Energy storage typical working condition generation method and device based on dynamic time sequence algorithm

Abstract

The application discloses a kind of energy storage typical working condition generation method and device based on dynamic time average sequence algorithm, the method includes the time series of multiple according to periodic sampling is formed set, initialization optimal time sequence, and time sequence in the set is successively carried out dynamic time regulation, find optimal path and increase path to optimal time sequence, calculate after sequentially, and obtain optimal time sequence mean, the above process can be obtained final optimal time sequence by repeated iteration, positive and negative value adjustment is carried out to the optimal time sequence, satisfies the charge-discharge capacity equal constraint of energy storage working condition.The working condition curve generated by the application can better extract the typical characteristics of original data, and has broad application prospect.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electrochemical energy storage, and particularly relates to a typical working condition generation method for energy storage based on a dynamic time average sequence algorithm. BACKGROUND

[0002] In the test of energy storage batteries, typical working conditions in actual operation need to be obtained for repeated testing. However, the working conditions of large-scale lithium battery energy storage power stations are complex, and simple averaging often cannot reflect the typical working conditions and is also easily affected by time offset to generate unreasonable working condition curves.

[0003] In the document Zhou Renjun et al. Source-load storage coordination optimization model with source-load similarity and curve fluctuation degree constraints [J]. Proceedings of the Chinese Society of Electrical Engineering, 2020, 40(13): 4092-4102, the time sequence similarity measurement method is improved, and the numerical and morphological similarity distance of the load curve and the new energy output curve is calculated as the new energy-load similarity index. The document uses dynamic time warping (DTW) to measure the distance of the load curve. In the document Li Yong et al. Distributed photovoltaic and energy storage optimization configuration based on joint time sequence scenario and source-grid-load coordination [J]. Transactions of Electrical Engineering Technology, 2022, 37(13): 3289-3303, a joint optimization configuration method for DPV and energy storage system (ESS) based on joint time sequence scenario and source-grid-load coordination is proposed. A light-load joint time sequence scenario generation method based on dynamic warping time characteristics is proposed to handle source-load uncertainty and time sequence correlation.

[0004] Patent CN201610222049.0 A kind of energy storage system typical working condition curve acquisition method and system, provide a kind of energy storage system typical working condition curve acquisition method and system, method reads data;Determine or read the charge-discharge power data of energy storage system, constitute charge-discharge power matrix;Get feature power value vector;Determine the typical power value of energy storage system;Excavate and integrate all time's typical power value, obtain energy storage system typical working condition curve.System includes communication module, data storage and management module, smooth controller module and typical working condition curve excavation module of mutual communication.The patent CN201510079728.2 a kind of energy storage battery typical working condition formulation method provides a kind of energy storage battery typical working condition formulation method, comprising the following steps: determining the actual operating condition performance parameter matrix of energy storage battery;Determine factor analysis mathematical model, and solve factor loading matrix;The distance between the factor scores of any two working condition segments is calculated;Reconstruct energy storage battery typical working condition.The patent CN201410665899.9 a kind of energy storage system smooth fluctuation typical working condition generation method and system proposes a kind of energy storage system smooth fluctuation typical working condition generation method and system, this method includes: obtaining the working data of energy storage system under smooth fluctuation mode;According to the smooth fluctuation curve generated after working data processing, and according to the corresponding feature distribution data generated by calculation according to the smooth fluctuation curve;The smooth fluctuation curve is segmented according to a unit time t, and a plurality of unit smooth fluctuation curve segments are obtained;Select a certain number a of unit smooth fluctuation curve segments, to form a plurality of combination smooth fluctuation curve segments with a time length of a x t;According to the feature distribution data corresponding to a plurality of combination smooth fluctuation curve segments, the error value between the feature distribution data corresponding to the smooth fluctuation curve is calculated;Select the feature distribution data corresponding to the combination smooth fluctuation curve segment with error value less than a certain numerical value, and synthesize the typical working condition diagram of energy storage system smooth fluctuation.

