A power distribution method based on a hybrid energy storage system and application thereof

Abstract

A power distribution method based on a hybrid energy storage system, comprising the following steps: CEEMD decomposition of wind power generation data; intrinsic mode function energy entropy calculation; power distribution based on fuzzy control optimization method. The present application proposes to decompose the wind and light power fluctuation based on the CEEMD method, and to optimize the energy storage configuration combination of power distribution after completing the energy entropy distribution by using the fuzzy control method. The present application helps to ensure the accuracy of power distribution and further improves the efficiency of fluctuating power suppression decomposition. The protection point is to make the SOC fluctuation range of super capacitor and battery below 10%, and the power is always maintained at about 50% of the battery SOC, so that both kinds of energy storage devices can play the maximum role, and the overall efficiency and service life of the hybrid energy storage system are improved.

Description

Technical Field

[0001] This invention relates to a power distribution method and its application, and more particularly to a power distribution method based on a hybrid energy storage system and its application. Background Technology

[0002] In the current context of increasingly tight power supply and frequent news of power rationing, in order to improve the service life and operational efficiency of various energy storage devices, it is necessary to propose an efficient and accurate power allocation method to smooth out power fluctuations caused by wind and solar power generation and maintain the stable and continuous operation of the power grid.

[0003] Complementary Ensemble Empirical Mode Decomposition (CEEMD) is a method for processing adaptive signals, derived stepwise from and optimized from Ensemble Empirical Mode Decomposition (EEMD). CEEMD builds upon EEMD by transforming the step of adding white noise in EEMD into two separate signals: one with white noise added and the other with white noise subtracted. These two signals are then simultaneously averaged using EEMD. This reduces the residual white noise in the reconstructed signal after EEMD decomposition, accelerates the decomposition iteration rate, and decreases the number of lumped averaging iterations from hundreds in EEMD to tens in CEEMD. Generally, the more lumped averaging iterations, the lower the noise in the reconstructed signal. Therefore, adaptive decomposition of fluctuating power based on the CEEMD method is more efficient and reliable.

[0004] In wind and solar power systems, high-frequency and low-frequency fluctuations are the main types of power fluctuations caused by unstable weather conditions. Correspondingly, there are two types of energy storage devices that can absorb or release energy. Power-type energy storage devices, such as supercapacitors, have high power density but low energy density, and their charging and discharging frequencies are relatively high. In contrast, energy-type energy storage devices have low power density but high energy density; batteries are a typical example of energy-type devices, with charging and discharging frequencies lower than supercapacitors. Therefore, a well-planned combination of both, leveraging their strengths and compensating for their weaknesses, to maximize their characteristics and advantages plays a crucial role in absorbing and mitigating power fluctuations.

[0005] Existing technologies, such as those by Guo Lingjuan, Wei Bin, and Han Xiaoqing, use the EEMD decomposition method to smooth wind power fluctuations, analyzing information entropy and energy entropy data to allocate and correct fluctuation power; Li Dayong, Du Mingliang, and Tian Chunguang, based on the EEMD method, obtain the first allocation order k1 by using the standardized modulus cumulative mean method on the Intrinsic Mode Function (IMF), and then adaptively calculate the second allocation power command k2; Fu Juxia, Chen Jie, and Teng Yangxin, based on the parameter requirements for power limits and time response speed in relevant Chinese national standards, use the EEMD method to decompose the original wind power data, enabling the energy storage device to execute relevant command actions in specific scenarios; Li Yang proposed an improved method based on the adaptive droop coefficient of inductor current, which stabilizes the bus voltage while adjusting the current and voltage, thereby improving the power allocation accuracy. However, the process of using the EEMD method in the above-mentioned existing technologies is mostly to first use EEMD to decompose the power signal into high and low frequencies to complete the primary power allocation, and then use fuzzy control to optimize and constrain the SOC status of the energy storage element to finally obtain the secondary power allocation result. Summary of the Invention

