Multi-scale adaptive wind power stabilizing method considering hybrid energy storage capacity

Abstract

The invention relates to the technical field of hybrid energy storage control, in particular to a multi-scale self-adaptive wind power stabilizing method considering hybrid energy storage capacity, which comprises the following steps: firstly, establishing a two-dimensional fuzzy control system, and dynamically adjusting the maximum allowable volatility according to the state of charge (SOC) of hybrid energy storage and the volatility of wind power; then, a sliding filtering algorithm based on a Fibonacci sequence is provided and is used for stabilizing wind power. On the basis, the number of decomposition layers and penalty factors in the VMD algorithm are optimized through INGO, optimized parameters are substituted into the VMD algorithm to decompose the power, needing to be stabilized, of the hybrid energy storage system HESS, and power distribution of the HESS is completed. And finally, through case analysis, the method not only can adaptively stabilize the wind power fluctuation, but also can effectively reduce the energy storage burden due to the optimization of the charge and discharge range of the HESS, thereby realizing the power distribution of the HESS, reducing the mode aliasing, avoiding the overcharge or overdischarge phenomenon of the HESS, and ensuring the safe and stable operation of the HESS.

Description

Technical Field

[0001] The present invention relates to the field of hybrid energy storage control technology, and in particular to a multi-scale adaptive wind power leveling method considering hybrid energy storage capacity. Background Art

[0002] As the global demand for renewable energy continues to rise, wind energy, as a typical green energy, is gradually becoming a core component of the modern energy structure. In recent years, thanks to the continuous innovation of wind power technology and the significant reduction in costs, the large-scale development and widespread application of wind energy have become realistic. However, the random volatility and intermittent characteristics of wind energy have brought many challenges to the stability of the power system. The output power of wind power is significantly affected by changes in wind speed. Its volatility and uncertainty pose a severe test to the dispatching and control of the power grid under high penetration scenarios, which is not conducive to wind power friendly grid-connected power generation.

[0003] The continuous development of energy storage technology has led to its promotion and application in the suppression of wind power fluctuations. As a power-type energy storage element, supercapacitors have the characteristics of high power density, low energy density, large charge and discharge rate, and wide working stability, which can effectively smooth high-frequency power fluctuations. Batteries, as a typical representative of energy-type energy storage devices, are characterized by high energy density and low power density, and are suitable for the smooth output of low-frequency power fluctuations. HESS combines power-type and energy-type energy storage elements to achieve complementary advantages in characteristics. Reasonable allocation of the power requirements of these two energy storage elements is the key to giving full play to the advantages of HESS.

[0004] Therefore, in order to solve the problem of wind power fluctuation, a method for HESS to stabilize wind power is needed. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and propose a multi-scale adaptive wind power smoothing method considering hybrid energy storage capacity, using HESS to smooth wind power fluctuations, improve the stability of the power grid, and promote the development of wind power technology.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A multi-scale adaptive wind power smoothing method considering hybrid energy storage capacity, including smoothing wind power fluctuations by a hybrid energy storage system and power allocation of the hybrid energy storage system;

[0008] The hybrid energy storage system adopts a multi-scale adaptive sliding filter (MSAMF) algorithm to smooth the wind power fluctuation. The hybrid energy storage system smoothes the wind power fluctuation according to the state of charge (SOC) of the hybrid energy storage and the fluctuation rate of wind power, establishes a two-dimensional fuzzy control system to dynamically adjust the maximum allowable fluctuation rate, and adopts a multi-scale adaptive sliding filter algorithm to smooth the wind power plan power according to the wind farm grid connection standard to obtain the grid-connected power that meets the grid connection standard and the power that the hybrid energy storage system needs to smooth.

[0009] The power distribution of the hybrid energy storage system adopts the improved goshawk optimization (INGO) algorithm to optimize the variational mode decomposition (VMD) algorithm. On the basis of the goshawk optimization (NGO) algorithm, segmented mapping is considered to improve the algorithm search ability. In the optimization process, adaptive weights and improved Levy flight strategy are introduced to improve the convergence speed of the NGO algorithm. Then, INGO is used to optimize the decomposition layer number K and penalty factor C in the VMD algorithm. The optimized parameters are substituted into the VMD algorithm to decompose the power that needs to be smoothed by the hybrid energy storage system, and the power distribution of the hybrid energy storage system HESS is completed.

[0010] Preferably, the hybrid energy storage system smoothes wind power fluctuations including the following steps:

[0011] Step 1: Read the planned wind power P at time t W (t);

[0012] Step 2: Initialize window value N=1;

[0013] Step 3: Calculate the maximum fluctuation power at 1 minute scale and the maximum fluctuation power at 10 minutes scale;

[0014] Step 4: Determine whether the maximum fluctuation rate under the two time scales in step 3 meets the grid connection standard. If it meets the grid connection standard, go to step 12; if it does not meet the grid connection standard, go to step 5;

[0015] Step 5: Calculate the difference ΔP(t) between the planned wind power at time t and the grid-connected power at time t-1;

[0016] ΔP(t)=P w (t)-P g (t-1) (1)

[0017] Among them, P w (t) is the planned wind power at time t, P g (t-1) is the grid-connected power at time t-1, and ΔP(t) is the difference between the planned wind power at time t and the grid-connected power at time t-1.

