Self-adaptive frequency collaborative supporting method and system for energy storage and new energy grid-connected device under weak power grid

Abstract

The invention discloses a self-adaptive frequency collaborative support method and system for an energy storage and new energy grid-connected device under a weak power grid, and the method comprises the steps: firstly determining the expandable active power, locking the frequency of a grid-connected point of the weak power grid, and inputting the frequency into a neuron model; and then constructing a fuzzy rule reasoning table, defuzzifying to obtain an adaptive neuron proportionality coefficient, determining an adaptive correction amount of a frequency reference value, and calculating a reference value of total active power. Safe operation and equipment protection of the system are considered at the same time, the power adjustment range and the power change rate of the unit are limited, an economic-frequency collaborative optimization objective function is established, a power reference value is obtained through calculation, and then a power instruction is issued; and finally, monitoring the actual output power of the station, and giving out early warning when the deviation exceeds a threshold value. According to the method, the correction amount of the frequency reference value is determined from the perspective of eliminating the frequency steady-state error of the weak power grid, the active power is distributed from the perspective of collaborative support, and the collaboration, economy and reliability of the output of the energy storage device and the new energy station are realized.

Description

Technical Field

[0001] The present invention belongs to the field of new energy grid-connected power generation, and particularly relates to an adaptive frequency collaborative support method for energy storage and new energy grid-connected devices under a weak grid. Background Art

[0002] In recent years, new energy power generation technologies such as photovoltaic and wind power in China have developed rapidly. However, due to constraints such as natural resources and environmental protection, new energy power stations are usually built in remote areas and are connected to the main grid through UHV technology and long-distance transmission lines. This makes the line impedance non-negligible, and a weak connection state appears between the new energy power station and the grid. With the increase in the new energy penetration rate, the grid gradually shows the characteristics of a "weak grid" with "high impedance, low inertia, and low anti-interference", and its frequency is easily fluctuated. The characteristics of "volatility" and "randomness" of new energy output will further exacerbate the frequency instability. Therefore, it is of great significance to study the frequency support technology for weak grids.

[0003] At present, there are two methods for the frequency support of new energy power stations: centralized and decentralized. Decentralized means that each unit independently adjusts to support the frequency of the weak grid. However, the decentralized control of the units is prone to conflicts, resulting in local optima and poor overall effects, and the collaborative effect of the power station cannot be exerted. Centralized means that the power station issues instructions to allocate power, and the constant coefficient averaging method is often used. This allocation method does not consider the actual operating conditions of the units and will cause waste of resources. The existing collaborative support strategies for energy storage and new energy power stations have the following problems:

[0004] 1. The overall control strategy design is complex, the frequency modulation speed is slow, there are many restrictions on energy storage, the dynamic performance is poor, and the flexibility of energy storage being plug-and-play and the synergy of peak shaving and valley filling are not fully utilized;

[0005] 2. The existing energy storage and new energy power stations mostly use primary frequency modulation with a steady-state error, and cannot adaptively adjust the controller parameters according to the actual frequency situation, and then adjust the frequency reference value to eliminate the steady-state error;

[0006] 3. The current weak grid frequency support strategy only considers the regulation of frequency, and does not consider the economy and robustness of the strategy while performing frequency modulation. Summary of the Invention

[0007] In view of the problems existing in the prior art, the present invention provides an adaptive frequency collaborative support method for energy storage and new energy grid-connected devices under a weak grid, which can consider the collaborative support of frequency by new energy power stations with additional energy storage, and achieve the optimal overall frequency modulation effect and reasonable utilization of resources.

[0008] To solve the above technical problems, the present invention provides the following technical solution: An adaptive frequency collaborative support method for energy storage and new energy grid-connected devices under a weak grid, comprising the following steps:

[0009] S1. Measure the wind speed and light intensity of the new energy power station, and evaluate the active power that the energy storage and the new energy power station can output in the short term;

[0010] S2. According to the active power that the energy storage and the new energy power station can output in the short term, use a pre-filter to obtain the fundamental positive-sequence complex vector of the grid-connected point voltage, and input it into a phase-locked device to lock the frequency of the grid-connected point of the weak grid;

[0011] S3. Input the frequency of the grid-connected point of the weak grid in step S2 into a neuron model, then construct a fuzzy rule inference table for various working conditions, defuzzify to obtain an adaptive neuron proportional coefficient, determine the adaptive correction amount of the frequency reference value, and then calculate the reference value of the total active power of the energy storage and the new energy power station through active-frequency control;

[0012] S4. Based on the active power that the energy storage and the new energy power station can output in the short term in step S1 and the reference value of the total active power of the energy storage and the new energy power station in step S3, considering the safe operation of the system and equipment protection, limit the power adjustment range and power change rate of the unit, establish an economic-frequency collaborative optimization objective function, calculate the power reference value of the energy storage and the new energy power station through an intelligent optimization algorithm, and finally send a power command by the microgrid central controller;

[0013] S5. Monitor the actual output active power of the unit in step S4 at a preset time. When the error between the actual output active power of the unit and the power reference value exceeds the threshold, an alarm is issued, and at the same time, the power command sent by the microgrid central controller is adjusted.

