A Planning Method and System for a Hybrid Time-Scale Hydrogen-Electricity Combined Energy Storage System

Abstract

The present invention relates to a hybrid time scale hydrogen-electricity combined energy storage system planning method, comprising: planning a daily net load net_P i,t Decomposed into daily fluctuating load #imgabs0# and daily benchmark load P i ; Obtain Ns typical daily fluctuating loads in the planning horizontal year#imgabs1# Obtain Ns typical daily benchmark loads P in the planning horizontal year k ; Obtain Ns typical daily loads P in the planning level year k,t , calculate the proportion π(k) of typical day k in the planning horizontal year; establish a hybrid time scale hydrogen-electric combined energy storage system planning model; solve and obtain the optimal planning capacity of each device in the hybrid time scale hydrogen-electric combined energy storage system. The present invention also discloses a hybrid time scale hydrogen-electric combined energy storage system planning system. The present invention establishes a hybrid time scale hydrogen-electric combined energy storage system planning model that takes into account technical and economic performance. The factors considered in the planning are closer to actual operation, and the planning results can significantly improve the technical and economic benefits of the planning.

Description

Technical Field

[0001] The present invention relates to the technical field of power system planning, and in particular to a planning method and system for a hydrogen-electric hybrid energy storage system with a hybrid time scale. Background Art

[0002] With the continuous grid connection of a high proportion of renewable energy, the power and energy imbalance problems in the power system caused by factors such as the uncertainty of its output and the large influence from the outside world are important problems faced by the current power system. At present, there are various means to adjust the power and energy imbalance between supply and demand in the power system, mainly traditional thermal power plants and hydropower plants. Traditional thermal power plants have a great impact on the environment and do not conform to the current path of sustainable development, so their proportion is gradually decreasing; while hydropower plants are greatly restricted by natural resources and geographical locations. Therefore, converting electric energy into hydrogen energy for storage has become the main flexibility adjustment resource in the current power system.

[0003] However, most of the current existing energy storage technology research mainly considers the intraday energy storage planning on a short time scale, lacking research on energy storage planning on a long time scale. And in the power system, both long-time-scale energy storage and planning technologies that cooperate with existing short-time-scale energy storage are needed.

[0004] Therefore, there is an urgent need for a planning method for a hydrogen-electric hybrid energy storage system that includes electrochemical energy storage, hydrogen energy storage (including an electrolytic water hydrogen production device, a hydrogen storage tank, and a fuel cell), and the superior power grid, forming a long-short-term collaborative energy storage strategy to provide a more scientific and engineering practical solution for the flexibility resource balance in the power system planning design, operation control, and optimal dispatching. Summary of the Invention

[0005] The primary object of the present invention is to provide a planning method for a hydrogen-electric hybrid energy storage system with a hybrid time scale that establishes a planning model considering technical economy, the factors considered in the planning are closer to the actual operation, and the planning results can significantly improve the technical and economic benefits of the planning.

[0006] To achieve the above object, the present invention adopts the following technical solutions: A planning method for a hydrogen-electric hybrid energy storage system with a hybrid time scale, the method includes the following steps in sequence:

[0007] (1) Decompose the daily net load net_P arranged in time series for the planning horizon year i,t into a daily fluctuating load and a daily reference load P i, where \(i\) represents the date sequence number within the planning horizon year, \(i = 1, 2, \ldots, N_i\), and \(N_i\) represents the number of days in the planning horizon year; \(t\) represents the time period sequence number for each day within the planning horizon year, \(t = 1, 2, \ldots, N_t\), and \(N_t\) represents the number of time periods for each day within the planning horizon year;

[0008] (2) For the daily fluctuating load obtained in step (1) Use the K - means clustering algorithm improved based on sensitive cluster centers to obtain \(N_s\) typical daily fluctuating loads within the planning horizon year

[0009] (3) For the daily reference load \(P\) obtained in step (1) i , use the weighted average method to obtain \(N_s\) typical daily reference loads \(P\) within the planning horizon year k ;