[0006] This invention aims to solve the problem of how to decompose the power fluctuations of wind and solar power generation using the CEEMD method, and to improve the power allocation configuration based on the primary allocation. The ultimate goal is to enable the energy storage system to operate continuously and efficiently, maintaining its stable operation. This invention proposes to use the CEEMD method to decompose the power fluctuations of wind and solar power generation, obtain its intrinsic mode function (IMF) and margin, perform a primary power allocation based on energy entropy, and then set constraints on the SOC of the energy storage devices based on fuzzy control, performing a secondary allocation and optimization to achieve the predetermined optimization objective. The technical solution is as follows:

[0007] A power allocation method based on a hybrid energy storage system, characterized by the following steps:

[0008] Step 1: CEEMD decomposition of wind power generation data;

[0009] Step 2: Calculation of energy entropy of intrinsic mode functions;

[0010] Step 3: Power allocation based on fuzzy control optimization method.

[0011] Beneficial effects

[0012] After using the CEEMD method to decompose the power fluctuations of wind and solar power generation and completing the secondary power allocation of related energy storage devices based on the fuzzy control method, the SOC state of the energy storage devices can be maintained within a reasonable range, reducing their power fluctuations. This has a positive impact on the smoothing of fluctuations in the entire wind power generation system and meets the normal operating output needs of the wind power system. Attached Figure Description

[0013] Figure 1 A diagram showing the steps of the CEEMD algorithm;

[0014] Figure 2 The input supercapacitor SOC;

[0015] Figure 3 The input supercapacitor ΔSOC SC Schematic diagram;

[0016] Figure 4 A schematic diagram of the input and output membership functions for fuzzy control of a supercapacitor;

[0017] Figure 5 A schematic diagram of the energy entropy of each IMF component;

[0018] Figure 6 A schematic diagram showing the energy entropy difference of each IMF component;

[0019] Figure 7 Adjusting the state of charge (SOC) before and after supercapacitor and battery;

[0020] Figure 8 This is a schematic diagram showing the power distribution between supercapacitors and batteries. Detailed Implementation

[0021] A power allocation method based on a hybrid energy storage system, characterized by the following steps:

[0022] Step 1: CEEMD decomposition of wind power generation data (see...) Figure 1 (as shown);

[0023] First, this method improves upon the EEMD method. The most widely used EEMD method is the ensemble empirical mode decomposition method, whose decomposition principle is that when the added white noise is uniformly distributed throughout the time-frequency space, this space is composed of different scale components segmented by the filter bank, allowing for stepwise decomposition. This step improves upon the well-known EEMD method by transforming the single-directional white noise signal into a pair of adaptive white noise signals with positive and negative polarities, based on multiple empirical mode decompositions of superimposed Gaussian white noise. Therefore, we can assume the original signal is x0(t), and the added bidirectional adaptive white noise signals are ±f(t), resulting in two new signals X1(t) and X2(t).

[0024]

[0025] The above process involves adding a pair of positive and negative noise once. However, in practical applications, the general procedure is to first set the average number NE in the program, which can be understood as the number of additions. Assuming NE = n, this means adding n pairs of positive and negative white noise ± f(t). After adding n pairs of white noise, EMD decomposition is performed on these n pairs of new signal sources, and the corresponding calculation results are obtained using the following formula:

[0026]

[0027] In the formula, IMF i Let i be the i-th IMF component after decomposition in X1, and let IMF be... j Let be the i-th IMF component after decomposition in X2; δ(t) and λ(t) are the decomposition remainders, which is one pair of EMD decomposition results in the algorithm. Assume there are n similar decomposition results in the entire process. To obtain the final decomposition result, we must first:

[0028]

[0029] The final decomposition results are as follows:

[0030]

[0031] In the formula: To find the q-th IMF variable obtained after averaging; That is, the remainder after calculating the total average; according to the above calculation, after n decompositions, the final decomposition remainder 'a' is... i The terms σ(t) and the remainder σ(t) are shown below:

[0032]

[0033] Step 2: Calculation of energy entropy of intrinsic mode functions;

[0034] Entropy measures the degree of disorder within a system; it is a measure of the state of a material system. Since Intrinsic Mode Functions (IMFs) also contain energy, energy entropy can be used to measure the total energy of each IMF. Combining energy entropy values, the largest energy entropy difference is used as the dividing point, i.e., the first distribution of power. Assume the energy entropy of each IMF function is E... i The total energy is E, and the remaining energy is generally not included in the formula. The respective calculation formulas are as follows:

[0035]

[0036]

[0037]

[0038] Where p i The energy entropy E represents the energy of each mode function. i The proportion of total energy E. Based on the calculated energy entropy of each IMF component, the difference in energy entropy between each two adjacent components is calculated. The interval where the maximum difference occurs is used as the boundary to distinguish between high-frequency and low-frequency bands. After verification, it can be set as the boundary point k for the primary distribution of power.

[0039] After determining the value of the dividing point k, the IMF components before k are regarded as high-frequency components. Because of their high frequency, high-power energy storage devices such as supercapacitors can be used to absorb or release energy. Conversely, all IMF components after k are regarded as low-frequency components and are absorbed or released by the battery.

[0040]

[0041] In the formula P sc This refers to the power that a supercapacitor needs to absorb or release; P bat It is the power that the battery needs to absorb or release; ε e is the IMF balance; n is the total number of IMF components.

[0042] Step 3: Power allocation based on fuzzy control optimization method.

[0043] After completing the above steps, the next main consideration is the safe range of the SOC value of the entire energy storage system. Simply put, the SOC of both the battery and the supercapacitor must be kept within a safe range simultaneously, serving as a constraint on the energy storage system. This paper will first calculate the change in the state of charge (ΔSOC) of the supercapacitor using the following formula. sc The calculation formula is as follows:

[0044]

[0045] Without considering the self-discharge rates of the two types of energy storage batteries, P in equation (10) sc (t) represents the high-frequency power command of the supercapacitor at time t; η c1 and η d1 These are the charging efficiency and discharging efficiency of the supercapacitor, respectively; E sc This is the rated operating capacity value of the supercapacitor. The above calculations can be used to read the state of charge of the supercapacitor during operation, based on P... sc The sign of (t) and the SOC state of the supercapacitor should be considered, and the charging and discharging command parameters should be adjusted appropriately according to the specific circumstances.

[0046] The input parameters of the fuzzy controller include the current state of charge (SOC) of the supercapacitor and the ΔSOC calculated according to equation (10). sc The value of fuzzy control is the adjustment coefficient K of the high-frequency power corresponding to the supercapacitor. p , (0 <K p <1). Therefore, (1-K p )P sc (t) represents the additional power that the battery needs to handle.

[0047] Then, we introduce the formula for calculating the change in the state of charge of a battery:

[0048]

[0049] Similarly, in the formula, P bat (t) represents the high-frequency power command of the battery at time t; η c2 and η d2 These are the charging efficiency and discharging efficiency of the battery, respectively; E bat This refers to the operating capacitance of the supercapacitor. During operation, the supercapacitor's state of charge is read, based on P... bat The sign of (t) and the SOC state of the battery are used to make appropriate adjustments to its charging and discharging process.

[0050] Then create a fuzzy controller whose input parameters are the current state of charge (SOC) of the battery and the ΔSOC calculated according to equation (11). bat The value of . Its output is the adjustment coefficient K of the battery corresponding to the low-frequency power. b (0 <K b <1).

[0051] Finally, P bat (t)K b and P bat (t)K b +(1-K p )P sc (t) Compare the two; if the difference between them does not exceed ±2%, they are considered the same, and the adjustment coefficient is selected as K. p or K b Both are acceptable; if the difference between the two is within ± (2% to 5%), the average of the two can be taken as the corresponding low-frequency power of the battery; if the difference between the two exceeds ±5%, the energy signal needs to be decomposed and new parameters need to be set, as shown in Table 1.