[0018] Step 6: According to the size of ΔP(t), determine whether the power fluctuation at this time is in the smoothing stage. If ΔP(t)≤k, go to step 7, otherwise go to step 8; where k is the dividing point between power fluctuation smoothness and power fluctuation mutation;

[0019] Step 7: At this time, the power fluctuation is relatively smooth, and the maximum allowable fluctuation rate α for 1 minute is set as the initial value;

[0020] Step 8: At this time, the power fluctuation is in the mutation stage. Read the current SOC state and ΔP(t) of the HESS, and establish a two-dimensional fuzzy control system. The input quantities are the normalized SOC value ω SOC (t) and ΔP(t) value η ΔP (t), the normalization method is as follows:

[0021]

[0022]

[0023] In the normalization process, if ΔP(t)>4k, let ΔP(t)=4k, similarly if ΔP(t)<4k, let ΔP(t)=-4k;

[0024] Among them, ω SOC The domain of (t) is [0,1], and the fuzzy subsets are {VS (extremely small), S (small), M (medium), B (large), VB (maximum)}; η ΔP The domain of (t) is [-1,1], and the fuzzy subsets are {NB (negative large), NM (negative medium), NS (negative small), Z (zero), PS (positive small), PM (positive medium), PB (positive large)}; the output is the maximum allowable volatility α for 1 minute;

[0025] Step 9: Add 1 to the window value N;

[0026] Step 10: Calculate the grid-connected power at time t using the MSAMF algorithm based on the Fibonacci sequence according to the window value N, the planned wind power at time t, and the grid-connected power at time t-1, t-2, …, t-N+1, as follows:

[0027]

[0028] Among them, a n is the Fibonacci sequence, n is the number of terms in the sequence, and N is the window value;

[0029] Then calculate the volatility α at 1 minute and 10 minutes scale 1min (t) and α 10min (t);

[0030] Step 11: Determine whether the volatility at the 1-minute and 10-minute scales meets the grid connection standard. If it does not meet the grid connection standard, go to step 9. If both meet the grid connection standard, go to step 12.

[0031] Step 12: Output the grid-connected power P at this time g (t), 1-minute and 10-minute volatility α 1min (t) and α 10min (t);

[0032] Step 13: Output the power P that the HESS needs to smooth out at this time HESS (t);

[0033] P HESS (t) = P W (t)-P g (t) (6)

[0034] Among them, P g (t) is the grid-connected power at time t;

[0035] Step 14: Let t=t+1, go to step 1, and smooth the wind power fluctuation at the next moment.

[0036] Preferably, the power distribution of the hybrid energy storage system comprises the following steps:

[0037] Based on the NGO algorithm, we consider using segmented mapping to generate a more evenly distributed initial solution set, improve the uniformity and quality of the initial set, avoid the algorithm from falling into the local optimum, and improve the global search ability of the algorithm. The segmented mapping function used is as follows:

[0038]

[0039] Among them, h is the control parameter of the mapping function, and x(t) is the random iteration value;

[0040] In the process of NGO optimization, the size of the weight will affect the convergence speed of the algorithm. Therefore, in order to improve the convergence speed of the NGO algorithm, adaptive weights and improved Levy flight strategies are introduced; in the early stage of algorithm iteration, adaptive weights are used to improve the global search ability of the algorithm and expand the search space of the algorithm; in the later stage of algorithm iteration, adaptive weights are used to improve the local search ability of the algorithm and find a better solution set. The expression of adaptive weights is as follows:

[0041]

[0042] Among them, ω is the adaptive weight, ω max and ω min are the maximum weight and the minimum weight respectively, generally ω max is 0.9, ωmin is 0.4, F i t is the objective function value of the i-th individual in the t-th iteration, is the maximum objective function value of all individuals at the tth iteration, The average value of the objective function of all individuals in the tth iteration, M is the total population.

[0043] Preferably, the power distribution of the hybrid energy storage system further comprises the following steps:

[0044] The Levy flight search strategy is a random behavior strategy used to simulate the step length and direction of a random walk or search process. The strategy combines short-distance exploratory advances with occasional long-distance walks. Short-distance advances can ensure that the northern goshawk can carefully search a small area around itself during hunting to achieve local optimality, while occasional long-distance walks can ensure that the northern goshawk can enter another area and surpass the local extreme value, thereby achieving global optimization. The formula of the Levy flight search strategy is as follows:

[0045]

[0046] Among them, u~N(0,σ u 2 ), v~N(0 , 1),

[0047] The formula of the INGO algorithm in the first stage is:

[0048]

[0049] Among them, x i,j is the position of the i-th northern goshawk in the j-th dimension, is the new position of the i-th northern goshawk in the j-th dimension in the first stage, x best,j is the optimal position of the northern goshawk in the jth dimension, P i,j is the position of the j-dimensional prey of the i-th northern goshawk, F Pi is the objective function value of the j-th dimension of the i-th northern goshawk, F i is the objective function;

[0050] The formula of the INGO algorithm in the second stage is:

[0051]

[0052] in, The new position of the northern goshawk in the second phase, F i new,P2 is the objective function value of the northern goshawk after the second stage update, x iis the position of the i-th northern goshawk, t and T are the current iteration number and the maximum iteration number, r and I are random numbers, R is the attack radius of the northern goshawk, and R gradually decreases with the increase of iteration number t, so that the algorithm continues to iterate until convergence.