[0014] Further, the foregoing includes the following sub-steps:

[0015] S101. According to the wind speed v measured by the wind speed sensor i , evaluate in real time the active power P m output by the current wind turbine in the short term, as follows:

[0016]

[0017] In the formula, P m is the total power generated by the wind turbine, m is the number of wind turbines, P i is the power generated by the i-th wind turbine, ρ is the air density, C p is the wind energy utilization coefficient, S is the swept area of the wind turbine rotor, v i is the wind speed measured by the sensor of the i-th wind turbine;

[0018] S102. According to the light intensity G measured by the light sensor, evaluate in real time the active power P pV output by the current photovoltaic array group in the short term, as follows:

[0019]

[0020] In the formula, P pV is the total power generated by the photovoltaic array group, n is the number of photovoltaic arrays, s is the number of photovoltaic cells in the photovoltaic array, and P pVi is the power generated by the i-th photovoltaic array, U oc is the open-circuit voltage of the photovoltaic cell, A is a constant, G is the light intensity, and I D is the diode current of the photovoltaic cell, and I sh is the series resistance current of the photovoltaic cell;

[0021] S103. Calculate the active power P G that can be output by the energy storage and new energy power station in the short term, as follows:

[0022]

[0023] In the formula, P ES is the output power of the energy storage device, and its value depends on the state of charge of the energy storage.

[0024] Furthermore, the aforementioned step S2 includes the following sub-steps:

[0025] S201. Convert the three-phase voltage at the weak grid connection point in the three-phase stationary coordinate system to the αβ coordinate system through Clark transformation:

[0026] Assume the voltage at the point of common coupling of the weak grid, that is, the PCC point, as follows:

[0027]

[0028] In the formula, U pccm is the voltage amplitude at the PCC point, and θ s is the actual phase angle at the PCC point;

[0029] S202. Convert the voltage in the three-phase stationary coordinate system to the αβ coordinate system through Clark transformation to obtain the α component U pccα and the β component U pccβ of the three-phase voltage at the grid connection point, as follows:

[0030]

[0031] S203. Use the α component of the voltage in the αβ coordinate system as the real axis and the β component as the imaginary axis to construct a voltage complex vector input pre-filter to obtain the fundamental positive sequence component of the voltage, and then perform Park transformation on it to obtain the fundamental positive sequence dq-axis components of the voltage. Multiply the β component in formula (5) by the unit imaginary part j and add the α component to obtain a complex electrical quantity, that is, the voltage complex vector u pccαβ = u pccα + jupccβ , input the voltage complex vector into the pre-filter G P (s) to obtain the fundamental positive-sequence synchronous component u of the voltage pccαβP :

[0032]

[0033] where k s is the damping coefficient and ω n is the rated angular frequency of the system. Pass the fundamental positive-sequence synchronous component u of the voltage pccαβP through the Park transformation to obtain the fundamental positive-sequence dq components of the voltage, and construct the complex vector u pccdqP = u pccdP + ju pccqP ;

[0034] S204. Derive the transfer function of the phase-locked device controller by designing a phase transfer function with fast dynamic response and zero overshoot in phase detection; the closed-loop transfer function G PLL (s) of the phase-locked device is as follows:

[0035]

[0036] where G θ (s) is the transfer function of the controller;

[0037] The transfer function G P (s) between phases after small-signal linearization of the pre-filter G D (s) is as follows:

[0038]

[0039] S205. Based on the transfer function G PLL (s) and the transfer function G D (s), the transfer function H(s) between the input phase and the output phase is obtained as:

[0040]

[0041] S206. According to the closed-loop transfer function G PLL (s) of the phase-locked device in step S204, the transfer function of the phase-locked device controller is obtained as follows:

[0042]

[0043] S207. Design the desired phase transfer function to inversely deduce the transfer function of the phase-locked device controller. The expression of the phase transfer function H(s) is:

[0044]

[0045] where a i , b j , m, and n are constants that satisfy the following constraints:

[0046]

[0047] Adopting the form of n first-order low-pass filters connected in series, the expression of the phase transfer function H(s) is:

[0048]

[0049] where T Hi is the time constant of the i-th filter;

[0050] S208. From the transfer function of the phase-locked device controller in step S206, the transfer function of the phase-locked device controller is obtained as:

[0051]

[0052] Input the fundamental positive-sequence q-axis component of the three-phase voltage into this phase-locked device, detect the current phase angle, and then obtain the current frequency f of the weak grid connection point.

[0053] Furthermore, the aforementioned step S3 includes the following sub-steps:

[0054] S301. Convert the rated frequency f n and the detected instantaneous frequency f(k) into the state quantities x 1 (k), x 2 (k) required by the artificial intelligence algorithm:

[0055]

[0056] where k is the iteration number and Δ is the difference;

[0057] S302. Establish a neuron model in the artificial intelligence algorithm, weighted sum the detected frequency ratio and integral state quantities according to the corresponding weight coefficients, and rollingly correct the weight coefficients to ensure the convergence of the algorithm;

[0058] Among them, the neuron model is as follows:

[0059]

[0060] where u(k) is the neuron output, that is, the correction amount of the frequency reference value, K is the proportional coefficient of the neuron, w(k) is the weight coefficient, η p , η i are the learning rates of proportion and integral respectively;

[0061] Normalize the weight coefficient w(k) as follows:

[0062]

[0063] In the formula, is the normalized weight coefficient;

[0064] Considering that the proportional and integral parameters are highly positively correlated with the deviation e(k) and the difference component Δe(k), simplify Equation (16) as follows:

[0065]

[0066] S303. Construct a fuzzy controller. Select triangular membership functions for both input and output variables, construct a fuzzy rule inference table, and input the proportional and integral state variables of the frequency into the fuzzy controller. Its output is used as the proportional coefficient of the neuron model;

[0067] Determine the adaptive neuron proportional coefficient according to the following logic to determine the fuzzy rule inference table

[0068] When the deviation of the frequency from the rated value is zero, the proportional coefficient of the neuron model remains unchanged;

[0069] When the deviation of the frequency from the rated value is the threshold A - S and the change rate is the threshold B - S, the proportional coefficient of the neuron model is set to the first preset interval;

[0070] When the deviation of the frequency from the rated value is the threshold C - M and the change rate is the threshold D - M, the proportional coefficient of the neuron model is set to the second preset interval;

[0071] When the deviation of the frequency from the rated value is the threshold E - B and the change rate is the threshold F - B, the proportional coefficient of the neuron model is set to the third preset interval. Use the centroid method for defuzzification to determine the adaptive neuron proportional coefficient The adaptive correction amount of the frequency reference value is:

[0072]

[0073] Use the frequency reference value correction amount to achieve bias - free adjustment of the frequency, eliminate the steady - state error existing in primary frequency modulation, and finally make the grid frequency stable at the rated value;

[0074] S304. Input the adaptive correction amount u of the frequency reference value obtained in step S303 into the active - frequency control to obtain the total active power reference value P of the energy storage and new - energy power stations ref , as follows:

[0075]

[0076] Where m 1 is the sag coefficient, and P is the actual power generated by the new energy power station.