[0052] Table 1 Supercapacitor Power Adjustment Scheme

[0053]

[0054]

[0055] Based on the calculated data, plot the input and output membership function graphs of the fuzzy controller corresponding to the supercapacitor, as follows: Figure 2 As shown. The commonly used fuzzy variable linguistic values ​​in fuzzy control are as follows: NB (Negative Big): negative big, NM (Negative Medium): negative medium, NS (Negative Small): negative small, ZO (Almost Zero): almost zero, PS (Positive Small): positive small, PM (Positive Medium): positive medium, PB (Positive Big): positive big.

[0056] Plot the input and output membership functions of the fuzzy controller corresponding to the supercapacitor, as follows: Figure 2 As shown in Figures 3 and 4. The fuzzy rule control table is shown in Table 2. Using the fuzzy control method can minimize the number of charge-discharge cycles of the battery, thereby extending the battery's service life, while maximizing the role of the supercapacitor and relieving the pressure of regulating the battery's charge-discharge power.

[0057] Table 2 Fuzzy Rule Control Table

[0058]

[0059] P bat (t)K b and P bat (t)K b +(1-K p )P sc (t) Compare the two; if the difference between them does not exceed ±2%, they are considered the same, and the adjustment coefficient is selected as K. p or K b Either is acceptable.

[0060]

[0061] If the difference between the two is within ±(2% to 5%), then the average of the two can be taken as the corresponding low-frequency power of the battery.

[0062]

[0063] In summary, the entire process is as follows:

[0064] First, the system performs CEEMD decomposition on the input wind power generation data, obtaining a set of IMF components and remainder terms. Then, using the energy entropy principle, it calculates the energy entropy values ​​of each order of IMF components and the differences between them, identifying the two IMF components with the largest differences and using them as primary power allocation points to divide the wind power generation into two groups: high-frequency energy and low-frequency energy. Next, it calculates the ΔSOC for each of the two groups of high-frequency and low-frequency energy. sc and ΔSOC bat The parameters K are respectively input into the two fuzzy controllers set in the fuzzy control optimization method for inference calculation, and two adjustment parameters K are obtained. p (0 <K p <1) and K b (0 <K b <1), and then the above comparison method is used to select the best one from the two to obtain the final allocation adjustment parameters. The secondary power allocation of the supercapacitor and the battery is carried out using equations (12) and (13).

[0065] Example

[0066] Step 1: Data Collection and CEEMD Decomposition of Wind Power Generation

[0067] This invention uses raw data from a wind farm with an installed capacity of 66MW. The sampling interval is 15 minutes, with 96 sampling points per day. Data mining is used to access 100 days of data. Based on the above formula, the final decomposition remainder and remainder terms are calculated after n decompositions. The IMF component curves and remainder term curves of each order are plotted using the CEEMD algorithm.

[0068] Step 2: Calculation of energy entropy of intrinsic mode functions

[0069] After plotting the IMF component curves of each order, the energy entropy value p of each IMF function is calculated using the above formula. The energy entropy difference between each IMF function can then be obtained by subtraction. The energy entropy of each IMF component and the energy entropy difference between each IMF component are as follows: Figure 5 , Figure 6 As shown.

[0070] As shown in the graph above, without considering the remainder, the difference between IMF4 and IMF5 is the largest among the IMF components of each order. This difference can be used as a dividing point to distinguish between high and low frequency power. Here, the power dividing point is taken as k=5. That is:

[0071]

[0072] Step 3: Power allocation based on fuzzy control optimization method.

[0073] Next, a simulation model of the hybrid energy storage system is performed to further verify whether the fluctuation range of the supercapacitor and battery SOC meets the limiting conditions. The specific initial parameter configuration of the entire hybrid energy storage system is shown in Table 3.