[0005] The above method does not carry out special research on the generation of energy storage battery typical working condition, and cannot guarantee that the charge-discharge capacity of typical working condition is the same, so it is difficult to use the curve to carry out cycle test on energy storage battery. SUMMARY

[0006] The present application aims to provide a kind of energy storage typical working condition generation method based on dynamic time average sequence algorithm, solve the problem that existing algorithm cannot extract working condition with time offset, and can guarantee that the curve obtained is the same in charge-discharge capacity.

[0007] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0008] In a first aspect of the present application, a typical working condition generation method for energy storage based on a dynamic time average sequence algorithm is provided. The working condition of an energy storage battery in a unit time period T can form a discrete time sequence q={I q1 ,I q2 ,…,I qn}, where each element in q is the current value of each sampling point in the unit time period T, n is the number of sampling points in the unit time period T, and the elements in q are arranged in chronological order from early to late; the time sequence set of the energy storage battery in m unit time periods T is Q={q1,q2,…,q m}, and the dynamic time average sequence algorithm is used to generate an optimal typical working condition for each discrete time sequence q k in Q, 1≤k≤m.

[0009] As a further optimization scheme of the present application, the dynamic time average sequence algorithm is as follows:

[0010] 1) Iteration number iter=1, initialize the typical working condition optimal time sequence C best =C best 0 , the sequence C best has a dimension of n, N is the maximum number of iterations, and ∈ is the termination iteration threshold;

[0011] 2) Let the discrete time sequence number k=1 in the set Q, and the time sequence C=C best iter-1 ;

[0012] 3) Generate the cumulative distance matrix γ best iter-1 between the time sequence C k and the kth time sequence q k in the time sequence set Q, γ k is an n*n matrix;

[0013] 4) Starting from the (n,n) position of the matrix γ k , search for the optimal path to (1,1), and for each point (g,h), 1 k (g-1,h-1), γ k (g,h-1), and γ k (g-1,h)}, the search method is to take the maximum value from the direction until reaching the (1,1) position, and the elements passed in the optimal path are added to the sequence C during the search process;

[0014] 5) If k≤m, the time sequence number k=k+1, and return to step 3), otherwise go to step 6);

[0015] 6) Take the mean of each element in sequence C to obtain the optimal sequence in the iterth iteration.

[0016] C best iter ;

[0017] 7) If iter ≤ N and C best iter The optimal sequence C in the (iter-1)th iteration best iter-1 If the Euclidean distance is greater than ω, then let the iteration number iter = iter + 1 and return to step 2); otherwise, proceed to step 8.

[0018] 8) Output C best iter C is the optimal time series for typical operating conditions. best .

[0019] As a further optimization of this application, the optimal time series C for typical operating conditions is initialized. best The method is as follows: From the set Q = {q1, q2, ..., q...} m Randomly select any time series as C. best 0 .

[0020] As a further optimization of this application, the time series C and the k-th time series q in the time series set Q are... k Generate cumulative distance matrix γ k The method is as follows:

[0021] 1) Generate γ k The data in the first row (i=1) and the first column (j=1) are represented by the following formula:

[0022] γ k (i,j)=d(I qi ,I cj )

[0023] In the above formula, d(I qi ,I cj ) indicates I qi with I cj The Euclidean distance;

[0024] 2) For points where n≥i>1 or n≥j>1, calculate γ according to the following formula. k (i,j):

[0025] γ k (i,j)=d(I qi ,I cj )+min{γ k (i-1,j-1),γ k(i-1,j),γ k (i,j-1)}

[0026] In the formula, I qi For sequence q k The elements, 1≤i≤n, I cj Let be the elements of sequence C, 1≤j≤n, γ k (i,j) represents the cumulative distance matrix γ k The element in the i-th row and j-th column, γ k (i-1,j-1) represents the cumulative distance matrix γ k The element in the (i-1)th row and (j-1)th column, γ k (i-1,j) represents the cumulative distance matrix γ k The element in the (i-1)th row and jth column, γ k (i,j-1) represents the cumulative distance matrix γ k The element in the i-th row and j-1-th column, i and j increase in order from earliest to latest time.