[0053] Preferably, the power distribution of the hybrid energy storage system further comprises the following steps:

[0054] INGO is used to optimize the decomposition layer number K and penalty factor C in the VMD algorithm, and the optimized parameters are substituted into the VMD algorithm to decompose the power that needs to be smoothed by the HESS, thus completing the power allocation of the HESS.

[0055] The VMD algorithm defines each intrinsic mode function (IMF) as an amplitude-frequency modulation function as follows:

[0056] s K (t) = a K (t)cos[φ K (t)] (15)

[0057] Among them, the phase φ K (t) is a non-decreasing function, the signal envelope a K (t) is non-negative, ensuring that the frequency and amplitude of the IMF are non-decreasing, thus ensuring the stability and interpretability of the signal;

[0058] VMD assumes that all IMF components are concentrated at their respective center frequencies ω K The narrowband signal near the Hilbert transform is used to obtain the analytical signal, then the signal is translated to the baseband, and then s is obtained by the square of the 2 norm of the gradient k The bandwidth of (t) is as follows:

[0059]

[0060] Among them, {s K}、{ω K} are the IMF component set and its corresponding center frequency set, δ(t) is the pulse function, * is the convolution symbol, s K (t) is the Kth IMF component, e -jωkt is the phasor of the center frequency on the complex plane, and f is the decomposed signal;

[0061] In order to transform the constrained problem into an unconstrained problem and thus find the global optimal solution, the Lagrange multiplier operator λ is introduced, and then the penalty factor C is used to alternately update the alternating direction method. , thereby determining the optimal solution of the above formula, and then using ω-ω KSubstituting the variable ω, the IMF components and center frequencies are finally obtained as follows:

[0062]

[0063] in, is the Fourier transform form of f, For K The Fourier transform form of (t) is: is the Fourier transform form of the Lagrange multiplier operator λ, is the center frequency of the current IMF component, ω is the center frequency;

[0064] After the INGO-VMD algorithm, the power that needs to be smoothed by the HESS is divided into several IMF components, among which the low-frequency component is smoothed by the energy storage battery and the remaining high-frequency component is smoothed by the supercapacitor, completing the power distribution of the HESS:

[0065]

[0066] Among them, P bat (t) is the sum of the low-frequency components that the energy storage battery needs to smooth out at time t, P SC (t) is the sum of the high-frequency components that the supercapacitor needs to suppress at time t, and b is the dividing point between the low-frequency component and the high-frequency component.

[0067] Compared with the prior art, the present invention has the following beneficial effects:

[0068] 1. The present invention adopts the MSAMF algorithm to smooth the wind power, obtains the wind power that meets the grid connection standard, alleviates the problem of excessive smoothing, and reduces the burden of HESS.

[0069] 2. In the process of stabilization, the present invention adopts a dual fuzzy control method, comprehensively considers the elastic fluctuation rate of wind power and the SOC threshold of HESS to optimize the stabilization algorithm, realizes the stable operation of HESS within the safety range, and ensures the reliability of the system.

[0070] 3. The present invention adopts INGO to optimize VMD parameters and obtains the optimal parameter value through iterative calculation. Compared with NGO, INGO has faster search speed and higher optimization accuracy. INGO-VMD is used to decompose the hybrid energy storage power, which gives full play to the characteristics of power-type energy storage and energy-type energy storage. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 It is the overall flow chart of the present invention;

[0072] Figure 2 This is a flow chart of HESS smoothing wind power fluctuation in the present invention;

[0073] Figure 3 It is a power allocation flow chart of HESS in the present invention;

[0074] Figure 4 It is a grid-connected power curve diagram before and after the leveling in the present invention;

[0075] Figure 5 It is the volatility graph of 1min scale before and after the stabilization in the present invention;

[0076] Figure 6 It is the volatility graph of 10min scale before and after the stabilization in the present invention;

[0077] Figure 7 This is a diagram showing the output of the HESS in the present invention for smoothing wind power fluctuations;

[0078] Figure 8 It is an iterative process diagram of the algorithm in the present invention;

[0079] Fig. 9 It is the time domain waveform diagram of the IMF after decomposition in the present invention;

[0080] Fig.10 It is the IMF marginal spectrum after decomposition in the present invention;

[0081] Fig.11 A power curve diagram of the energy storage battery and supercapacitor in the present invention for smoothing the output fluctuations required;

[0082] Fig.12 is a SOC curve diagram of the energy storage battery in the present invention;

[0083] Fig.13 It is the SOC curve diagram of the supercapacitor in the present invention. DETAILED DESCRIPTION

[0084] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings, so that those skilled in the art can better understand the advantages and features of the present invention, thereby making a clearer definition of the protection scope of the present invention. The embodiments described in the present invention are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention.