[0010] (4) Combine the \(N_s\) typical daily fluctuating loads obtained in steps (2) and (3) and the \(N_s\) typical daily reference loads \(P\) k , to obtain the \(N_s\) typical daily loads \(P\) within the planning horizon year k,t , meanwhile, calculate the proportion \(\pi(k)\) of the typical day \(k\) within the planning horizon year;

[0011] (5) Based on the data obtained in step (4), establish a hybrid - time - scale hydrogen - electricity combined energy storage system planning model for hydrogen energy storage on an annual time scale and electrochemical energy storage on a daily time scale;

[0012] (6) Input the data obtained in step (4) and the electricity price and hydrogen price data into the hybrid - time - scale hydrogen - electricity combined energy storage system planning model, and use a mixed - integer non - linear programming solver to solve it to obtain the optimal planned capacity of each device in the hybrid - time - scale hydrogen - electricity combined energy storage system.

[0013] The specific content of step (1) is as follows: Obtain the renewable energy power generation \(newP\) i,t , the electrical load \(P'\) i,t and the hydrogen load \(H\) i,t at the \(t\) - th time period of the \(i\) - th day within the planning horizon year, and convert the hydrogen load into an electrical load according to formula (1) to obtain the comprehensive electricity consumption load \(P\) i,t at the \(t\) - th time period of the \(i\) - th day within the planning horizon year:

[0014] P i,t =k h2p H i,t +P' i,t (1)

[0015] where \(k\) h2p is the hydrogen - electricity conversion coefficient;

[0016] Obtain the net load nwt_P at the planned horizontal year i,t , as shown in formula (2):

[0017] nwt_P i,t = newP i,t - P i,t (2)

[0018] Decompose the net load net_P at the planned horizontal year i,t into the intra-day fluctuating load and the intra-day reference load P i according to formula (3):

[0019]

[0020] Let the intra-day fluctuating load at the first time period of each day within the planned horizontal year be the intra-day reference load P i , that is

[0021] The specific steps of step (2) include the following steps:

[0022] (2a) Take the intra-day fluctuating load within the planned horizontal year as the sample data set, denoted as P = {P1, P2,..., P Ni}; where Calculate the Euclidean distance dist(P i and P j ) between any two data objects P i , P j :

[0023]

[0024] Use formula (5) to calculate the average distance MeanDist between data objects in the sample data set P:

[0025]

[0026] In the formula, is the number of combinations of taking 2 data objects from Ni data objects;

[0027] (2b) The area centered on P i with the average distance MeanDist as the radius is called the neighborhood of P i . At the same time, calculate the density parameter density(P i ) of P i as shown in formula (6):

[0028]

[0029] In formula (6), u(dist(P i ,P j )) represents a function, and there is:

[0030]

[0031] Obtain the density parameters of all Ni data objects; add the top M data objects with the largest density parameters to the candidate point set D, D = {D i |density(P i )}, i = 1, 2,..., M, and M is 0.8Ni;

[0032] (2c) Define a loop variable r and initialize r = 1; define the number of clusters as k c , and initialize k c = 1; define the initial maximum average similarity between clusters as AMS r-1 , and initialize AMS r-1 = 0; define the set of cluster centers as A r-1 , and initialize it as an empty set; use the candidate point set D described in step (2b) as the (r - 1)-th candidate point set D r-1 ;

[0033] (2d) Select two data objects P r-1 and P max，1 with the largest corresponding density parameters from the (r - 1)-th candidate point set D max，2 , use their respective density parameters as weights, and obtain the r-th cluster center C r according to formula (8):

[0034] C r = density(P max，1 )·P max，1 + density(P max，2 )·P max，2 (8)

[0036] Add C r to the set of cluster centers A r-1 , to obtain the r-th set of cluster centers A r ; at the same time, delete the density parameters corresponding to the two data objects P max，1 and P max，2 from the candidate point set D r-1 , to obtain the r-th candidate point set D r ;