[0074] Table 3 Parameters of Hybrid Energy Storage Equipment

[0075] parameter Supercapacitor storage battery Rated capacity / (MW*h) 5.9 6.6 SOC range / % 25-85 25-80 Initial SOC / % 50 50 Rated charging power / kW 4450 1620 Rated discharge power / kW 4555 2875 Charge / discharge efficiency / % 98 98

[0076] Based on the aforementioned fuzzy control rules and the parameter configuration specifications in Table 3, the SOC adjustment curves of the two energy storage devices before and after charging and discharging can be obtained, as follows: Figure 7 As shown. Simultaneously, the original power data graph and its high and low values ​​are plotted.

[0077] Simultaneously, draw the original power data diagram and a schematic diagram of the distribution of high and low frequency power between the supercapacitor and the battery, such as... Figure 8 As shown.

[0078] according to Figure 7 After adjusting the system's power distribution, the SOC curves of both energy storage devices were optimized to some extent, with reduced peak and valley values ​​and increased stability. Specifically, the battery's SOC changed slowly, stabilizing around 0.58, and remained relatively stable, meeting the aforementioned constraints. In contrast, the supercapacitor's SOC changed more significantly, stabilizing around 0.40 towards the end, with the overall curve approaching 0.50. This indicates that the supercapacitor primarily handles sudden power changes in the later stages, providing high-frequency power to the external system and allowing the energy storage system's state to fluctuate slightly around the desired value.

[0079] exist Figure 8 It can be observed that supercapacitors have significantly more charge-discharge cycles than batteries, primarily handling the high-frequency components of power fluctuations, while batteries mainly handle the low-frequency components. This CEEMD method for power distribution fully leverages the characteristics and advantages of both energy storage devices, enabling peak shaving and valley filling, smoothing fluctuations, maintaining the stable operation of the wind power system, and extending the overall lifespan of the energy storage system.

[0080] Finally, this invention proposes a method based on CEEMD to decompose the power fluctuations of wind and solar power generation, obtaining their intrinsic mode functions (IMFs) and margins. Then, using fuzzy control, after completing the initial allocation of energy entropy, the energy storage configuration is optimized for power allocation, achieving optimal system planning within a suitable SOC range. This invention helps ensure the accuracy of power allocation and further improves the efficiency of power fluctuation smoothing and decomposition. The key protection lies in controlling the SOC of the energy storage devices within a certain range. Ultimately, the scheme also ensures that the SOC fluctuation range of both the supercapacitor and the battery is below 10%, and the battery capacity is consistently maintained at approximately 50% of the battery's SOC. This allows both energy storage devices to function at their maximum potential, achieving the previously predetermined goals and improving the overall efficiency and lifespan of the hybrid energy storage system.