[0085] A multi-scale adaptive wind power smoothing method considering hybrid energy storage capacity includes a hybrid energy storage system smoothing wind power fluctuations and power allocation of the hybrid energy storage system.

[0086] See also Figure 1The overall flow chart of the multi-scale adaptive wind power smoothing method considering the hybrid energy storage capacity is shown in the figure. First, a two-dimensional fuzzy control system is established to dynamically adjust the maximum allowable volatility according to the state of charge (SOC) of the hybrid energy storage and the volatility of wind power. Then, a sliding filter algorithm based on the Fibonacci sequence is proposed for smoothing wind power. On this basis, INGO is used to optimize the number of decomposition layers and penalty factors in the VMD algorithm, and the optimized parameters are substituted into the VMD algorithm to decompose the power that the HESS needs to smooth, and complete the power allocation of the HESS.

[0087] Among them, HESS smoothes the fluctuation of wind power: according to the state of charge (SOC) of the hybrid energy storage and the fluctuation rate of wind power, a two-dimensional fuzzy control system is established to dynamically adjust the maximum allowable fluctuation rate. According to the grid-connected standards of wind farms, a multi-scale adaptive sliding filtering algorithm is used to smooth the planned wind power to obtain the grid-connected power that meets the grid-connected standards and the power that the HESS needs to smooth.

[0088] See also Figure 2 The flow chart of HESS to smooth wind power fluctuations, the specific steps are as follows:

[0089] Step 1: Read the planned wind power P at time t W (t);

[0090] Step 2: Let the initial window value N at time t = 1;

[0091] Step 3: Calculate the maximum fluctuation power at 1 minute scale and the maximum fluctuation power at 10 minutes scale;

[0092] Step 4: Determine whether the maximum fluctuation rate under the two time scales in step 3 meets the grid connection standard. If it meets the grid connection standard, go to step 12; if it does not meet the grid connection standard, go to step 5;

[0093] Step 5: Calculate the difference ΔP(t) between the planned wind power at time t and the grid-connected power at time t-1;

[0094] ΔP(t)=P w (t)-P g (t-1) (1)

[0095] Among them, P w (t) is the planned wind power at time t, P g (t-1) is the grid-connected power at time t-1, ΔP(t) is the difference between the planned wind power at time t and the grid-connected power at time t-1;

[0096] Step 6: According to the size of ΔP(t), determine whether the power fluctuation at this time is in the smoothing stage. If ΔP(t)≤k, go to step 7, otherwise go to step 8; where k is the dividing point between power fluctuation smoothness and power fluctuation mutation;

[0097] Step 7: At this time, the power fluctuation is relatively smooth, and the maximum allowable fluctuation rate α for 1 minute is set as the initial value;

[0098] Step 8: At this time, the power fluctuation is in the mutation stage. Read the current SOC state and ΔP(t) of the HESS, and establish a two-dimensional fuzzy control system. The input quantities are the normalized SOC value ω SOC (t) and ΔP(t) value η ΔP (t), the normalization method is as follows:

[0099]

[0100] In the normalization process, if ΔP(t)>4k, let ΔP(t)=4k, similarly if ΔP(t)<4k, let ΔP(t)=-4k;

[0101] Among them, ω SOC The domain of (t) is [0,1], and the fuzzy subsets are {VS (extremely small), S (small), M (medium), B (large), VB (maximum)}; η ΔP The domain of (t) is [-1,1], and the fuzzy subset is {NB (negative large), NM (negative medium), NS (negative small), Z (zero), PS (positive small), PM (positive medium), PB (positive large)}; the output is the maximum allowable volatility α for 1 minute, and the established fuzzy control rules are shown in Table 1;

[0102] Step 9: Add 1 to the window value N;

[0103] Step 10: Calculate the grid-connected power at time t using the MSAMF algorithm based on the Fibonacci sequence according to the window value N, the planned wind power at time t, and the grid-connected power at time t-1, t-2, …, t-N+1, as follows:

[0104]

[0105]

[0106] Among them, a n is the Fibonacci sequence, n is the number of terms in the sequence, and N is the window value;

[0107] Then calculate the volatility α at 1 minute and 10 minutes scale 1min (t) and α 10min (t);

[0108] Step 11: Determine whether the volatility at the 1-minute and 10-minute scales meets the grid connection standard. If it does not meet the grid connection standard, go to step 9. If both meet the grid connection standard, go to step 12.

[0109] Step 12: Output the grid-connected power P at this time g (t), 1-minute and 10-minute volatility α 1min (t) and α 10min (t);

[0110] Step 13: Output the power P that the HESS needs to smooth out at this time HESS (t);

[0111] P HESS (t) = P W (t)-P g (t) (6)

[0112] Among them, P g (t) is the grid-connected power at time t;

[0113] Step 14: Let t=t+1, go to step 1, and smooth the wind power fluctuation at the next moment.