[0077] Furthermore, the aforementioned step S4 includes the following sub-steps:

[0078] S401. Establish an economic-frequency coordinated optimization objective function:

[0079] max J = max(r(r)(p m (t) + p pv (t) + p ES (t)) - (R b + R ES ))

[0080] min J = min(R b + R ES + τ(f ref - f( t )) 2 )

[0081] R b = r p (t)λ p + r m (t)λ m + r n (t)λ n (21)

[0082] Where r(t) is the day-ahead electricity price, R b is the day-ahead planned reserve cost, R ES is the energy storage cost, τ is the robustness weight, r p , r m , r n are the day-ahead reserve plan prices, λ p is the positive spinning reserve capacity, λ m is the negative spinning reserve capacity, λ n is the non-spinning reserve capacity;

[0083] S402. Combining step S401, considering the safe operation of the system and equipment protection, taking the power balance constraint of the new energy power station, the generator output constraint, and the capacity of the energy storage device as constraint conditions, and solving the reference values of the active power of the energy storage device and the new energy generator set;

[0084] Power balance constraint: p m (t) + p PV (t) + p ES (t) = p ref (t) (22) Wind power generation output and ramping constraint:

[0085]

[0086] In the formula, P m (t) is the wind power generation power within time period t; H m_p is the allowable upward ramp rate of the wind power generation system; H m_m is the allowable downward ramp rate of the wind power generation system, and both of the two ramp rate limits are positive values;

[0087] Photovoltaic power generation output and ramp constraint:

[0088] 0 < p PV (t) < P PV

[0089] -H PV_m < |p PV (t) - p PV (t - 1) < H PV_p (24) In the formula, P PV (t) is the photovoltaic power generation power within time period t; H PV_p is the allowable upward ramp rate of the photovoltaic power generation system; H PV_m is the allowable downward ramp rate of the photovoltaic power generation system, and both of the two ramp rate limits are positive values.

[0090] Energy storage device capacity constraint:

[0091] p ES (t) = [SOC(t) - SOC 0 P ES

[0092] SOC min < SOC(t) < SOC max (25)

[0093]

[0094] In the formula, SOC(t) is the state of charge of the energy storage device within time period t, SOC 0 is the initial state of charge of the energy storage device, S m is the total capacity of the energy storage device, P ES is the total power of the energy storage device, SOC max is the upper limit of the state of charge of the energy storage device, SOC min is the lower limit of the state of charge of the energy storage device, δ is the charge retention ability, η C is the charging efficiency, less than 1, η F is the discharging efficiency, less than 1;

[0095] S403. Use the intelligent optimization algorithm in MATLAB to solve the active power reference values of the energy storage device and the new energy power station simultaneously, and finally send the power command by the microgrid central controller.

[0096] Further, the specific step S5 is as follows: Monitor the execution of the active power target command at fixed time intervals. If the following equation is satisfied, send a warning to the microgrid central controller to check the execution of the command.

[0097] E 1 = |p m,t - p m，t-τ | ≥ δ 1

[0098] E 2 = |p pV，t - p pV,t-τ ≥ δ 2 (27)

[0099] In the formula, p m,t is the active power output by the wind turbine at time t, p m,t-τ is the active power output by the wind turbine at time t - τ, p pV,t is the active power output by the photovoltaic array at time t, p pV,t-τ is the active power output by the photovoltaic array at time t - τ, δ 1 and δ 2 are the power deviation thresholds. If the deviation exceeds the threshold, a warning will be issued immediately.

[0100] On the other hand, the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of any method described in the present invention are implemented.

[0101] On yet another aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of any method described in the present invention are implemented.

[0102] Compared with the prior art, the beneficial technical effects of the present invention adopting the above technical solutions are as follows:

[0103] 1. Give full play to the fast charge and discharge ability of the energy storage device, with a fast response speed, and can quickly increase power to support the frequency at the initial stage of power shortage.

[0104] 2. Adopt an adaptive frequency reference value correction amount, which can adapt to various conditions of a weak power grid, eliminate the steady-state error of primary frequency modulation at the same time, and finally stabilize the frequency to the rated value.

[0105] 3. While adjusting the frequency, economy and robustness are considered, achieving multi-objective optimization of frequency modulation. Description of the Drawings

[0106] Figure 1 is a schematic diagram of the overall process of the present invention.

[0107] Figure 2 is the membership function graph of the frequency deviation e of the present invention.

[0108] Figure 3 is the membership function graph of the rate of change of frequency deviation Δe of the present invention.

[0109] Figure 4 is the neuron proportionality coefficient of the present invention of the membership function graph.

[0110] Figure 5 is the frequency response curve graph of the present invention.

[0111] Figure 6 is a schematic diagram of the new power output of the station and energy storage of the present invention.

[0112] Figure 7 is the neuron proportionality coefficient of the present invention curve graph. Detailed Embodiment

[0113] To better understand the technical content of the present invention, specific embodiments are given and described in conjunction with the accompanying drawings as follows.

