limit protection mode based on the electrical status of the PCC point, and determines the operating condition: The grid-connected steady-state mode is as follows: When |df / dt|≤ the set threshold of 0.05Hz / s, the power grid is judged to be in steady state, the energy storage is in standby or SOC maintenance state, and the gas turbine undertakes the base load.

[0060] The transient support mode is as follows: When |df / dt|> the set threshold, it is determined that a power imbalance has occurred in the power grid, and steps S1-S4 are immediately triggered. The energy storage outputs compensation power in millisecond-level response.

[0061] The SOC over-limit protection mode is as follows: When the energy storage SOC exceeds the preset safety range, falling below 20% or exceeding 90%, the weight of the SOC constraint term in the MPC optimization function automatically increases, forcing the energy storage to prioritize the SOC recovery strategy to avoid overcharging and over-discharging.

[0062] The communication architecture of the system is as follows: The state perception module, predictive control unit, MPC decision unit and energy storage converter communicate with each other via high-speed real-time industrial Ethernet EtherCAT or high-speed serial bus. The communication cycle does not exceed the MPC control cycle ≤50ms, ensuring that the total delay from signal sampling to instruction execution is less than 30ms.

[0063] Furthermore, step 2 relies on any external load forecast data.

[0064] Implementation Method 3 This embodiment provides a gas-energy storage dynamic decoupling predictive control system as described in Embodiment 1, which is applied to an industrial park microgrid.

[0065] In a microgrid in an industrial park, the gas turbine has a rated power of 2MW, and the energy storage system has a capacity of 1MW / 1MWh. The status sensing module is installed at the low-voltage side incoming cabinet of the 10kV / 400V transformer, which serves as the grid connection common coupling point (PCC).

[0066] During normal system operation, the gas turbine bears the base load of the park, while the energy storage system remains within the target SOC range and is in standby or low-power regulation mode. The state sensing module continuously collects the three-phase voltage signal at the PCC point, obtains the phase angle θ through the PLL, and calculates the frequency f, frequency deviation Δf, and frequency change rate df / dt in real time.

[0067] When the upstream power grid experiences a power deficit due to heavy load or reduced power output, the rate of change of frequency at the PCC point, df / dt, changes first. For example, df / dt changes from near 0 to -0.5 Hz / s. The state-aware module captures this trend within milliseconds and sends df / dt and the current Δf to the predictive control unit.

[0068] The predictive control unit estimates the short-term frequency state using the UKF algorithm and predicts that the frequency may drop within the next 200ms. The MPC decision unit solves the energy storage compensation power sequence in a rolling manner based on the predicted frequency trajectory, energy storage SOC, energy storage power limit, and ramp rate limit, and sends the first control variable to the energy storage converter PCS.

[0069] After receiving the power command, the energy storage converter PCS increases the energy storage discharge power to provide transient power compensation before the gas turbine's mechanical power responds significantly, thereby reducing the frequency drop. After the gas turbine governor activates and gradually increases the mechanical power output, the MPC decision unit reduces the energy storage compensation power, causing the energy storage system to gradually exit transient support and recover to the target charge range according to the SOC state.

[0070] When the system detects that the energy storage SOC is lower than the set lower limit or higher than the set upper limit, the MPC decision unit increases the weight of the SOC recovery target in the optimization function, restricts the energy storage from continuing deep charging and discharging, and prioritizes the execution of the SOC recovery strategy.

[0071] Through the above control process, the energy storage system mainly undertakes transient high-frequency power compensation at the millisecond to hundred-millisecond level, while the gas turbine mainly undertakes medium-low frequency and steady-state power output, thereby realizing dynamic decoupling and coordinated control between the gas turbine and the energy storage system.

Claims

1. A dynamic decoupling predictive control system for gas-energy storage, characterized in that, The system includes a gas turbine power generation unit, an energy storage unit, an energy storage converter PCS, an AC bus, and a grid-connected common coupling point PCC. It also includes a state perception module, a predictive control unit, a model predictive control decision unit, and a gas turbine control unit. The voltage sampling terminal of the state sensing module is connected to the AC bus at the grid common coupling point PCC, and is used to collect the three-phase voltage signal at the PCC point in real time, and extract the grid frequency f, frequency deviation Δf and frequency change rate df / dt based on the three-phase voltage signal. The prediction control unit is communicatively connected to the state perception module and is used to receive at least the frequency change rate df / dt and the current frequency deviation Δf, and to estimate the future short-term frequency evolution state based on the power grid frequency response model, and output the predicted frequency trajectory. The model predictive control decision unit is communicatively connected to the predictive control unit. It is used to receive the predicted frequency trajectory and, in combination with the energy storage state of charge (SOC), energy storage charging and discharging power limits, power change rate limits, and PCS operating constraints, to solve the energy storage compensation power sequence in the finite prediction time domain. The control signal input terminal of the energy storage converter is communicatively connected to the output terminal of the model predictive control decision unit, and is used to receive the first value of the energy storage compensation power command, and control the energy storage unit to output or absorb compensation power before the electrical quantity at the PCC point is significantly deviated due to load fluctuations. The gas turbine control unit is used to adjust the mechanical power output of the gas turbine according to the system steady-state power demand, low-frequency power deviation, or frequency deviation signal. In the early stages of transient disturbances, the energy storage unit undertakes high-frequency transient power compensation based on the df / dt feedforward prediction results. After the gas turbine governor and mechanical power gradually respond, the gas turbine undertakes the medium- and low-frequency and steady-state power demands, and the energy storage unit gradually reduces the compensation power and restores the SOC, thereby realizing the dynamic decoupling of the gas turbine and the energy storage system based on the response frequency band and time scale.