[0114] Table 1 Fuzzy control table

[0115]

[0116] The actual power data of a wind farm in one day is taken as the research object, with an installed capacity of 50MW, a sampling period of 24 hours, and a sampling rate of 1 minute. The HESS is composed of energy storage batteries and supercapacitors to smooth the fluctuation of wind power.

[0117] The adaptive moving average filter (IMAF) algorithm and the MSAMF algorithm are used to smooth the planned wind power, and the maximum allowable fluctuation rate α in 1 minute is 1min Take 4%, the volatility coefficient k in the MSAMF algorithm = 8% × PN. Figure 4 The two filtering algorithms are shown in the figure before and after the grid-connected power curve. As can be seen from the figure, the IMAF algorithm has the phenomenon of over-smoothing in the two time periods of 500-512min and 535-545min, which brings a greater burden to the HESS, while the MSAMF algorithm improves the wind power tracking capability in these two time periods and alleviates the phenomenon of over-smoothing.

[0118] See also Figure 5 and Figure 6They are the 1-minute and 10-minute volatility of the IMAF algorithm and the MSAMF algorithm before and after smoothing, respectively. It can be seen from the figure that the volatility before smoothing does not meet the grid-connected requirements, and the volatility of the two algorithms after smoothing both meet the grid-connected requirements. However, when the volatility before smoothing is too high, the wind power tracking capability of the IMAF algorithm is relatively weak, and the volatility after smoothing is low, with over-smoothing. However, the MSAMF algorithm has a strong wind power tracking capability, which alleviates the over-smoothing phenomenon.

[0119] See also Figure 7 The output of HESS using IMAF algorithm and MSAMF algorithm to smooth wind power fluctuations. It can be seen from the figure that after using the MSAMF algorithm to smooth out the fluctuations, the charging and discharging power of HESS is reduced, especially when the wind power fluctuations are large, such as the two time periods of 500-512min and 535-545min in the figure. Compared with the IMAF algorithm, the energy storage output of the MSAMF algorithm is significantly reduced.

[0120] Power allocation of HESS: The power allocation of HESS uses the improved Northern Goshawk Optimization (INGO) algorithm to optimize the variational mode decomposition (VMD) algorithm. Based on the Northern Goshawk Optimization (NGO) algorithm, the use of segmented mapping is considered to improve the algorithm search capability. In the optimization process, adaptive weights and improved Levy flight strategies are introduced to improve the convergence speed of the NGO algorithm. Then, INGO is used to optimize the decomposition layers and penalty factors in the VMD algorithm. The optimized parameters are substituted into the VMD algorithm to decompose the power that the HESS needs to smooth, and the power allocation of the HESS is completed.

[0121] See also Figure 3 This is the power allocation flow chart of HESS. The specific steps are as follows:

[0122] Based on the NGO algorithm, we consider using segmented mapping to generate a more evenly distributed initial solution set, improve the uniformity and quality of the initial set, avoid the algorithm from falling into the local optimum, and improve the global search ability of the algorithm. The segmented mapping function used is as follows:

[0123]

[0124] Among them, h is the control parameter of the mapping function, and x(t) is the random iteration value;

[0125] In the process of NGO optimization, the size of the weight will affect the convergence speed of the algorithm. Therefore, in order to improve the convergence speed of the NGO algorithm, adaptive weights and improved Levy flight strategies are introduced; in the early stage of algorithm iteration, adaptive weights are used to improve the global search ability of the algorithm and expand the search space of the algorithm; in the later stage of algorithm iteration, adaptive weights are used to improve the local search ability of the algorithm and find a better solution set. The expression of adaptive weights is as follows:

[0126]

[0127] Among them, ω is the adaptive weight, ω max and ω min are the maximum weight and the minimum weight respectively, generally ω max is 0.9, ω min is 0.4, F i t is the objective function value of the i-th individual in the t-th iteration, is the maximum objective function value of all individuals at the tth iteration, The average value of the objective function of all individuals in the tth iteration, M is the total population.

[0128] The Levy flight search strategy is a random behavior strategy used to simulate the step length and direction of a random walk or search process. The strategy combines short-distance exploratory advances with occasional long-distance walks. Short-distance advances can ensure that the northern goshawk can carefully search a small area around itself during hunting to achieve local optimality, while occasional long-distance walks can ensure that the northern goshawk can enter another area and surpass the local extreme value, thereby achieving global optimization. The formula of the Levy flight search strategy is as follows:

[0129]

[0130] Where β is a constant, v~N(0,1),

[0131] The formula of the INGO algorithm in the first stage is:

[0132]

[0133] Among them, x i,j is the position of the i-th northern goshawk in the j-th dimension, is the new position of the i-th northern goshawk in the j-th dimension in the first stage, x best,j is the optimal position of the northern goshawk in the jth dimension, P i,j is the position of the j-dimensional prey of the i-th northern goshawk, F Piis the objective function value of the j-th dimension of the i-th northern goshawk, F i is the objective function;

[0134] The formula of the INGO algorithm in the second stage is:

[0135]

[0136] in, The new position of the northern goshawk in the second phase, F i new,P2 is the objective function value of the northern goshawk after the second stage update, x i is the position of the i-th northern goshawk, t and T are the current iteration number and the maximum iteration number, r and I are random numbers, R is the attack radius of the northern goshawk, and R gradually decreases with the increase of iteration number t, so that the algorithm continues to iterate until convergence.