[0114] In the present invention, various aspects of the present invention are described with reference to the drawings, and many illustrative embodiments are shown in the drawings. The embodiments of the present invention are not limited to those described in the drawings. It should be understood that the present invention can be implemented by any one of the various concepts and embodiments introduced above, and the concepts and embodiments described in detail below, because the concepts and embodiments disclosed in the present invention are not limited to any embodiment. In addition, some aspects disclosed in the present invention can be used alone, or in any suitable combination with other aspects disclosed in the present invention.

[0115] This example is based on a new energy station composed of a wind turbine generator set, a photovoltaic array group, and energy storage batteries. The frequency of the weak grid connection point is quickly detected through a pre-filter and a phase-locked device. The parameters of the controller are optimized using a neuron model and a fuzzy rule inference table to obtain the total frequency reference value and the required power adjustment amount. Considering the power constraints of the unit and the energy storage, an economic-frequency collaborative optimization objective function is established to reasonably distribute the active power. The power command is issued by the microgrid central controller, and finally, the actual active power output of the unit is monitored at fixed time intervals, and an alarm is issued when the deviation from the reference frequency exceeds the threshold.

[0116] Reference Figure 1, the present invention discloses an adaptive frequency collaborative support method for an energy storage and new energy grid connection device under a weak power grid, including the following steps:

[0117] S1. Measure the real-time wind speed through a high-precision wind speed sensor to calculate the total available active power capacity of the wind turbine generator set, and measure the light intensity through the photovoltaic modules connected in series with optimizers to calculate the total available active power capacity of the photovoltaic power station, and then obtain the total available active power capacity of the energy storage and new energy power station. Step S1 includes the following sub-steps:

[0118] S101. According to the wind speed v sensed by each fan blade measured by the high-precision wind speed sensor i , evaluate the active power P output by the current wind turbine generator set in the short term in real time m , as follows:

[0119]

[0120] In the formula, P m is the total power generated by the wind turbine generator set, m is the number of wind turbines, P i is the power generated by the i-th wind turbine, ρ is the air density, C p is the wind energy utilization coefficient, S is the swept area of the wind turbine rotor, v i is the wind speed measured by the sensor of the i-th wind turbine; affected by the wake effect, although the wind turbines are in the same environment, the wind speeds sensed by the fan blades at different positions are different.

[0121] S102. According to the light intensity G measured by the photovoltaic modules connected in series with optimizers, evaluate the active power P output by the current photovoltaic array group in the short term in real time pV , as follows:

[0122]

[0123] In the formula, P pV is the total power generated by the photovoltaic array group, n is the number of photovoltaic arrays, s is the number of photovoltaic cells in the photovoltaic array, P pVi is the power generated by the i-th photovoltaic array, U oc is the open-circuit voltage of the photovoltaic cell, A is a constant, G is the light intensity, I D is the diode current of the photovoltaic cell, I sh is the series resistance current of the photovoltaic cell;

[0124] S103. Since the light intensity received by the photovoltaic arrays in the same environment is the same, the power generated by the photovoltaic array group is the sum of the powers generated by each photovoltaic array. Therefore, calculate the active power P that can be output by the energy storage and new energy power station in the short term G , as follows:

[0125]

[0126] In the formula, P ES is the output power of the energy storage device, and its value depends on the state of charge of the energy storage.

[0127] S2. According to the active power that can be output by the energy storage and the new energy power station in the short term, use a pre-filter to obtain the fundamental positive-sequence complex vector of the grid connection point voltage, and input it into the phase-locked device to lock the frequency of the grid connection point of the weak grid; Step S2 includes the following sub-steps:

[0128] S201. Convert the three-phase voltage of the grid connection point of the weak grid in the three-phase stationary coordinate system to the αβ coordinate system through Clark transformation:

[0129] Let the voltage at the point of common coupling of the weak grid, that is, the PCC point, be as follows:

[0130]

[0131] In the formula, U pccm is the voltage amplitude at the PCC point, and θ s is the actual phase angle at the PCC point;

[0132] S202. Convert the voltage in the three-phase stationary coordinate system to the αβ coordinate system through Clark transformation to obtain the α component U pccα and the β component U pccβ , as follows:

[0133]

[0134] S203. Take the α component of the voltage in the αβ coordinate system as the real axis and the β component as the imaginary axis to construct a voltage complex vector and input it into the pre-filter to obtain the fundamental positive-sequence component of the voltage. Then perform Park transformation on it to obtain the fundamental positive-sequence dq-axis components of the voltage. Multiply the β component in formula (5) by the unit imaginary part j and add the α component to obtain a complex electrical quantity, that is, the voltage complex vector u pccαβ = u pccα + ju pccβ . In order to improve the accuracy and dynamic performance of phase and frequency detection and suppress harmonic interference, input the voltage complex vector into the pre-filter G P (s) to obtain the fundamental positive-sequence synchronous component u pccαβP of the voltage:

[0135]

[0136] In the formula, k s is the damping coefficient, and ω n is the rated angular frequency of the system. The fundamental positive-sequence synchronous component u pccαβP of the voltage is obtained through Park transformation to obtain the fundamental positive-sequence dq components of the voltage, and a complex vector upccdqP = u pccdP + ju pccqP ;

[0137] S204. Derive the transfer function of the phase-locked device controller by designing a phase transfer function with fast dynamic response and zero overshoot in phase detection; the closed-loop transfer function G PLL (s) is as follows:

[0138]

[0139] In the formula, G θ (s) is the transfer function of the controller;

[0140] The pre-filter G P (s) after small-signal linearization gives the transfer function G D (s) between phases as follows:

[0141]

[0142] S205. Based on the transfer function G PLL (s) and the transfer function G D (s), the transfer function H(s) between the input phase and the output phase is obtained as:

[0143]

[0144] S206. According to the closed-loop transfer function G PLL (s) of the phase-locked device in step S204, the transfer function of the phase-locked device controller is obtained as follows:

[0145]

[0146] S207. Design the desired phase transfer function to inversely deduce the transfer function of the phase-locked device controller. The expression of the phase transfer function H(s) is:

[0147]

[0148] In the formula, a i , b j , m, and n are constants that satisfy the following constraints:

[0149]

[0150] In the form of n first-order low-pass filters connected in series, the expression of the phase transfer function H(s) is:

[0151]

[0152] In the formula, T Hiis the time constant of the i-th filter;

[0153] S208. From the transfer function of the phase-locked device controller in step S206, the transfer function of the phase-locked device controller is obtained as:

[0154] Input the fundamental positive-sequence q-axis component of the three-phase voltage into the phase-locked device, detect the current phase angle, and then obtain the current frequency f of the weak grid connection point.