2. The system according to claim 1, characterized in that, The state sensing module calculates the angular frequency using the phase angle output by the phase-locked loop (PLL) and obtains the frequency signal based on the angular frequency; it further obtains the frequency change rate df / dt through sliding time window fitting, differential calculation, or filtering differentiation; the sampling frequency of the state sensing module is higher than 1kHz, and the calculation time window of the frequency change rate df / dt is no greater than 5ms.

3. The system according to claim 1, characterized in that, The prediction control unit has a built-in unscented Kalman filter (UKF) to filter out noise and local disturbances in the acquired signal and to estimate the state of the frequency evolution trajectory within a time window of 100ms to 500ms.

4. The system according to claim 1, characterized in that, The objective function constructed by the model predictive control decision unit is: in, N The total number of steps within the finite prediction time domain. f is the energy storage compensation power for the previous control step. pred (k) is the prediction frequency of the k-th prediction step, f ref The SOC(k) is the rated frequency or dynamic frequency reference value, and the SOC(k) is the energy storage state of charge. ref For the target state of charge of energy storage, P ess (k) represents the energy storage compensation power, q f q soc and q p These are the weighting coefficients corresponding to frequency deviation, SOC deviation, and power change rate, respectively. The MPC optimization process must satisfy at least the following constraints: in, Let SOC be the battery current at step k. min and SOC max These are the lower and upper limits of energy storage SOC, respectively, P ess,min and P ess,max These represent the lower and upper limits of energy storage charging and discharging power, respectively. max This represents the upper limit of the rate of change of energy storage power. This represents the upper limit of the battery's allowable current.

5. The system according to claim 1, characterized in that, The state perception module, predictive control unit, MPC decision unit and energy storage converter communicate with each other via high-speed real-time industrial Ethernet or high-speed serial bus. The communication cycle does not exceed the MPC control cycle, ensuring that the total delay from signal sampling to instruction execution is less than 30ms.

6. The system according to claim 1, characterized in that, The system automatically switches its operating mode based on the electrical status of the PCC point: When |df / dt|≤ the set threshold, the power grid is judged to be in steady state, the energy storage is in standby or SOC maintenance state, and the gas turbine undertakes the base load; When |df / dt|> the set threshold, it is determined that a power imbalance has occurred in the power grid, and the energy storage is immediately triggered to output compensation power in a transient support mode with a millisecond-level response. When the energy storage SOC exceeds the preset safe range, the weight of the SOC recovery target is increased in the MPC optimization function to limit the energy storage from continuing deep charging and discharging, and enter the SOC over-limit protection mode.

7. A dynamic decoupling predictive control method for gas-energy storage, characterized in that, Using the gas-storage dynamic decoupling predictive control system as described in any one of claims 1-6, the method includes the following steps: Step 1: The state sensing module detects the three-phase voltage in real time at the PCC point at a sampling frequency higher than 1kHz, and calculates and outputs the frequency change rate df / dt signal without waiting for a complete power frequency cycle. Step 2: The prediction control unit uses only the df / dt and the current Δf as observation values, inputs them into the unscented Kalman filter (UKF), performs optimal estimation of the system state, and outputs the predicted frequency trajectory; Step 3: The Model Predictive Control Decision Unit directly and uniquely uses the predicted frequency trajectory as the time-varying reference trajectory in its cost function. In each control cycle, it solves the optimal energy storage compensation power sequence, and the control cycle of the MPC is less than the effective variation cycle of the df / dt signal. Step 4: The energy storage converter outputs compensation power before the gas turbine governor operates, according to the issued command, to fill the power vacuum period and achieve decoupling of the response time scale between energy storage and gas turbine.

8. The method according to claim 7, characterized in that, Step 1 involves automatically switching the operating mode based on the electrical status of the PCC point. When |df / dt|≤ the set threshold, the power grid is judged to be in steady state, the energy storage is in standby or SOC maintenance state, and the gas turbine undertakes the base load; When |df / dt|> the set threshold, it is determined that a power imbalance has occurred in the power grid, and the energy storage is immediately triggered to output compensation power in a transient support mode with a millisecond-level response. When the energy storage SOC exceeds the preset safe range, the weight of the SOC recovery target is increased in the MPC optimization function to limit the energy storage from continuing deep charging and discharging, and enter the SOC over-limit protection mode.

9. The method according to claim 7, characterized in that, Step 2 relies on any external load forecast data.

10. A gas-energy storage dynamic decoupling predictive control system as described in any one of claims 1-6, characterized in that, The system is applied to microgrids in industrial parks.

Images