An adaptive power allocation method for hybrid energy storage considering battery response speed

Abstract

The present invention discloses a hybrid energy storage adaptive power allocation method considering the battery response speed, including the following steps: first, using a sliding average filtering algorithm to calculate the optimal window; according to the optimal window trajectory combined with the Markov chain model, the probability decision matrix of the sliding average filter adaptive window is obtained, and the optimal step length of the optimal window data under the fixed step length is calculated by exhaustive method, and the scheme decision formula is calculated according to the fixed step length formula; according to the characteristics of the optimal window trajectory and the scheme decision formula, a suitable adaptive filtering method is selected between the probability decision matrix and the fixed step length method to allocate the hybrid energy storage power. Based on the consideration of the response speed of the battery in the low and medium frequency bands, the present invention combines the characteristics of the hybrid energy storage historical data set, analyzes and obtains a hybrid energy storage adaptive filtering method, reduces the capacity configuration of the lithium battery while ensuring the stable operation of the system, improves the economy, and effectively ensures the stable operation of the hybrid energy storage system.

Description

Technical Field

[0001] The invention relates to a hybrid energy storage adaptive power distribution method taking battery response speed into consideration, belonging to the technical field of new energy storage system methods. Background Art

[0002] Wind power generation and photovoltaic power generation can not only alleviate the increasing energy demand in society, but also meet the concept of sustainable development of society. However, the intermittent and uncertain nature of wind power generation and photovoltaic power generation increases the difficulty of energy scheduling in new energy systems. Although the connection of wind power generation and photovoltaic power generation to the grid can alleviate certain energy demand, due to the widespread use of power electronic devices in new energy power generation and the low inertia characteristics of power electronic devices, the power grid is less friendly to new energy. With the increase of installed capacity of wind power generation and photovoltaic power generation in the power grid, the fluctuation of new energy power generation will gradually have a greater impact on the power grid. The combination of energy storage system with wind power generation and photovoltaic power generation can effectively reduce the intermittent and uncertain nature of new energy power generation and improve the friendliness of the power grid to new energy. The use of energy storage can effectively compensate for or absorb the power deviation caused by the prediction accuracy in the process of new energy power generation.

[0003] Limited by the response speed of the battery, in the case of large fluctuations in the power deviation of new energy, the battery cannot absorb or compensate for all the power alone. At this time, the hybrid energy storage system composed of batteries and supercapacitors can effectively compensate and absorb the power deviation. The operating principle of the hybrid energy storage system composed of batteries and supercapacitors is to divide the power deviation into three parts: low, medium and high frequency. The battery is responsible for the exchange of energy in the medium and low frequency bands, and the supercapacitor is responsible for the exchange of energy in the medium and high frequency bands. The main principle of the sliding average filter algorithm in the hybrid energy storage system is to complete the calculation of the battery exchange power through the sliding average filter window. When the battery exchange power is determined, the exchange of the remaining energy in the hybrid energy storage system is completed by the supercapacitor. Although the hybrid energy storage system can effectively improve the reliability of new energy power generation, its high configuration cost, operation and maintenance cost limit its further promotion and utilization, and the source of this part of the cost is mainly caused by the increase and replacement of battery capacity configuration. Therefore, it is necessary to adopt a suitable adaptive filtering technology that can perform power distribution, so as to complete the switching of the filter window during the operation of the hybrid energy storage system, reduce the capacity demand of the battery, and give full play to the role of the supercapacitor. Summary of the invention

[0004] In view of the problems existing in the above-mentioned prior art, the present invention provides a hybrid energy storage adaptive power allocation method taking into account the battery response speed, and further can provide an adaptive power allocation method for hybrid energy storage economy.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is: a hybrid energy storage adaptive power allocation method considering battery response speed, comprising the following steps;

[0006] Step 1: Determine the lower limit of the sliding average filter window according to the response speed of the battery, and consider that the battery needs to be responsible for the exchange of low-frequency energy, thereby determining the upper limit of the sliding average filter window;