[0037] (2e) Select a sample D r from the candidate point set D i such that the corresponding Pi is the farthest from the clustering center C r in A, and this P r is used as the (r + 1)-th clustering center C i and added to the set of clustering centers to obtain the (r + 1)-th set of clustering centers A r+1 ; meanwhile, the density parameter corresponding to this P r+1 is deleted from the set of candidate points D i to obtain the (r + 1)-th set of candidate points D r ; r+1 ;

[0038] (2f) Assign k + 1 to k;

[0039] (2g) Calculate the Euclidean distances between the data objects P r+1 corresponding to the remaining data objects in the (r + 1)-th set of candidate points D j and each clustering center in the (r + 1)-th set of clustering centers A r+1 , and assign each data object to the class where the clustering center with the closest Euclidean distance is located to obtain k clusters; use Equation (9) to obtain the average distance s i between the clustering center C i of the i-th cluster and the data object P j in the i-th cluster: i

[0040]

[0041] In Equation (9), Ki is the total number of data objects in the i-th cluster;

[0042] (2h) Use Equation (10) to obtain the distance d i between the clustering center C j of the i-th cluster and the clustering center C i,j of the j-th cluster:

[0043] d i,j = dist(C i , C j ) (10)

[0044] where i = 1, 2,..., k; j = 1, 2,..., k;

[0045] Use Equation (11) to obtain the r-th average maximum inter-class similarity AMS r :

[0046]

[0047] Meanwhile, judge whether AMS r < AMS r-1Whether it holds. If it holds, go to step (2i); otherwise, go to step (2l);

[0048] (2i) Update the cluster center C′ using Equation (12) i :

[0049]

[0050] (2j) Calculate the sum of the Euclidean distances between any data object corresponding to a density parameter in the (r + 1)-th set of candidate points D r+1 and all updated cluster centers respectively, to obtain a set of the sums of the Euclidean distances between the data objects corresponding to all density parameters and all updated cluster centers. Select the data object corresponding to the maximum value from the set of the sums of the Euclidean distances as the (r + 2)-th cluster center C r+2 and put it into the (r + 1)-th set of cluster centers A r+1 to obtain the (r + 2)-th set of cluster centers A r+2 ;

[0051] (2k) Delete the density parameter corresponding to the (r + 2)-th cluster center C r+2 from the (r + 1)-th set of candidate points D r+1 to obtain the (r + 2)-th set of candidate points D r+2 ;

[0052] (2l) After assigning r + 1 to r, go to step (2f);

[0053] (2m) Let Ns = k, then use the Ns cluster centers corresponding to AMS r-1 as the initial cluster centers of the K-means clustering algorithm to obtain Ns typical daily load fluctuations

[0054] Specifically, step (3) means: According to the clustering result of step (2), allocate the daily reference load P i in step (1) to each cluster to obtain the daily reference load P i,k of the planned horizontal year, i = 1, 2,..., Ni; k = 1, 2,..., Ns. Calculate the reference load P k of Ns typical days using the weighted average method, i = 1, 2,..., Ns:

[0055]

[0056] where: Nk is the number of data objects included in the k-th typical day, k = 1, 2,..., Ns.

[0057] The specific content of step (4) is as follows: combining the Ns typical daily fluctuating loads obtained in steps (2) and (3) and the Ns typical daily reference loads P k , to obtain the Ns typical daily loads P k,t in the planned horizon:

[0058]

[0059] where: Nk is the number of data objects included in the k-th typical day, k = 1, 2,..., Ns; t = 1, 2,..., Nt;

[0060] Meanwhile, calculate the proportion π(k) of the typical day k in the planned horizon according to formula (14):

[0061]

[0062] where: Ni represents the number of days in the planned horizon.