[0027] As a further optimization of this application, the method for adding the elements traversed in the optimal path to sequence C during the search process is as follows: Let C = {I} before one search process. c1 ,I c2 ,…,I cn If the sampled value I needs to be processed after the search is complete... ct’ Add to the t-th element I ct In the case where 1 ≤ t ≤ n, then add I ct’ The subsequent sequence C is represented as C = {I} c1 ,I c2 ,…,{I ct ,I ct’},…,I cn}

[0028] As a further optimization of this application, the method for averaging each element of sequence C is as follows: Let C = {I} c1 ,I c2 ,…,I cn}, for the t-th element I ct 1≤t≤n, calculate the mean I according to the following formula. ct_avg :

[0029]

[0030] In the above formula, b is a set I. ct In the middle element, B is I ct The number of elements in the middle.

[0031] As a further optimization of the present application, the maximum iteration number N takes the number of sequences m in the set Q, and the termination iteration threshold ∈ = max{||q k ||,q k ∈Q} / S, ||q k || is the modulus of the time series q k , 1≤k≤m; S is a large number set, which can be taken as 10000.

[0032] In order to ensure that the charge and discharge curve area is equal, as a further optimization of the present application, the generated typical working condition optimal time series C best , the working condition curve is adjusted by using the following formula:

[0033]

[0034]

[0035]

[0036]

[0037] In the above formula, I cpos_x is the xth positive element in C best , g is the number of positive elements in C best , I cpos_y is the yth negative element in C best , h is the number of negative elements in C best , C pos_sum and C neg_sum are the sum of positive and negative elements in C best , respectively, I cpos_x_new and I cneg_y_new are the adjusted values of I cpos_x and I cpos_y , respectively.

[0038] The second aspect of the present application proposes a dynamic time average sequence algorithm-based energy storage typical working condition generation device, comprising a time series set acquisition module and a typical working condition generation module.

[0039] The time series set acquisition module is used to acquire the time series set Q = {q1, q2, …, q m} of the energy storage battery in m unit time periods T.

[0040] The typical working condition generation module is used to generate a typical working condition optimal time series using the dynamic time average sequence algorithm for any discrete time series q k in Q, 1≤k≤m.

[0041] As a further optimization scheme of the present application, the energy storage typical working condition generating device based on the dynamic time average sequence algorithm further comprises a working condition curve adjusting module configured to adjust the optimal time sequence C of the typical working condition generated by the typical working condition generating module best The working condition curve is adjusted to ensure that the charge and discharge curve areas are equal.

[0042] Compared with the prior art, the method provided by the present application can effectively solve the complex and variable operation conditions of the battery, generate a typical working condition for extracting the working characteristics, and the charge and discharge amounts in the working condition are equal, i.e., the net charge in the cycle period is zero, which can be repeatedly tested for multiple cycles, and has a wide application prospect. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0044] Figure 1 A flowchart of a typical working condition generating method of an energy storage battery based on a dynamic time average sequence algorithm is shown in an example embodiment.

[0045] Figure 2 A typical working condition of energy storage obtained using the mean method is shown in an example embodiment.

[0046] Figure 3 A typical working condition of energy storage obtained using the dynamic time average sequence algorithm is shown in an example embodiment. DETAILED DESCRIPTION

[0047] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0048] The present application provides a typical working condition generating method of an energy storage battery based on a dynamic time average sequence algorithm, as shown in Figure 1 The method comprises the following steps:

[0049] The working condition of the energy storage battery in a unit time period T can form a discrete time sequence q={I q1 ,I q2 ,…,I qn}, each element in q is the current value at sampling time, n is the number of sampling points in a unit time period T, and the elements in q are arranged in order from early to late time; the time sequence set of the energy storage battery in m unit time periods T is Q = {q1, q2,..., qm}, each time sequence q in Q is generated by sampling the current value of the energy storage battery at each sampling point in a unit time period T, and the elements in q are arranged in order from early to late time. m}, each element in q is the current value at sampling time, n is the number of sampling points in a unit time period T, and the elements in q are arranged in order from early to late time; the time sequence set of the energy storage battery in m unit time periods T is Q = {q1, q2,..., qm}, each time sequence q in Q is generated by sampling the current value of the energy storage battery at each sampling point in a unit time period T, and the elements in q are arranged in order from early to late time. k , 1≤k≤m, the typical working condition optimal time sequence is generated using a dynamic time average sequence algorithm:

[0050] 1) Iteration number iter = 1, initialize the typical working condition optimal time sequence C best = C best 0 , the dimension of sequence C best is n, N is the maximum number of iterations, and ∈ is the termination iteration threshold;

[0051] 2) Let the time sequence number k = 1 in the set Q, and the time sequence C = C best iter-1 ;

[0052] 3) The time sequence C best iter-1 is compared with the kth time sequence q k in the time sequence set Q, and the cumulative distance matrix γ k is generated, γ k is an n*n matrix;

[0053] 4) Starting from the (n,n) position of the matrix γ k , search for the optimal path to the (1,1) position, and for the (g,h)

[0054] position, 1 k (g-1,h-1), γ k (g,h-1), and γ k (g-1,h)} from the direction of the maximum value, until reaching the (1,1) position, and the elements passed in the optimal path are added to the sequence C during the search process;

[0055] 5) If k≤m, the time sequence number k = k+1, and return to step 3), otherwise go to step 6);

[0056] 6) Take the average of each element in the sequence C to obtain the optimal sequence C

[0057] C best iter of the iter iteration;

[0058] 7) If iter≤N and C best iter is the optimal sequence Cbest iter-1 If the Euclidean distance of is greater than, let the time series index iter = iter + 1, return to step 2), otherwise enter step 8);

[0059] 8) Output C best iter Typical working condition optimal time series C best .

[0060] Initialize typical working condition optimal time series C best The method is: randomly take any one time series from the set Q = {q1, q2, …, q m} as C best 0 .

[0061] I qi is the element of sequence q k , 1≤i≤n, I cj is the element of sequence C, 1≤j≤n, i and j respectively represent the row number and column number of the cumulative distance matrix γ k , i and j increase in the order of time from early to late, and the time series C is the kth time series q k Generate the cumulative distance matrix γ k The method is:

[0062] 1) Generate the first row i = 1 and the first column j = 1 data of γ k , using the following formula:

[0063] γ k (i,j) = d(I qi ,I cj )

[0064] In the formula, d(I qi ,I cj ) represents the Euclidean distance of I qi and I cj ;

[0065] 2) For points where n≥i>1 or n≥j>1, calculate γ k (i,j) according to the following formula:

[0066] γ k (i,j) = d(I qi ,I cj ) + min{γ k (i-1,j-1), γ k (i-1,j), γ k (i,j-1)}.

[0067] The method for adding the elements in the optimal path to the sequence C during the search process is as follows: let C={I c1 ,I c2 ,…,I cn} before the first search process, each element in C is a sampling value or a set of sampling values, if the sampling value I ct ’ needs to be added to the tth element I ct in C after the search is finished, 1≤t≤n, then the sequence after the addition is represented as C’={I c1 ,I c2 ,…,{I ct ,I ct’},…,I cn}, and C’ is used to replace the original sequence C to become the new sequence C=C’.

[0068] The method for taking the average of each element in the sequence C is as follows: let C={I c1 ,I c2 ,…,I cn}, each element in C is a sampling value or a set of sampling values, and the average of each element I ct , 1≤t≤n, is calculated according to the following formula:

[0069]

[0070] In the above formula, b is an element in the set I ct , and B is the number of elements in I ct .

[0071] The maximum iteration number N is the number of sequences m in the set Q, and the termination iteration threshold ∈=max{||q k ||,q k ∈Q} / S, ||q k || is the modulus value of the time sequence q k , 1≤k≤m; and S is a large number set, which is 10000 in this embodiment.