[0007] Step 2: Using the trial and error method based on the historical data of the hybrid energy storage system to determine the optimal window trajectory under the historical data;

[0008] Step 3: Based on the historical data of hybrid energy storage, an exhaustive method is used to determine the optimal step size of the regulation window under the fixed step size scheme;

[0009] Step 4: According to the optimal window trajectory determined in step 2, the probability decision matrix is ​​determined using the Markov chain model;

[0010] Step 5: Based on the optimal step length determined under the fixed step length scheme in step 3, the decision formula for the scheme can be obtained;

[0011] Step 6: According to the characteristics of the optimal window trajectory data and the scheme decision formula, select a suitable scheme between the fixed step length scheme and the probability decision matrix scheme to determine the filter window under the adaptive sliding average filter.

[0012] Furthermore, the specific implementation method of step 2 is to traverse the corresponding battery output power of all windows at each sampling point, calculate the corresponding battery demand capacity until the sampling point, and select the window value corresponding to the minimum capacity as the optimal window value corresponding to the sampling point.

[0013] Furthermore, the stability and continuity of the optimal window trajectory obtained under the historical data in step 2 are analyzed using a first-order forward difference algorithm;

[0014] The calculation formula of the first-order forward difference sequence of the historical optimal window trajectory is shown in formula (1);

[0015] Y(t)=W(t+1)-W(t)t=1,...,n-1 (1);

[0016] Where W(t) is the optimal window value at sampling point t, Y(t) is the first-order forward difference value at sampling point t; n is the number of sampling points;

[0017] Through the first-order forward difference sequence, the corresponding difference sequence mean and variance can be calculated.

[0018]

[0019] Where μ is the mean of the difference sequence; δ is the variance of the difference sequence.

[0020] Furthermore, the step length determination scheme in step 3 is specifically as follows:

[0021] The sliding average filter window is continuously iterated with a fixed step length. In order to obtain a fixed step length with higher credibility, the fixed step length is set by exhaustive method for the historical data of hybrid energy storage. Specifically, on the basis of the known upper and lower limits of the sliding average filter window, all step lengths are traversed to comprehensively select the optimal step length under the fixed step length scheme. Under the fixed step length scheme, the window will be continuously iterated and updated. The specific iteration formula is shown in formula (3);

[0022]

[0023] Where step is the optimal step size selected under the fixed step size scheme; inv is the step size scaling factor; win(t) is the sliding average filter window value at sampling point t; win min —Lower limit of sliding average filter window; win max —The upper limit of the sliding average filter window; sig(t)—the step direction correction coefficient at the sampling point t;

[0024] The step length direction correction coefficient is determined according to formula (4);

[0025]

[0026] Where Eup(t) is the capacity demand value of the maximum value of the filter window at time t; Edown(t) is the capacity demand value of the minimum value of the filter window at time t; P win,max (t)—battery exchange power at time t when the filtering window is at its maximum value; P win,min (t)—battery exchange power at time t when the filtering window is at the minimum value;

[0027] Whether the step size is iteratively updated under the fixed step size scheme requires Table 1 to determine;

[0028] Table 1: Window update judgment procedure

[0029]

[0030] Where E(t) is the capacity demand value at time t under the adaptive filtering window.

[0031] Furthermore, in the step 4, the adaptive filtering scheme of the Markov chain constructing the probability decision matrix uses the probability statistics method to analyze the optimal window trajectory, obtains the probability of each window corresponding to the next window by the statistical analysis method, and completes the setting of the adaptive filtering window; the specific implementation method is:

[0032] Step 41: Calculate the maximum value win_max of the optimal window trajectory sequence;

[0033] Step 42: construct a window trajectory change matrix win1 (win1 is a matrix of win_max*win_max);

[0034] Step 43: Count the frequency of jumps to window values ​​from 1 to win_max under different window values ​​according to the optimal trajectory sequence. For example, for the window value at time t and its corresponding window value at the next time t+1, the number of corresponding values ​​of the win1 matrix is ​​increased by 1, as shown in formula (5).