[0155] S3. Input the frequency of the weak grid connection point in step S2 into the neuron model, then construct a fuzzy rule inference table for various working conditions, defuzzify to obtain an adaptive neuron proportional coefficient, determine the adaptive correction amount of the frequency reference value, and then calculate the reference value of the total active power of the energy storage and new energy power station through active-frequency control;

[0156] Step S3 includes the following sub-steps:

[0157] S301. Convert the rated frequency f n and the detected instantaneous frequency f(k) into the state quantities x 1 (k), x 2 (k) of the frequency ratio and integral of the artificial intelligence algorithm:

[0158]

[0159] In the formula, k is the iteration number, and Δ is the difference;

[0160] S302. Establish a neuron model in the artificial intelligence algorithm, weighted sum the detected frequency ratio and integral state quantities according to the corresponding weight coefficients, and roll-correct the weight coefficients to ensure the convergence of the algorithm;

[0161] To overcome the problems of divergence and unclear goals in traditional Hebb learning, a supervised Hebb learning rule is adopted, and a neuron model is established accordingly. The neuron model is as follows:

[0162]

[0163] In the formula, u(k) is the neuron output, that is, the correction amount of the frequency reference value, K is the neuron proportional coefficient, w(k) is the weight coefficient, and η p and η i are the learning rates of the ratio and integral respectively;

[0164] To ensure the convergence of the algorithm, the weight coefficient w(k) is normalized as follows:

[0165]

[0166] In the formula, is the normalized weight coefficient;

[0167] Considering that the proportional and integral parameters are highly positively correlated with the deviation e(k) and the difference component Δe(k), Equation (16) is simplified as follows:

[0168]

[0169] S303. Construct a fuzzy controller. The triangular membership function is selected for both the input and output variables. Construct a fuzzy rule inference table. Input the state variables of the proportional and integral of the frequency into the fuzzy controller, and its output is used as the proportional coefficient of the neuron model;

[0170] Since the frequency of a weak power grid is vulnerable to disturbances and the working conditions are complex, in order to achieve parameter adaptive adjustment for different working conditions, the proportional coefficient of the neuron is obtained by fuzzy inference Specifically as follows:

[0171] The proportional coefficient K of the neuron can directly affect the values of the proportional and integral parameters, and thus indirectly affect the control effect of the controller. In addition, the proportional coefficient K determines the response intensity of the controller to the error signal. A larger K can enhance the response speed of the system, but may cause overshoot and instability; a smaller K can reduce the system fluctuations, but may lead to slow response. The proportional coefficient K affects the amplitude of weight update and the sensitivity of network output. By adjusting K, the convergence speed and learning accuracy of the network can be improved.

[0172] Therefore, it is considered to adaptively adjust the proportional coefficient of the neuron by using fuzzy control according to the relationship between the frequency deviation and the frequency change rate to obtain Select the triangular membership function, such as Figure 2 , Figure 3 , Figure 4 as shown.

[0173] Determine the fuzzy rule inference table according to the following logic:

[0174] When the frequency deviation from the rated value is zero, the proportional coefficient of the neuron model remains unchanged;

[0175] When the frequency deviation from the rated value is very small, at the threshold A - S, and the change rate is also very small, at the threshold B - S, the proportional coefficient of the neuron model is set very small to the first preset interval;

[0176] When the frequency deviation from the rated value is moderate, at the threshold C - M, and the change rate is also moderate, at the threshold D - M, the proportional coefficient of the neuron model is moderate and set to the second preset interval;

[0177] When the frequency deviates significantly from the rated value, reaching the threshold E - B, and the rate of change is also very large, reaching the threshold F - B, the proportional coefficient of the neuron model is set to the third preset interval. The membership degree relationship of more specific variables is shown in the fuzzy rule inference table in Table 1. The centroid method is used for defuzzification to determine the adaptive neuron proportional coefficient

[0178] Table 1

[0179]

[0180] Note: Among them, NB, NM, NS, Z, PS, PM, PB represent Negative Big, Negative Medium, Negative Small, Zero, Positive Small, Positive Medium, Positive Big respectively.

[0181] The adaptive correction amount of the frequency reference value is:

[0182]

[0183] Using the new frequency reference value correction amount can achieve bias - free adjustment of the frequency, eliminate the steady - state error problem existing in primary frequency modulation, and finally make the power grid frequency stable at the rated value.

[0184] S304. Input the adaptive correction amount u of the frequency reference value obtained in step S303 into the active - frequency control to obtain the total active power reference value P of the energy storage and new - energy power station ref , as follows:

[0185]

[0186] In the formula, m 1 is the droop coefficient, and P is the actual power generated by the new - energy power station.