[0072] In order to ensure that the areas of the charge and discharge curves are equal, the optimal time sequence C best of the typical working condition generated in this embodiment is adjusted according to the following formula:

[0073]

[0074]

[0075]

[0076]

[0077] In the above formula, I cpos_xC best The x-th element with a positive value in the array, g is C best The median is the number of positive elements, I cpos_y C best The y-th element with a negative value, h is C best The median is the number of negative elements, C pos_sum and C neg_sum C respectively best The median is the sum of elements with positive and negative values, I cpos_x_new and I cneg_y_new I cpos_x and I cpos_y The adjusted value.

[0078] This application also provides a typical operating condition generation device for energy storage batteries based on a dynamic time-averaged series algorithm, including a time series set acquisition module and a typical operating condition generation module.

[0079] The time series set acquisition module is used to obtain the time series set Q = {q1,q2,…,q} of the energy storage battery over m unit time periods T. m};

[0080] The typical operating condition generation module is used to generate any discrete time sequence q in Q. k The optimal time series for typical operating conditions is generated using a dynamic time averaging series algorithm, where 1 ≤ k ≤ m.

[0081] To ensure equal charging and discharging curve areas, the energy storage typical operating condition generation device based on the dynamic time-averaging sequence algorithm in this embodiment is further equipped with an operating condition curve adjustment module, which is used to adjust the optimal time series C of the typical operating conditions generated by the typical operating condition generation module. best Adjust the operating curves to ensure that the areas of the charge and discharge curves are equal.

[0082] Based on four months of on-site operation data of the liquid-cooled chamber of a peak-shaving energy storage project, Figure 2 The example shown illustrates typical operating conditions of energy storage obtained using the mean method. It can be seen that the operating condition curve generated by the mean method cannot reflect the constant rate operation of the energy storage battery. Figure 3 An exemplary embodiment is shown, which uses a dynamic time-averaged series algorithm to obtain a typical operating condition curve for energy storage. This curve reflects the operating characteristics of the energy storage battery and has equal charge and discharge amounts, meeting the requirements for repeated charge and discharge cycles of the operating condition curve.

[0083] Exemplary embodiments of this application have been specifically shown and described above. It should be understood that this application is not limited to the detailed structures, arrangements, or implementation methods described herein; rather, this application is intended to cover various modifications and equivalent arrangements contained within the spirit and scope of the appended claims.

Claims

1. A method for generating typical operating conditions of energy storage based on dynamic time average sequence algorithm, characterized in that, The working condition of the energy storage battery in a unit time period T forms a discrete time sequence q = {I q1 , q2 , …,I qn}, each element in q is a current value of each sampling point in the unit time period T, and n is the number of sampling points in the unit time period T; and a time sequence set of the energy storage battery in m unit time periods T is Q = {q1, q2, …, q m}, any discrete time sequence q k An optimal time sequence of a typical working condition is generated by using a dynamic time average sequence algorithm, 1≤k≤m; The dynamic time average sequence algorithm is as follows: 1) Let the iteration number iter = 1, initialize the typical operating condition optimal time series C best = C best 0 , C best n is the dimensionality, N is the maximum number of iterations, and ∈ is the termination iteration threshold. 2) Let k = 1, time series C = C best iter-1 ; 3) time series C best iter-1 with q k Generating cumulative distance matrix γ k ; 4) Starting from the (n,n) position of the matrix γ k , search for the optimal path to the (1,1) position, adding the elements passed in the optimal path to the sequence C during the search; where the search method for the (g,h) position is min{γ k (g-1,h-1), γ k (g,h-1), γ k (g-1,h)} maximum direction until reaching the (1,1) position, 1 5) If k<=m, then let k=k+1, return to step 3), otherwise enter step 6); 6) Take the average of each element of the sequence C to obtain the optimal sequence of the iter iteration C best iter ； 7) if iter≤N and C best iter the optimal sequence C best iter-1 of the (iter-1)th iteration is greater than ε, let the iteration number iter = iter+1, return to step 2), otherwise enter step 8); 8) output C best iter optimal time series C for typical working conditions best .

2. The energy storage typical operating condition generation method based on dynamic time average sequence algorithm of claim 1, wherein, Initialization of typical working condition optimal time series C best The method is: randomly taking any one time series from the set Q = {q1, q2, …, q m} as C best 0 .