[0035] win1(W(t),W(t+1))=win1(W(t),W(t+1))+1 (5);

[0036] Step 44: performing a row-wise sum operation on the obtained window trajectory change matrix to obtain a window matrix win2 (win2 is a matrix of win_max*win_max+1, and the last column is the sum of the elements in the row);

[0037] Step 45, according to formula (6), the final probability decision matrix win3 can be obtained;

[0038]

[0039] Furthermore, in step 6, when the fixed step length scheme is determined, a scheme decision formula is constructed, and a suitable filtering scheme is selected through the scheme decision formula. The specific formula is shown in formula (7);

[0040]

[0041] The beneficial effects of the present invention are as follows: the present invention uses a trial-and-error method to traverse the corresponding capacity requirements under all windows and obtain the optimal window trajectory sequence; a probability decision matrix is ​​constructed through a Markov chain to realize the update operation of the adaptive sliding average filter window. Considering the rapidity problem of the fixed step-length formula, the scheme decision formula is determined according to the fixed step-length formula. Taking into account the characteristics of the Markov chain with good rapidity and poor stability, and the fixed step-length formula with poor rapidity and good stability, the first-order forward difference is used to analyze the optimal window trajectory sequence, and the variance of the first-order forward difference sequence is used to reflect the continuity and stability of the optimal window trajectory sequence, and the appropriate adaptive filtering scheme is selected through the scheme decision formula. . BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A technical block diagram of a hybrid energy storage adaptive power allocation method considering battery response speed of the present invention;

[0043] Figure 2 is an optimal window trajectory diagram of data set 1 in an embodiment of the present invention;

[0044] Figure 3 is an optimal window trajectory diagram of data set 2 in an embodiment of the present invention;

[0045] Figure 4 A diagram showing battery capacity requirements under different step lengths of a data set in an embodiment of the present invention;

[0046] Figure 5 Graph showing battery capacity requirements under different step lengths of data set 2 in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail below through the accompanying drawings and embodiments. However, it should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the scope of the present invention.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which the present invention belongs. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0049] This embodiment is shown in Example 1.

[0050] A hybrid energy storage adaptive power allocation method considering the battery response speed, wherein the hybrid energy storage system is composed of batteries and supercapacitors, wherein the battery is responsible for the exchange of energy in the low and medium frequency bands, and the supercapacitor is responsible for the exchange of energy in the high and medium frequency bands. The characteristic of the sliding average algorithm is that the larger the sliding window, the fewer high-frequency components pass through, and the smaller the sliding window, the more high-frequency components pass through. The lower limit value of the sliding average filter window is determined according to the battery response speed, and considering that the battery needs to be responsible for the exchange of low-frequency energy, the upper limit value of the sliding average filter window is determined, and the upper limit value of the filter window is 12, and the lower limit value is 4.

[0051] The optimal window trajectory under the historical data of the hybrid energy storage system is determined by trial and error. The specific implementation method is to traverse the corresponding battery output power under all windows at each sampling point, calculate the corresponding battery demand capacity until the sampling point, and select the window corresponding to the minimum capacity as the optimal window value corresponding to the sampling point.

[0052] In order to better illustrate the difference between the adaptive filtering scheme under the fixed step size formula and the adaptive filtering scheme under the probability decision matrix, two data sets are used for comparative analysis. Figure 2 It shows the optimal step size trajectory in dataset 1. Figure 3 Shows the optimal step size trajectory in dataset 2.

[0053] In the fixed step length scheme, a fixed step length is used to continuously iterate the sliding average filter window. In order to obtain a fixed step length with high credibility, the fixed step length is set by exhaustive method for the historical data of hybrid energy storage. Based on the known upper and lower limits of the sliding average filter window, all step lengths are traversed to comprehensively select the optimal step length under the fixed step length scheme, where the step length scaling coefficient is 0.01. All the results of the data set traversal under the fixed step length scheme are as follows Figure 4 As shown in , the data step length is between 100-200, which requires the lowest capacity. Therefore, the step length for data set 1 is 150. All the results of traversing data set 2 under the fixed step length scheme are shown in Figure 5 As shown, the fixed step length has the lowest capacity requirement when it is between 1 and 99, so the fixed step length for data set 2 is 50.