[0187] S4. Based on the short - term available active power of the energy storage and new - energy power station in step S1 and the total active power reference value of the energy storage and new - energy power station in step S3, considering the safe operation of the system and equipment protection, limit the power adjustment range and power change rate of the unit, establish an economic - frequency coordinated optimization objective function, calculate the power reference value of the energy storage and new - energy power station through an intelligent optimization algorithm, and finally issue a power command by the micro - grid central controller; specifically, it includes the following sub - steps:

[0188] S401. When allocating the active power reference value of the power station, considering both achieving frequency stability and combining with the day - ahead electricity price to achieve high total system economic benefits, establish an economic - frequency coordinated optimization objective function:

[0189] max J=max(r(t)(p m (t)+p pv (t)+p ES(t))-(R b +R ES ))

[0190] min J=min(R b +R ES +τ(f ref -f(t)) 2 )

[0191] R b =r p (t)λ p +r m (t)λ m +r n (t)λ n (21)

[0192] In the formula, r(t) is the day-ahead electricity price, R b is the day-ahead planned reserve cost, R ES is the energy storage cost, τ is the robustness weight, r p 、r m 、r n are the day-ahead reserve plan prices, λ p is the positive spinning reserve capacity, λ m is the negative spinning reserve capacity, λ n is the non-spinning reserve capacity;

[0193] S402. Combining with step S401, considering the safe operation of the system and equipment protection, taking the power balance constraint, the generator output constraint of the new energy power station, and the capacity of the energy storage device as constraint conditions, solve the reference values of the active power of the energy storage device and the new energy generator set;

[0194] Power balance constraint: p m (t)+p PV (t)+p ES (t)=p ref (t) (22)

[0195] Wind power generation output and ramping constraint:

[0196]

[0197] In the formula, P m (t) is the wind power generation power within period t; H m_p is the allowable upward ramping rate of the wind power generation system; H m_m is the allowable downward ramping rate of the wind power generation system, and both ramping rate limits are positive values;

[0198] Photovoltaic power generation output and ramping constraint:

[0199] 0<p PV(t) < PV

[0200] -H PV_m <|p PV (t)-p PV (t - 1) < H PV_p (24)

[0201] Wherein, P PV (t) is the photovoltaic power generation power within time period t; H PV_p is the allowable upward ramp rate of the photovoltaic power generation system; H PV_m is the allowable downward ramp rate of the photovoltaic power generation system, and both ramp rate limits are positive values.

[0202] For the mathematical model of the energy storage device, in order to make full use of its ability to quickly absorb and emit power and achieve the functions of plug-and-play and peak shaving and valley filling, the following ideas are considered for modeling: when the power generated by wind power and photovoltaics is greater than the total power required for frequency support, the energy storage absorbs the excess power until the energy storage device is full and reaches the upper limit of the battery charge; when the power generated by wind power and photovoltaics is less than the total power required for frequency support, the power deficit is borne by the energy storage, and the energy storage discharges to support the power until it drops to the lower limit of the device's battery charge; when the power generated by wind power and photovoltaics is equal to the total power required for frequency support, the energy storage neither absorbs nor emits power.

[0203] Energy storage device capacity constraint:

[0204] p ES (t) = [SOC(t) - SOC 0 R ES

[0205] SOC min <SOC(t) < SOC max (25)

[0206]

[0207] Wherein, SOC(t) is the state of charge of the energy storage device within time period t, SOC 0 is the initial state of charge of the energy storage device, S m is the total capacity of the energy storage device, P ES is the total power of the energy storage device, SOC max is the upper limit of the state of charge of the energy storage device, SOC min is the lower limit of the state of charge of the energy storage device, δ is the charge retention ability, η C is the charging efficiency, less than 1, η F is the discharging efficiency, less than 1.

[0208] S403. Use the intelligent optimization algorithm in MATLAB to jointly solve for the reference active power values of the energy storage device and the new energy power station, and finally send the power command by the microgrid central controller.

[0209] S5. Monitor the actual output active power of the unit in step S4 according to the preset time. When the error between the actual output active power of the unit and the power reference value exceeds the threshold, an alarm is issued, and at the same time, the power command sent by the microgrid central controller is adjusted. Specifically as follows: Monitor the execution of the active power target command at fixed time intervals. If the following equation is satisfied, a warning is sent to the microgrid central controller to check the execution of the command.

[0210] E 1 =|p m，t -p m,t -τ≥δ 1

[0211] E 2 =|p pV,t -p pV,t-τ ≥δ 2 (27)

[0212] In the formula, p m,t is the active power output by the wind turbine at time t, p m,t-τ is the active power output by the wind turbine at time t - τ, p pV,t is the active power output by the photovoltaic array at time t, p pV,t-τ is the active power output by the photovoltaic array at time t - τ, δ 1 and δ 2 are the power deviation thresholds. If the deviation exceeds the threshold, a warning is immediately issued.

[0213] The following further illustrates this method with the results of specific embodiments. Table 2: Key simulation parameters of the new energy power station with energy storage gives the key simulation parameters of the new energy power station with energy storage:

[0214] Table 2

[0215]

[0216] Figure 5 Shows the frequency response curves of the comparison between this method and the traditional primary frequency modulation method under the condition that the power grid has a power deficit of 7.8 MW. It can be seen that the traditional method has a deeper frequency drop, a slower response rate, and finally cannot stabilize to the rated value, with a steady-state error, which is not conducive to the normal operation of the system; this method has a fast response rate, a small maximum frequency difference, and the frequency can stabilize to the rated value. Figure 6It shows the new power output of the station and energy storage. It can be seen that at the initial stage of power deficit, the energy storage quickly generates power to support the frequency. As the frequency gradually recovers, the output gradually decreases, and the power output of the wind turbines and photovoltaic power stations gradually increases. Figure 7 It shows the neuron proportionality coefficient of the change curve. It can be seen that with the change of frequency, the neuron proportionality coefficient is adaptively adjusted to effectively support the frequency.

[0217] On the other hand, the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of any one of the methods of the present invention are implemented.