[0081] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A power allocation method based on a hybrid energy storage system, characterized by: Includes the following steps: Step 1: CEEMD decomposition of wind power generation data; First, the most widely used EEMD method currently refers to the ensemble empirical mode decomposition method. Its decomposition principle is that when the added white noise is uniformly distributed throughout the time-frequency space, this space is composed of components of different scales segmented by the filter bank, thus undergoing gradual decomposition. An improvement on the EEMD method is made by transforming the only unidirectional white noise signal into a pair of adaptive white noise signals with positive and negative values, based on multiple empirical mode decompositions of superimposed Gaussian white noise. Assuming the original signal is... The subsequent added bidirectional adaptive white noise signal is ± Two new signals were obtained. and : ； Set the average number of iterations NE, assuming NE = n, which means adding n times of positive and negative white noise ± After adding n times of white noise, EMD decomposition is performed on these n pairs of new signal sources, and the corresponding calculation results are obtained using the following formula: ； In the formula Let i be the i-th IMF component after decomposition in X1. Let i be the i-th IMF component after decomposition in X2; , To decompose the remainder, this is one pair of EMD decomposition results in the algorithm. Assume there are a total of n similar decomposition results throughout the process. Next, to obtain the final decomposition result, we must first: ； The final decomposition results are as follows: ； In the formula: To find the q-th IMF variable obtained after averaging; That is, the remainder after calculating the total average; according to the above calculation, after n decompositions, the final decomposition remainder is... and remainder As shown below: ； Energy entropy is used to measure the total energy possessed by each IMF function. Combining the energy entropy values, the largest difference in energy entropy is taken as the dividing point, i.e., the first distribution of power. Assume the energy entropy of each IMF function is as follows: The total energy is E, and the remaining energy terms are not included in the formula; their respective calculation formulas are as follows: ； in Energy entropy representing each mode function Total energy The proportion of the power; based on the calculated energy entropy of each IMF component, the energy entropy difference between each two adjacent components is calculated, and the interval where the maximum difference occurs is used as the boundary to distinguish between high frequency and low frequency bands. After verification, it is set as the boundary point k of the first power distribution. After determining the value of the dividing point k, the IMF components before k are considered high-frequency components, and because of their high frequency, supercapacitors are used to absorb or release their energy. Conversely, all IMF components after k are considered low-frequency components and are absorbed or released by the battery. ； In the formula This refers to the power that the supercapacitor needs to absorb or release; This refers to the power that the battery needs to absorb or release. It is the IMF's surplus; It is the total number of IMF components; Step 2: Calculation of energy entropy of intrinsic mode functions; Step 3: Power allocation based on fuzzy control optimization method.

2. The power allocation method based on a hybrid energy storage system according to claim 1, characterized in that: The formula for calculating the change in state of charge (ΔSOCsc) of a supercapacitor is as follows: ； Without considering the self-discharge rates of the two types of energy storage batteries, in equation (10) This represents the high-frequency power command of the supercapacitor at time t; and These are the charging efficiency and discharging efficiency of the supercapacitor, respectively. This refers to the rated operating capacity of the supercapacitor; the state of charge of the supercapacitor during operation is read through the above calculations, based on... The positive and negative values ​​of the capacitor and the state of charge (SOC) of the supercapacitor should be considered, and the charging and discharging command parameters should be adjusted appropriately according to the specific circumstances.

3. The power allocation method based on a hybrid energy storage system according to claim 1, characterized in that: Step three further includes the following: The input parameters of the fuzzy controller include the current state of charge (SOC) value of the supercapacitor and the value calculated according to equation (10). The value; the output of fuzzy control is the adjustment coefficient of the high-frequency power corresponding to the supercapacitor. ,0< <1; then, This refers to the additional power that the battery needs to handle; Then, the formula for calculating the change in battery state of charge is introduced: ； In the formula This represents the high-frequency power command of the battery at time t; and These are the charging efficiency and discharging efficiency of the battery, respectively. This refers to the operating capacity of the supercapacitor. Supercapacitors read their state of charge during operation, according to The positive and negative values ​​of the battery and its SOC state are used to adjust the charging and discharging process appropriately; then a fuzzy controller is created, whose input parameters are the current SOC value of the battery and the value calculated according to equation (11). The value; Its output is the adjustment coefficient of the battery corresponding to low-frequency power. ,0< <1; Finally, and If the two are compared and the difference does not exceed ±2%, they are considered the same, and the adjustment coefficient is selected. or Either is acceptable; ； If the difference between the two is within ± (2%~5%), then the average value of the two can be taken as the corresponding low-frequency power of the battery. ； If the difference between the two exceeds ±5%, the energy signal needs to be re-decomposed and new parameters need to be set.

4. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, wherein the program, when executed, controls the device where the non-volatile storage medium is located to perform the method described in any one of claims 1 to 3.

5. An electronic device, characterized in that, It includes a processor and a memory; the memory stores computer-readable instructions, and the processor is configured to execute the computer-readable instructions, wherein the computer-readable instructions, when executed, perform the method according to any one of claims 1 to 3.

Images