[0054] The adaptive filtering scheme of Markov chain construction probability decision matrix uses probability statistics method to analyze the optimal window trajectory, obtains the probability of each window corresponding to the change to the next window by statistical analysis method, and completes the setting of adaptive filtering window.

[0055] Considering the continuity and rapidity of the fixed-step scheme, the optimal window cannot be tracked quickly when the optimal window trajectory changes rapidly. When the fixed-step scheme is determined, a scheme decision formula is constructed to select a suitable filtering scheme. The scheme decision formula for data set 1 is shown in formula (8), and the scheme decision formula for data set 2 is shown in formula (9).

[0056]

[0057]

[0058] According to the optimal trajectory under historical data, the stability and continuity of the optimal trajectory are analyzed using the first-order forward difference algorithm. The variance value of the first-order forward difference sequence corresponding to data set 1 is 0.8603, and the variance value of the first-order forward difference sequence corresponding to data set 2 is 2.8376.

[0059] According to the characteristics of the optimal trajectory data and the scheme decision formula, a suitable scheme is selected between the fixed step length scheme and the probability decision matrix scheme to determine the filter window under the adaptive sliding average filter. For data set 1, referring to formula (8), it can be found that the fixed step length scheme is more appropriate. For data set 2, referring to formula (9), the probability decision matrix scheme is more appropriate. In data set 2, the continuity of the optimal window trajectory is poor and the fluctuation is more violent. At this time, the optimal capacity value obtained by the fixed step length algorithm is 1.2347*10 6Wh, the capacity value obtained after using the decision matrix is ​​1.2318*10 6 Wh, it can be found that for data set 2, after adopting the probability decision matrix scheme, the capacity demand has been significantly reduced. Table 2 shows the probability decision matrix obtained by Markov chain for data set 2.

[0060] Table 2 is the Markov chain probability decision matrix of data set 2, which is as follows;

[0061] 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.8689 0.9697 0.9824 0.9863 0.9891 0.9906 0.9928 0.9937 1 0 0 0 0 0.3019 0.3224 0.8672 0.9308 0.9523 0.9580 0.9750 0.9784 1 0 0 0 0 0.2045 0.2254 0.2480 0.8261 0.9130 0.9420 0.9549 0.9646 1 01 0 0 0 0.1810 0.1897 0.1983 0.2047 0.8556 0.9332 0.9612 0.9698 1 0 0 0 0 0.1199 0.1295 0.1343 0.1391 0.1535 0.8561 0.9353 0.9544 1 0 0 0 0 0.1146 0.1198 0.1198 0.1250 0.1276 0.1328 0.8281 0.9245 1 0 0 0 0 0.1228 0.1304 0.1304 0.1304 0.1330 0.1381 0.1432 0.8440 1 0 0 0 0 0.0884 0.0909 0.0909 0.0909 0.0909 0.0934 0.0960 0.0960 1 0 0 0 0 0.0675 0.0733 0.0768 0.0787 0.0808 0.0834 0.0893 0.0992 1