3. The energy storage typical operating condition generation method based on dynamic time average sequence algorithm of claim 1, wherein, The cumulative distance matrix γ k The generation method is as follows: 1) Generate the element of the first row i = 1 and first column j = 1 of the matrix Y k using the following formula: gamma k (i,j) = d(I qi ,I cj ) where d(I qi ,I cj ) denotes the Euclidean distance between I qi and I cj . For elements of n>=i>1 or n>=j>1, the following formula is used: gamma k (i,j) = d(I qi ,I cj )+ min{gamma k (i-1,j-1), gamma k (i-1,j), gamma k (i,j-1)}; where I qi is an element of the sequence q k , 1≤i≤n, I cj is an element of the sequence C, 1≤j≤n, γ k (i,j) denotes the element in the ith row and jth column of the cumulative distance matrix γ k , γ k (i-1,j-1) denotes the element in the (i-1)th row and (j-1)th column of the cumulative distance matrix γ k , γ k (i-1,j) denotes the element in the (i-1)th row and jth column of the cumulative distance matrix γ k , γ k (i,j-1) denotes the element in the ith row and (j-1)th column of the cumulative distance matrix γ k .

4. The energy storage typical operating condition generation method based on dynamic time average sequence algorithm of claim 1, wherein, The method for adding the elements passed in the optimal path to the sequence C in the search process is: let C={I c1 ,I c2 ,…,I cn} before the first search process, if I ct’ needs to be added to the tth element I ct after the search is finished, 1≤t≤n, then the sequence C after I ct’ is added is expressed as C={I c1 ,I c2 ,…,{I ct ,I ct’},…,I cn}.

5. The energy storage typical operating condition generation method based on dynamic time average sequence algorithm of claim 1, wherein, The method of taking the average of each element of sequence C is: let C = {I c1 , c2 , cn , …, I ct , n} and the average of the tth element I ct_avg , 1≤t≤n is calculated according to the following formula: where b is the set I ct where b is the set I ct where b is the set I 6. The method of claim 1, wherein the method is based on a dynamic time average sequence algorithm. The maximum iteration number N takes the sequence number m in the set Q; the termination iteration threshold ∈ = max{||q k ||, q k ∈ Q} / S, ||q k || is the modulus value of the time sequence q k , and S is a set value.

7. The energy storage typical operating condition generation method based on dynamic time average sequence algorithm of claim 1, wherein: The method further comprises adjusting the generated typical operating condition optimal time series C best using the following formula: where I cpos_x is C best is the xth positive element in I best is the number of positive elements in I cneg_y is C best is the yth negative element in I best is the number of negative elements in I pos_sum and C neg_sum are C best is the sum of positive and negative elements in I cpos_x_new and I cneg_y_new are I cpos_x and I cneg_y is the adjusted value.

8. A dynamic time average sequence algorithm-based energy storage typical operating condition generating device, characterized in that, The method comprises a time series set acquisition module and a typical working condition generation module. The time sequence set obtaining module is configured to obtain a time sequence set Q={q1, q2, …, qm} of the energy storage battery in m unit time periods T. m}; The typical working condition generating module is configured to generate a typical working condition qkfrom any one of the discrete time sequences q in Q k generating a typical working condition optimal time sequence using the dynamic time average sequence algorithm as claimed in any one of claims 1 to 7, 1≤k≤m.

9. The energy storage typical operating condition generating device based on dynamic time average sequence algorithm according to claim 8, characterized in that, The working condition curve adjusting module is further included, and is used for adjusting the typical working condition optimal time sequence C generated by the typical working condition generating module best The working condition curve is adjusted by using the following formula: where I cpos_x is C best is the xth positive element in I best is the number of positive elements in I cneg_y is C best is the yth negative element in I best is the number of negative elements in I pos_sum and C neg_sum are C best is the sum of positive and negative elements in I cpos_x_new and I cneg_y_new are I cpos_x and I cneg_y is the adjusted value.

Images