[0137] See also Figure 8 The figure shows the iterative process of the two algorithms, INGO and NOG. It can be seen from the figure that the fitness function value of INGO at the 8th iteration is 0.1805, and the fitness function value of NGO at the 13th iteration is 0.1883. It can be seen that INGO has a faster optimization speed and calculation accuracy than NGO. By optimizing the VMD parameters by INGO, the optimal K value is 11 and the optimal C value is 9983.

[0138] INGO is used to optimize the decomposition layer number K and penalty factor C in the VMD algorithm, and the optimized parameters are substituted into the VMD algorithm to decompose the power that needs to be smoothed by the HESS, thus completing the power allocation of the HESS.

[0139] The VMD algorithm defines each intrinsic mode function (IMF) as an amplitude-frequency modulation function as follows:

[0140] s k (t) = a k (t)cos[φ k (t)] (15)

[0141] Among them, the phase φ k (t) is a non-decreasing function, the signal envelope a k (t) is non-negative, ensuring that the frequency and amplitude of the IMF are non-decreasing, thus ensuring the stability and interpretability of the signal;

[0142] VMD assumes that all IMF components are concentrated at their respective center frequencies ω K The narrowband signal near the Hilbert transform is used to obtain the analytical signal, then the signal is translated to the baseband, and then s is obtained by the square of the 2 norm of the gradient KThe bandwidth of (t) is as follows:

[0143]

[0144] Among them, {s k}、{ω K} are the IMF component set and its corresponding center frequency set, δ(t) is the pulse function, * is the convolution symbol, s K (t) is the Kth IMF component, e -jωkt is the phasor of the center frequency on the complex plane, and f is the decomposed signal;

[0145] In order to transform the constrained problem into an unconstrained problem and thus find the global optimal solution, the Lagrange multiplier operator λ is introduced, and then the penalty factor C is used to alternately update the alternating direction method. In order to determine the optimal solution of the above formula, we can use ω-ω K Substituting the variable ω, the IMF components and center frequencies are finally obtained as follows:

[0146]

[0147] in, is the Fourier transform form of f, For K The Fourier transform form of (t) is is the Fourier transform form of the Lagrange multiplier operator λ, is the center frequency of the current IMF component, ω is the center frequency;

[0148] After the INGO-VMD algorithm, the power that needs to be smoothed by the HESS is divided into several IMF components, among which the low-frequency component is smoothed by the energy storage battery and the remaining high-frequency component is smoothed by the supercapacitor to complete the power distribution of the HESS.

[0149] The optimal parameters obtained by using INGO to optimize VMD are the power P required by HESS. Hess See also Fig. 9 The time domain waveforms of the 11 IMFs after decomposition are shown in Figure 2. Fig.10 Perform Hilbert transform on IMF to get the corresponding marginal spectrum. As can be seen from the figure, the frequency distinction characteristics of different IMFs are obvious and easy to distinguish.

[0150] By observation Fig.10From the aliasing degree of each mode, it can be clearly seen that the mode aliasing phenomenon between IMF8 and IMF9 is the slightest, so IMF8 is used as the boundary of the high and low frequency components of HESS, IMF1~IMF8 are allocated as low frequency components to the energy storage battery, and IMF9~IMF11 are allocated as high frequency components to the supercapacitor to complete the power distribution of HESS, as shown in Fig.11 As shown, the power output required for the energy storage battery and supercapacitor to smooth out the fluctuations.

[0151]

[0152] Among them, P bat (t) is the sum of the low-frequency components that the energy storage battery needs to smooth out at time t, P SC (t) is the sum of high-frequency components that the supercapacitor needs to smooth out at time t.

[0153] The rated capacity of the energy storage battery used is 2MWh, and the rated capacity of the supercapacitor is 0.06MWh. Fig.12 The SOC curve of the energy storage battery shows that after the MSAMF algorithm is smoothed and the INGO-VMD is decomposed, the SOC of the energy storage battery is in the reasonable range of 20% to 80%. However, after the IMAF algorithm is smoothed and the INGO-VMD is decomposed, the SOC of the energy storage battery exceeds the limit 5 times, and even the SOC is 0, which will reduce the life of the energy storage battery. Fig.13 The SOC curve of the supercapacitor shows that the SOC of the supercapacitor after the two smoothing algorithms is in the reasonable range of 20% to 80%. Although the SOC of the supercapacitor after the MSAMF algorithm smoothing and INGO-VMD decomposition is relatively high, it is also within a reasonable range. The results show that after the proposed strategy smoothing and allocation, the SOC of the HESS can be maintained within a reasonable range, avoiding overcharging and over-discharging.

[0154] In summary, the present invention can not only adaptively smooth out wind power fluctuations, but also effectively reduce the energy storage burden due to the optimization of the charging and discharging range of the HESS, realize the power distribution of the HESS, reduce mode aliasing, avoid overcharging or over-discharging of the HESS, and ensure the safe and stable operation of the HESS.