[0062] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modification, equivalent substitution or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A hybrid energy storage adaptive power allocation method considering battery response speed, characterized in that: The steps include: Step 1: Determine the lower limit of the sliding average filter window according to the response speed of the battery, and consider that the battery needs to be responsible for the exchange of low-frequency energy, thereby determining the upper limit of the sliding average filter window; Step 2: Using the trial and error method based on the historical data of the hybrid energy storage system to determine the optimal window trajectory under the historical data; Specifically, the stability and continuity of the optimal window trajectory obtained under the historical data in step 2 are analyzed using a first-order forward difference algorithm; The calculation formula of the first-order forward difference sequence of the historical optimal window trajectory is shown in formula (1); Y(t)=W(t+1)-W(t) t=1,…,n-1 (1); Where W(t) is the optimal window value at sampling point t, Y(t) is the first-order forward difference value at sampling point t; n is the number of sampling points; Through the first-order forward difference sequence, the corresponding difference sequence mean and variance can be calculated. Where μ is the mean of the difference sequence; δ is the variance of the difference sequence; Step 3: Based on the historical data of hybrid energy storage, an exhaustive method is used to determine the optimal step size of the regulation window under the fixed step size scheme; The specific step length scheme in step 3 is: The sliding average filter window is continuously iterated with a fixed step length. In order to obtain a fixed step length with higher credibility, the fixed step length is set by exhaustive method for the historical data of hybrid energy storage. Specifically, on the basis of the known upper and lower limits of the sliding average filter window, all step lengths are traversed to comprehensively select the optimal step length under the fixed step length scheme. Under the fixed step length scheme, the window will be continuously iterated and updated. The specific iteration formula is shown in formula (3); Where step is the optimal step size selected under the fixed step size scheme; inv is the step size scaling factor; win(t) is the sliding average filter window value at sampling point t; win min —Lower limit of sliding average filter window; win max —The upper limit of the sliding average filter window; sig(t)—the step direction correction coefficient at the sampling point t; The step length direction correction coefficient is determined according to formula (4); Where Eup(t) is the capacity demand value of the maximum value of the filter window at time t; Edown(t) is the capacity demand value of the minimum value of the filter window at time t; P win,max (t)—battery exchange power at time t when the filtering window is at its maximum value; P win,min (t)—battery exchange power at time t when the filtering window is at the minimum value; Whether the step size is iteratively updated under the fixed step size scheme requires Table 1 to determine; Table 1: Window update judgment procedure Where E(t) is the capacity demand value at time t under the adaptive filtering window; Step 4: According to the optimal window trajectory determined in step 2, the probability decision matrix is ​​determined using the Markov chain model; The adaptive filtering scheme of the Markov chain constructing the probability decision matrix in step 4 uses the probability statistics method to analyze the optimal window trajectory, obtains the probability of each window corresponding to the next window by the statistical analysis method, and completes the setting of the adaptive filtering window; the specific implementation method is: Step 41: Calculate the maximum value win_max of the optimal window trajectory sequence; Step 42: construct a window trajectory change matrix win1, where win1 is a matrix of win_max*win_max; Step 43: Count the frequencies of window values ​​jumping to 1 to win_max under different window values ​​according to the optimal trajectory sequence. At this time, for the window value at time t and its corresponding window value at the next time t+1, the number of corresponding values ​​of the win1 matrix is ​​increased by 1, as shown in formula (5); win1(W(t),W(t+1))=win1(W(t),W(t+1))+1(5); Step 44: perform a row-wise sum operation on the obtained window trajectory change matrix to obtain a window matrix win2, where win2 is a win_max*win_max+1 matrix, and the last column is the sum of the elements in the row; Step 45, according to formula (6), the final probability decision matrix win3 can be obtained; Step 5: Based on the optimal step length determined under the fixed step length scheme in step 3, the decision formula for the scheme can be obtained; Step 6: According to the characteristics of the optimal window trajectory data and the solution decision formula, a suitable solution is selected between the fixed step length solution and the probability decision matrix solution to determine the filter window under the adaptive sliding average filter; In step 6, when the fixed step length scheme is determined, a scheme decision formula is constructed, and a suitable filtering scheme is selected through the scheme decision formula. The specific formula is shown in formula (7); 2. A hybrid energy storage adaptive power allocation method considering battery response speed according to claim 1, characterized in that: The specific implementation method of step 2 is to traverse the corresponding battery output power in all windows at each sampling point, calculate the corresponding battery demand capacity until the sampling point, and select the window value corresponding to the minimum capacity as the optimal window value corresponding to the sampling point.

Images