[0218] On yet another aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods of the present invention are implemented.

[0219] Although the present invention has been described above with preferred embodiments, it is not intended to limit the present invention. Those with ordinary knowledge in the technical field to which the present invention pertains can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to what is defined by the claims.

Claims

1. An adaptive frequency collaborative support method for energy storage and new energy grid-connected devices under weak power grid, characterized in that: The following steps are involved: S1. Measure the wind speed and light intensity of the new energy station, and evaluate the short-term output active power of the energy storage and new energy station; S2. According to the short-term output active power of the energy storage and new energy stations, the positive sequence complex vector of the fundamental voltage at the grid connection point is obtained by using the pre-filter, and the frequency of the weak grid connection point is locked by the phase-locking device; S3, input the frequency of the weak grid connection point in step S2 into the neuron model, then construct a fuzzy rule inference table for various working conditions, defuzzify to obtain an adaptive neuron proportional coefficient, determine the adaptive correction amount of the frequency reference value, and then calculate the reference value of the total active power of the energy storage and new energy station through active-frequency control; S4, based on the short-term output active power of the energy storage and new energy stations in step S1, and the reference value of the total active power of the energy storage and new energy stations in step S3, considering the safe operation of the system and equipment protection, limiting the power adjustment range and power change rate of the unit, establishing the economic-frequency collaborative optimization objective function, calculating the power reference value of the energy storage and new energy stations through the intelligent optimization algorithm, and finally issuing the power command by the microgrid central controller; S5. Monitor the actual output active power of the unit in step S4 according to the preset time. When the error between the actual output active power of the unit and the power reference value exceeds the threshold, an alarm is issued and the microgrid central controller is adjusted to issue a power command.

2. The adaptive frequency collaborative support method for energy storage and new energy grid-connected devices under weak power grid according to claim 1 is characterized in that: Step S1 includes the following sub-steps: S101, wind speed v measured by wind speed sensor i , real-time evaluation of the active power P output by the current wind turbine in the short term m , as follows: Where P m is the total power generated by the wind turbine, m is the number of wind turbines, P i is the power generated by the i-th fan, ρ is the air density, C p is the wind energy utilization coefficient, S is the wind turbine rotor swept area, v i is the wind speed measured by the sensor of the i-th wind turbine; S102, based on the light intensity G measured by the light sensor, real-time evaluation of the active power P output by the current photovoltaic array group in the short term pV , as follows: Where P pV is the total power generated by the photovoltaic array group, n is the number of photovoltaic arrays, s is the number of photovoltaic cells in the photovoltaic array, P pVi is the power generated by the ith photovoltaic array, U oc is the open circuit voltage of the photovoltaic cell, A is a constant, G is the light intensity, I D is the photovoltaic cell diode current, I sh is the series resistance current of the photovoltaic cell; S103. Calculate the short-term output active power P of energy storage and new energy stations G , as follows: Where P ES The output power of the energy storage device depends on the state of charge of the energy storage.

3. The adaptive frequency collaborative support method for energy storage and new energy grid-connected devices under weak power grid according to claim 1 is characterized in that: Step S2 includes the following sub-steps: S201. Convert the three-phase voltage of the weak power grid connection point in the three-phase stationary coordinate system to the αβ coordinate system through Clark transformation: Assume that the voltage of the weak power grid common connection point, i.e., PCC point, is as follows: Where U pccm is the voltage amplitude at the PCC point, θ s is the actual phase angle of the PCC point; S202, convert the voltage in the three-phase stationary coordinate system to the αβ coordinate system through Clark transformation, and obtain the α component U of the three-phase voltage at the grid connection point pccα and the β component U pccβ , as follows: S203, take the α component of the voltage in the αβ coordinate system as the real axis and the β component as the imaginary axis, construct a voltage complex vector input prefilter, obtain the voltage fundamental positive sequence component, and then perform Park transformation on it to obtain the voltage fundamental positive sequence dq axis component, multiply the β component in formula (5) by the unit imaginary part j and add the α component to obtain the complex electrical quantity, that is, the voltage complex vector u pccαβ =u pccα +ju pccβ , input the voltage complex vector into the pre-filter G P (s), and obtain the fundamental positive sequence synchronous component u of the voltage pccαβP : In the formula, k s is the damping coefficient, ω n is the rated angular frequency of the system. pccαβP The fundamental positive sequence dq component of the voltage is obtained by Park transformation, and the complex vector u is constructed pccdqP =u pccdP +ju pccqP ; S204, by designing a phase transfer function of a fast dynamic response of phase detection zero overshoot, deriving a transfer function of a phase-locked device controller; a closed-loop transfer function G of the phase-locked device PLL (s) is as follows: In the formula, G θ (s) is the transfer function of the controller; Prefilter G P (s) After the small signal is linearized, the phase transfer function G is obtained D (s) is as follows: S205, based on the transfer function G PLL (s) and the transfer function G D (s), the transfer function H(s) between the input phase and the output phase is obtained as: S206: According to the closed-loop transfer function G of the phase-locked device in step S204 PLL (s), the transfer function of the phase-locked device controller is as follows: S207, design the desired phase transfer function to reversely deduce the transfer function of the phase-locked device controller. The expression of the phase transfer function H(s) is: In the formula, a i , b j , m, n are constants, satisfying the following constraints: Using n first-order low-pass filters in series, the expression of the phase transfer function H(s) is: Where, T Hi is the time constant of the i-th filter; S208, from the transfer function of the phase-locked device controller in step S206, the transfer function of the phase-locked device controller is obtained as follows: The fundamental positive sequence q-axis component of the three-phase voltage is input into the phase-locked device, the current phase angle is detected, and then the current frequency f of the weak power grid connection point is obtained.