[0155] The description and practice disclosed in the present invention are easy to think and understand for ordinary technicians in the technical field, and several improvements and modifications can be made without departing from the principles of the present invention. Therefore, modifications or improvements made without departing from the spirit of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A multi-scale adaptive wind power leveling method considering hybrid energy storage capacity, characterized in that: Including hybrid energy storage system to smooth wind power fluctuations and power distribution of hybrid energy storage system; The hybrid energy storage system smoothes the wind power fluctuation by adopting a multi-scale adaptive sliding filter MSAMF algorithm. The hybrid energy storage system smoothes the wind power fluctuation by establishing a two-dimensional fuzzy control system to dynamically adjust the maximum allowable fluctuation rate according to the charge state SOC of the hybrid energy storage and the fluctuation rate of wind power. According to the wind farm grid connection standard, a multi-scale adaptive sliding filter algorithm is used to smooth the wind power plan power to obtain the grid-connected power that meets the grid connection standard and the power that the hybrid energy storage system needs to smooth. The power distribution of the hybrid energy storage system adopts the improved goshawk optimization INGO algorithm to optimize the variational mode decomposition VMD algorithm. On the basis of the goshawk optimization NGO algorithm, segmented mapping is considered to improve the algorithm search ability. In the optimization process, adaptive weights and improved Levy flight strategy are introduced to improve the convergence speed of the NGO algorithm. Then, INGO is used to optimize the decomposition layer number K and penalty factor C in the VMD algorithm. The optimized parameters are substituted into the VMD algorithm to decompose the power that needs to be smoothed by the hybrid energy storage system, and the power distribution of the hybrid energy storage system HESS is completed.

2. A multi-scale adaptive wind power leveling method considering hybrid energy storage capacity according to claim 1, characterized in that: The hybrid energy storage system for smoothing wind power fluctuations comprises the following steps: Step 1: Read the planned wind power P at time t W (t); Step 2: Initialize window value N=1; Step 3: Calculate the maximum fluctuation power at 1 minute scale and the maximum fluctuation power at 10 minutes scale; Step 4: Determine whether the maximum fluctuation rate under the two time scales in step 3 meets the grid connection standard. If it meets the grid connection standard, go to step 12; if it does not meet the grid connection standard, go to step 5; Step 5: Calculate the difference ΔP(t) between the planned wind power at time t and the grid-connected power at time t-1; ΔP(t)=P w (t)-P g (t-1) (1) Among them, P w (t) is the planned wind power at time t, P g (t-1) is the grid-connected power at time t-1, ΔP(t) is the difference between the planned wind power at time t and the grid-connected power at time t-1; Step 6: According to the size of ΔP(t), determine whether the power fluctuation at this time is in the smoothing stage. If ΔP(t)≤k, go to step 7, otherwise go to step 8; where k is the dividing point between power fluctuation smoothness and power fluctuation mutation; Step 7: At this time, the power fluctuation is relatively smooth, and the maximum allowable fluctuation rate α for 1 minute is set as the initial value; Step 8: At this time, the power fluctuation is in the mutation stage. Read the current SOC state and ΔP(t) of the HESS, and establish a two-dimensional fuzzy control system. The input quantities are the normalized SOC value ω SOC (t) and ΔP(t) value η ΔP (t), the normalization method is as follows: In the normalization process, if ΔP(t)>4k, let ΔP(t)=4k, similarly if ΔP(t)<4k, let ΔP(t)=-4k; Among them, ω SOC The domain of (t) is [0,1], and the fuzzy subsets are {VS (extremely small), S (small), M (medium), B (large), VB (maximum)}; η ΔP The domain of (t) is [-1,1], and the fuzzy subsets are {NB (negative large), NM (negative medium), NS (negative small), Z (zero), PS (positive small), PM (positive medium), PB (positive large)}; the output is the maximum allowable volatility α for 1 minute; Step 9: Add 1 to the window value N; Step 10: According to the window value N, the planned wind power at time t and the grid-connected power at time t-1, t-2, …, t-N+1, the MSAMF algorithm based on the Fibonacci sequence is used to calculate the grid-connected power at time t, as follows: Among them, a n is the Fibonacci sequence, n is the number of terms in the sequence, and N is the window value; Then calculate the volatility α at 1 minute and 10 minutes scale 1min (t) and α 10min (t); Step 11: Determine whether the volatility at the 1-minute and 10-minute scales meets the grid connection standard. If it does not meet the grid connection standard, go to step 9. If both meet the grid connection standard, go to step 12. Step 12: Output the grid-connected power P at this time g (t), 1-minute and 10-minute volatility α 1min (t) and α 10min (t); Step 13: Output the power P that the HESS needs to smooth out at this time HESS (t); P HESS (t)=P W (t)-P g (t) (6) Among them, P g (t) is the grid-connected power at time t; Step 14: Let t=t+1, go to step 1, and smooth the wind power fluctuation at the next moment.