[0063] The specific content of step (5) includes the following steps:

[0064] (5a) Establish a hybrid time-scale hydrogen-electricity combined energy storage system planning model for hydrogen energy storage on an annual time scale and electrochemical energy storage on a daily time scale. The objective function of this model is to minimize the comprehensive annualized cost C ann of the hybrid time-scale hydrogen-electricity combined energy storage system, including equipment investment cost C inv , operation and maintenance cost C ope and curtailment penalty cost C pun . The expression is as follows:

[0065] C ann = C inv + C ope + C pun (15)

[0066] The equipment investment cost C inv is shown in formula (16):

[0067]

[0068] In the formula: x q is the capacity of the q-type equipment, c inv,q is the unit investment cost of the q-type equipment, Nq is the number of types of q-type equipment; r c is the discount rate; N pl is the planned service life of the equipment;

[0069] The operation and maintenance cost C ope is shown in formula (17):

[0070]

[0071] Where: Ns is the number of typical days; is the unit power cost of purchasing electricity from the grid at time t; is the unit power cost of selling electricity to the grid at time t; is a 0-1 state variable for purchasing electricity from the grid at time t. When the value is 0, it means not working; when the value is 1, it means working; is a 0-1 state variable for selling electricity to the grid at time t. When the value is 0, it means not working; when the value is 1, it means working;

[0072] The penalty cost C for curtailment of wind and solar power pun As shown in Equation (18):

[0073]

[0074] Where: is the curtailment of wind / solar power at time t within typical day k; is the unit cost of curtailment of wind / solar power at time t within typical day k;

[0075] (5b) The constraints of the hybrid time-scale hydrogen-electricity combined energy storage system planning model are as follows:

[0076] The constraints of the electrochemical energy storage are:

[0077]

[0078] Where, represents the charging power of the electrochemical energy storage in the t-th period within typical day k; represents the discharging power of the electrochemical energy storage in the t-th period within typical day k; represents the rated power of the electrochemical energy storage; is a 0-1 state variable for charging the electrochemical energy storage in the t-th period within typical day k. When the value is 0, it means not working; when the value is 1, it means working; is a 0-1 state variable for discharging the electrochemical energy storage in the t-th period within typical day k. When the value is 0, it means not working; when the value is 1, it means working;

[0079] The energy state level of the electrochemical energy storage at different times within typical day k is expressed as:

[0080]

[0081] Where, represents the energy state level of the electrochemical energy storage in the (t + 1)-th period within typical day k; represents the initial energy state level of the electrochemical energy storage within typical day k; η EESIndicates the charge-discharge power coefficient of the electrochemical energy storage; Δt E is the operating time interval of the electrochemical energy storage within the typical day k; t′ represents the t′-th period, with values ranging from 1 to Nt, and Nt is taken as 24 h;

[0082] The energy state level of the electrochemical energy storage at the initial state of each day should be equal to the energy state level at the initial state of the next day, that is, the energy state levels of charging and discharging of the electrochemical energy storage device within the typical day k are equal, which is expressed as:

[0083]

[0084] The interaction power between the superior power grid and the hybrid time-scale hydrogen-electricity combined energy storage system is less than or equal to the maximum electric power of the power tie line:

[0085]

[0086] In the formula, is the purchased electric power of the power system in the t-th period within the typical day k; is the sold electric power of the power system in the t-th period within the typical day k; is the maximum interaction power of the power tie line between the superior power grid and the hydrogen-electricity combined energy storage system; is the 0-1 state variable of the power system's power purchase in the t-th period within the typical day k. When the value is 0, it means not working, and when the value is 1, it means working; is the 0-1 state variable of the power system's power sale in the t-th period within the typical day k. When the value is 0, it means not working, and when the value is 1, it means working;

[0087] The upper and lower limit constraints of the input power of the electrolytic water hydrogen production device in each period are:

[0088]

[0089] In the formula, represents the input power of the electrolytic water hydrogen production device in the t-th period within the typical day k; represents the rated power of the electrolytic water hydrogen production device; represents the operating state of the electrolytic water hydrogen production device in the t-th period within the typical day k, where 1 means operating and 0 means stopped;