4. The adaptive frequency collaborative support method for energy storage and new energy grid-connected devices under weak power grid according to claim 1 is characterized in that: Step S3 includes the following sub-steps: S301, set the rated frequency f n The detected instantaneous frequency f(k) is converted into the frequency proportion and integral state quantities x1(k) and x2(k) required by the artificial intelligence algorithm: In the formula, k is the number of iterations, Δ is the difference; S302, establish a neuron model in the artificial intelligence algorithm, weight the detected frequency ratio and the integral state quantity according to the corresponding weight coefficient, and roll-correct the weight coefficient to ensure the convergence of the algorithm; Among them, the neuron model is as follows: Where u(k) is the neuron output, i.e., the correction value of the frequency reference value, K is the proportional coefficient of the neuron, w(k) is the weight coefficient, η p , η i are the learning rates for proportion and integration respectively; The weight coefficient w(k) is normalized as follows: In the formula, is the normalized weight coefficient; Considering that the proportional and integral parameters are positively correlated with the deviation e(k) and the difference Δe(k), equation (16) is simplified as follows: S303, constructing a fuzzy controller, selecting triangular membership functions for both input and output variables, constructing a fuzzy rule inference table, inputting the frequency proportion and integral state quantity into the fuzzy controller, and using its output quantity as the proportional coefficient of the neuron model; Determine the fuzzy rule inference table according to the following logic to determine the adaptive neuron proportional coefficient When the frequency deviates from the rated value of zero, the proportionality coefficient of the neuron model remains unchanged; When the frequency deviation from the rated value is the threshold value AS, and the rate of change is the threshold value BS, the proportional coefficient of the neuron model is set to the first preset interval; When the frequency deviates from the rated value by a threshold value CM and the rate of change is a threshold value DM, the proportional coefficient of the neuron model is set to a second preset interval; When the frequency deviation from the rated value is the threshold EB, and the rate of change is the threshold FB, the proportional coefficient of the neuron model is set to the third preset interval, and the centroid method is used for defuzzification to determine the adaptive neuron proportional coefficient. The adaptive correction value of the frequency reference value is: The frequency reference value correction is used to achieve non-biased frequency adjustment, eliminate the steady-state error existing in primary frequency regulation, and ultimately stabilize the grid frequency to the rated value; S304: Input the adaptive correction value u of the frequency reference value obtained in step S303 into the active power-frequency control to obtain the total active power reference value P of the energy storage and new energy station. ref , as follows: Where m1 is the droop coefficient, and P is the actual power generated by the new energy station.

5. The adaptive frequency collaborative support method for energy storage and new energy grid-connected devices under weak power grid according to claim 1 is characterized in that: Step S4 includes the following sub-steps: S401. Establishing the economic-frequency collaborative optimization objective function: max J=max(r(t)(p m (t)+p pv (t)+p ES (t))-(R b +R ES )) min J′=min(R b +R ES +τ(f ref -f(t) 2 ) R b =r p (t)λ p +r m (t)λ m +r n (t)λ n (21) Where r(t) is the day-ahead electricity price, R b is the day-ahead planning reserve cost, R ES is the energy storage cost, τ is the robustness weight, r p 、r m 、r n is the day-ahead reserve plan price, λ p is the positive spinning reserve capacity, λ m is the negative spinning reserve capacity, λ n is non-rotating reserve capacity; S402, combined with step S401, considering the safe operation of the system and equipment protection, taking the power balance constraint of the new energy station, the unit output constraint, and the capacity of the energy storage device as constraints, solving the active power reference value of the energy storage device and the new energy unit; Power balance constraint: p m (t)+p PV (t)+p ES (t) = p ref (t) (22) Wind power output and climbing constraints: Where P m (t) is the wind power generation power in time period t; H m_p H is the allowable climbing rate of the wind power generation system; m_m is the lower ramp rate allowed by the wind power generation system, and both ramp rate limits are positive values; Photovoltaic power generation output and climbing constraints: 0<p PV (t)<P PV -H PV_m <|p PV (t)-p PV (t-1)|<H PV_p (24) Where P PV (t) is the photovoltaic power generation power in time period t; H PV_p H is the allowable climbing rate of the photovoltaic power generation system; PV_m is the lower ramp rate allowed for the photovoltaic power generation system, and both ramp rate limits are positive values. Energy storage device capacity constraints: p ES (t)=[SOC(t)-SOC0]P ES SOC mim <SOC(t)<SOC max (25) In the formula, SOC(t) is the state of charge of the energy storage device in time period t, SOC0 is the initial state of charge of the energy storage device, S m is the total capacity of the energy storage device, P ES is the total power of the energy storage device, SOC max SOC is the upper limit of the state of charge of the energy storage device. min is the lower limit of the charge state of the energy storage device, δ is the charge retention capacity, η C is the charging efficiency, less than 1, η F is the discharge efficiency, which is less than 1; S403, using the intelligent optimization algorithm in MATLAB to solve the active power parameters of the energy storage device and the new energy station The power command is finally issued by the microgrid central controller.

6. The adaptive frequency collaborative support method for energy storage and new energy grid-connected devices under weak power grid according to claim 1 is characterized in that: Step S5 specifically includes: monitoring the execution of the active target instruction at fixed time intervals. If the following equation is satisfied, an early warning is issued to the microgrid central controller to check the execution of the instruction. E1=|p m，t -p m，t-τ |≥δ1 E2=|p pV,t -p pV，t-τ |≥δ2 (27) In the formula, p m,t is the active power output of the wind turbine at time t, p m,t-τ is the active power output of the wind turbine at time t-τ, p pV,t is the active power output by the photovoltaic array at time t, p pV,t-τ is the active power output by the photovoltaic array group at time t-τ, δ1 and δ2 are the power deviation thresholds, and an early warning will be issued immediately if the deviation exceeds the threshold.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images