3. A multi-scale adaptive wind power leveling method considering hybrid energy storage capacity according to claim 1, characterized in that: The power distribution of the hybrid energy storage system comprises the following steps: Based on the NGO algorithm, we consider using segmented mapping to generate a more evenly distributed initial solution set, improve the uniformity and quality of the initial set, avoid the algorithm from falling into the local optimum, and improve the global search ability of the algorithm. The segmented mapping function used is as follows: Among them, h is the control parameter of the mapping function, and x(t) is the random iteration value; In the process of NGO optimization, the size of the weight will affect the convergence speed of the algorithm. Therefore, in order to improve the convergence speed of the NGO algorithm, adaptive weights and improved Levy flight strategies are introduced; in the early stage of algorithm iteration, adaptive weights are used to improve the global search ability of the algorithm and expand the search space of the algorithm; in the later stage of algorithm iteration, adaptive weights are used to improve the local search ability of the algorithm and find a better solution set. The expression of adaptive weights is as follows: Among them, ω is the adaptive weight, ω max and ω min are the maximum weight and minimum weight respectively, ω max is 0.9, ω min is 0.4, is the objective function value of the i-th individual in the t-th iteration, is the maximum objective function value of all individuals at the tth iteration, The average value of the objective function of all individuals in the tth iteration, M is the total population.

4. A multi-scale adaptive wind power leveling method considering hybrid energy storage capacity according to claim 3, characterized in that: The power distribution of the hybrid energy storage system further comprises the following steps: The Levy flight search strategy is a random behavior strategy used to simulate the step length and direction of a random walk or search process. The strategy combines short-distance exploratory advances with long-distance walks. Short-distance advances ensure that the northern goshawk can carefully search a small area around itself during hunting to achieve local optimality, while long-distance walks ensure that the northern goshawk can enter another area and surpass the local extreme value, thereby achieving global optimization. The formula of the Levy flight search strategy is as follows: Where β is a constant, v~N(0,1), The formula of the INGO algorithm in the first stage is: Among them, x i,j is the position of the i-th northern goshawk in the j-th dimension, is the new position of the i-th northern goshawk in the j-th dimension in the first stage, x best,j is the optimal position of the northern goshawk in the jth dimension, P i,j is the position of the j-dimensional prey of the i-th northern goshawk, is the objective function value of the j-th dimension of the i-th northern goshawk, F i is the objective function; The formula of the INGO algorithm in the second stage is: in, The new position of the Northern Goshawk for the second phase, is the objective function value of the northern goshawk after the second stage update, x i is the position of the i-th northern goshawk, t and T are the current iteration number and the maximum iteration number, r and I are random numbers, R is the attack radius of the northern goshawk, and R gradually decreases with the increase of iteration number t, so that the algorithm continues to iterate until convergence.

5. A multi-scale adaptive wind power leveling method considering hybrid energy storage capacity according to claim 4, characterized in that: The power distribution of the hybrid energy storage system further comprises the following steps: INGO is used to optimize the decomposition layer number K and penalty factor C in the VMD algorithm, and the optimized parameters are substituted into the VMD algorithm to decompose the power that needs to be smoothed by the HESS, thus completing the power allocation of the HESS. The VMD algorithm defines each intrinsic mode function IMF as an amplitude modulation frequency modulation function as follows: s K (t)=a K (t)cos[φ K (t)] (15) Among them, the phase φ K (t) is a non-decreasing function, the signal envelope a K (t) is non-negative, ensuring that the frequency and amplitude of the IMF are non-decreasing, thus ensuring the stability and interpretability of the signal; VMD assumes that all IMF components are concentrated at their respective center frequencies ω K The narrowband signal near the Hilbert transform is used to obtain the analytical signal, then the signal is translated to the baseband, and then s is obtained by the square of the 2 norm of the gradient K The bandwidth of (t) is as follows: Among them, {s K }、{ω K } are the IMF component set and its corresponding center frequency set, δ(t) is the pulse function, * is the convolution symbol, s K (t) is the Kth IMF component, e -jωkt is the phasor of the center frequency on the complex plane, and f is the decomposed signal; In order to transform the constrained problem into an unconstrained problem and thus find the global optimal solution, the Lagrange multiplier operator λ is introduced, and then the penalty factor C is used to alternately update the alternating direction method. In order to determine the optimal solution of the above formula, we can use ω-ω K Substituting the variable ω, the IMF components and center frequencies are finally obtained as follows: in, is the Fourier transform form of f, For K The Fourier transform form of (t) is is the Fourier transform form of the Lagrange multiplier operator λ, is the center frequency of the current IMF component, ω is the center frequency; After the INGO-VMD algorithm, the power that needs to be smoothed by the HESS is divided into several IMF components, among which the low-frequency component is smoothed by the energy storage battery and the remaining high-frequency component is smoothed by the supercapacitor, completing the power distribution of the HESS: Among them, P bat (t) is the sum of the low-frequency components that the energy storage battery needs to smooth out at time t, P SC (t) is the sum of the high-frequency components that the supercapacitor needs to suppress at time t, and b is the dividing point between the low-frequency component and the high-frequency component.

Images