[0090] The start-up and shut-down time constraints of the electrolytic water hydrogen production device are expressed as:

[0091]

[0092] In the formula, represents the operating state of the electrolytic water hydrogen production device in the (t - 1)-th period within the typical day k; It represents the operating state of the electrolytic water hydrogen production device in the k-th period within a typical day k; TO represents the minimum start-up time allowed for the electrolytic water hydrogen production device; TS represents the minimum shutdown time allowed for the electrolytic water hydrogen production device;

[0093] After linearizing the upper and lower limit constraints of the output power of the fuel cell in each period using the big M method, we get:

[0094]

[0095] In the formula, It represents the output power of the fuel cell in the t-th period within a typical day k; It represents the rated power of the fuel cell; It represents the operating state of the fuel cell in the t-th period within a typical day k;

[0096] The continuous start-up and shutdown duration constraints of the fuel cell are:

[0097]

[0098] In the formula, It represents the operating state of the fuel cell in the (t - 1)-th period within a typical day k;

[0099] The energy state levels of the hydrogen storage tank at different times within a typical day k are expressed as:

[0100]

[0101] In the formula, It represents the energy state level of the hydrogen storage tank in the (t + 1)-th period within a typical day k; It represents the initial energy state level of the hydrogen storage tank within a typical day k; η ED It represents the power coefficient of the electrolytic water hydrogen production device; η FC It represents the power coefficient of the fuel cell; Δt H Is the operating time interval of the hydrogen energy storage;

[0102] The energy state levels of hydrogen charging and discharging of the hydrogen storage tank are equal within a year, which is expressed as:

[0103]

[0104] The power balance constraint is satisfied in the t-th period within a typical day k:

[0105]

[0106] In the formula, They respectively represent the wind power output and photovoltaic power output in the t-th period within a typical day k.

[0107] Another object of the present invention is to provide a system for a method of planning a hydrogen-electric hybrid energy storage system with a mixed time scale, including:

[0108] An electrolytic water hydrogen production device, which uses water as a raw material. Under the action of direct current, water is decomposed into hydrogen and oxygen. The separated hydrogen is controlled by a regulating valve and output, and sent to a hydrogen storage tank;

[0109] An electrochemical energy storage device, which converts electrical energy into chemical energy for storage and converts chemical energy into electrical energy;

[0110] A hydrogen fuel cell, which converts the chemical reaction of hydrogen energy into electrical energy;

[0111] A hydrogen storage tank, which is a pressure vessel made of metal materials such as stainless steel and aluminum alloy and is used to store hydrogen;

[0112] When there is surplus power in the power grid, the electrical energy directly charges the electrochemical energy storage device to convert the electrical energy into chemical energy for storage; or, the electrical energy is converted into hydrogen by the electrolytic water device and stored in the hydrogen storage tank;

[0113] When the power grid power is insufficient, the electrochemical energy storage device directly converts chemical energy into electrical energy and releases it to the power grid, or the hydrogen storage tank converts hydrogen into electrical energy through a hydrogen fuel cell and releases it to the power grid.

[0114] From the above technical solutions, the beneficial effects of the present invention are as follows: First, in view of the multi-time scale characteristics of the flexibility requirements in a power system with a high proportion of renewable energy, the present invention constructs a hydrogen-electric hybrid energy storage system with a mixed time scale and establishes a planning model for the hydrogen-electric hybrid energy storage system with a mixed time scale considering technical economy. Compared with the existing single-time scale energy storage system planning method, the factors considered in the planning are closer to the actual operation, and the planning results can significantly improve the technical and economic benefits of the planning; Second, the improved K-means clustering algorithm based on sensitive clustering centers adopted by the present invention divides the outlier data into a separate cluster, reducing the influence of the outlier data on the cluster division result, avoiding the clustering center moving away from the data dense area and tending to the outlier data, and improving the quality of clustering; Third, the present invention decomposes the load data in the planning horizon year into intra-day fluctuating load and intra-day reference load by time series, and adopts different processing methods for the two different loads: for the intra-day fluctuating load, the improved K-means clustering algorithm based on sensitive clustering centers is used, and the optimal number of clusters k is automatically determined by comparing the value of the average maximum inter-class similarity index AMS, solving the problem that the k value needs to be given in advance in the traditional k-means algorithm, improving the quality of clustering, and thus well solving the problem of time series coupling of multiple operation scenarios in the hydrogen-electric hybrid energy storage system in the long time scale. Description of the Drawings

[0115] Figure 1It is the flowchart of the method of the present invention;

[0116] Figure 2 It is the schematic diagram of the typical daily load calculation process;

[0117] Figure 3 It is the flowchart of the improved K - means clustering algorithm for sensitive clustering centers in the present invention;

[0118] Figure 4 It is the schematic diagram of the hydrogen - electric hybrid energy storage system in the present invention;

[0119] Figure 5(a) 、 5(b) 、5(c) respectively and one - to - one correspond to the wind, light, and load power time - series diagrams of Region 1 in Embodiment 1;

[0120] Figure 6(a) 、 6(b) 、6(c) respectively and one - to - one correspond to the wind, light, and load power time - series diagrams of Region 2 in Embodiment 1;

[0121] Figure 7 It is the charge - discharge power time - series diagram of Region 1 in Embodiment 1;

[0122] Figure 8 It is the hydrogen charge - discharge power time - series diagram of Region 1 in Embodiment 1;

[0123] Figure 9 It is the energy state level diagram of the hydrogen storage tank in Region 1 in Embodiment 1;

[0124] Figure 10 It is the purchased - sold electric power time - series diagram of Region 1 in Embodiment 1;

[0125] Figure 11 It is the charge - discharge power time - series diagram of Region 2 in Embodiment 1;

[0126] Figure 12 It is the hydrogen charge - discharge power time - series diagram of Region 2 in Embodiment 1;

[0127] Figure 13 It is the energy state level diagram of the hydrogen storage tank in Region 2 in Embodiment 1;

[0128] Figure 14 It is the purchased - sold electric power time - series diagram of Region 2 in Embodiment 1. Specific implementation manner

[0129] As Figure 1 shown, a planning method for a hydrogen - electric hybrid energy storage system with a hybrid time scale, the method includes the following steps in sequence:

[0130] (1) Decompose the daily net load net_P arranged in time series for the planned horizontal year i,t into daily fluctuating load and the daily reference load P i , where i represents the date sequence number within the planning horizon year, i = 1, 2, …, Ni, and Ni represents the number of days in the planning horizon year; t represents the time period sequence number of each day within the planning horizon year, t = 1, 2, ..., Nt, and Nt represents the number of time periods of each day within the planning horizon year;

[0131] (2) For the daily fluctuating load obtained in step (1) Use the improved K - means clustering algorithm based on sensitive cluster centers to obtain Ns typical daily fluctuating loads within the planning horizon year

[0132] (3) For the daily reference load P obtained in step (1) i , use the weighted average method to obtain Ns typical daily reference loads P within the planning horizon year k ;

[0133] (4) Combine the Ns typical daily fluctuating loads obtained in steps (2) and (3) and the Ns typical daily reference loads P k to obtain Ns typical daily loads P within the planning horizon year k,t , and at the same time, calculate the proportion π(k) of the typical day k within the planning horizon year;

[0134] (5) Based on the data obtained in step (4), establish a hybrid - time - scale hydrogen - electricity combined energy storage system planning model for hydrogen energy storage on an annual time scale and electrochemical energy storage on a daily time scale;

[0135] (6) Input the data obtained in step (4) and the electricity price and hydrogen price data into the hybrid - time - scale hydrogen - electricity combined energy storage system planning model, and use a mixed - integer non - linear programming solver to solve it to obtain the optimal planning capacity of each device in the hybrid - time - scale hydrogen - electricity combined energy storage system.

[0136] The specific content of step (1) is: Obtain the renewable energy power generation newP i,t , the electrical load P′ i,t and the hydrogen load H i,t at the t - th time period of the i - th day within the planning horizon year, convert the hydrogen load into an electrical load according to formula (1) to obtain the comprehensive electrical load P i,t at the t - th time period of the i - th day within the planning horizon year:

[0137] P i,t = k h2p H i,t + P′ i,t (1)

[0138] where k h2pis the hydrogen-electricity conversion coefficient;

[0139] Obtain the net load net_P for the planned horizontal year i,t , as shown in formula (2):

[0140] net_P i,t = newP i,t - P i,t (2)

[0141] Decompose the net load net_P for the planned horizontal year i,t into the intra-day fluctuating load and the intra-day reference load P i according to formula (3):

[0142]

[0143] Let the intra-day fluctuating load at the first time period of each day within the planned horizontal year be the intra-day reference load P i , that is

[0144] As Figure 2 , 3 shown, the said step (2) specifically includes the following steps:

[0145] (2a) Take the intra-day fluctuating load within the planned horizontal year as the sample data set, denoted as P = {P1, P2,..., P Ni}; where Calculate the Euclidean distance dist(P i and P j ) between any two data objects: i , P j ):

[0146]

[0147] Use formula (5) to calculate the average distance MeanDist between data objects in the sample data set P:

[0148]

[0149] In the formula, is the combination number of taking 2 data objects from Ni data objects;

[0150] (2b) The area centered on P i with the average distance MeanDist as the radius is called the neighborhood of P i , and at the same time, calculate the density parameter density(P i ) of P i)As shown in formula (6):

[0151]

[0152] In formula (6), u(dist(P i ,P j )) represents a function, and there is:

[0153]

[0154] Obtain the density parameters of all Ni data objects; add the top M data objects with the largest density parameters to the candidate point set D, D = {D i |density(P i )}, i = 1, 2,..., M, and M is 0.8Ni;

[0155] (2c) Define a loop variable r and initialize r = 1; define the number of clusters as k c , and initialize k c = 1; define the initial maximum average inter-class similarity as AMS r-1 , and initialize AMS r-1 = 0; define the set of cluster centers as A r-1 , and initialize it as an empty set; use the candidate point set D described in step (2b) as the (r - 1)-th candidate point set D r-1 ;

[0156] (2d) Select two data objects P r-1 and P max，1 with the largest corresponding density parameters from the (r - 1)-th candidate point set D max，2 , use their respective density parameters as weights, and obtain the r-th cluster center C r according to formula (8):

[0157] C r = density(P max，1 )·P max，1 + density(P max，2 )·P max，2 (8)

[0159] Add C r to the cluster center A r-1 set to obtain the r-th cluster center set A r ; at the same time, delete the density parameters corresponding to the two data objects P max，1 and P max，2 from the candidate point set D r-1 to obtain the r-th candidate point set D r ;

[0160] (2e) Select a sample \(D\) from the alternative point set \(D\). r such that the corresponding \(P\) i has the farthest distance from the cluster center \(C\) i in \(A\) r ; and take this \(P\) r as the \((r + 1)\)-th cluster center \(C\) i and add it to the cluster center set to obtain the \((r + 1)\)-th cluster center set \(A\) r+1 ; meanwhile, delete the density parameter corresponding to this \(P\) r+1 from the alternative point set \(D\) i to obtain the \((r + 1)\)-th alternative point set \(D\) r ; r+1

[0161] (2f) Assign \(k + 1\) to \(k\).

[0162] (2g) Calculate the Euclidean distances between the data objects \(P\) r+1 corresponding to the remaining data objects in the \((r + 1)\)-th alternative point set \(D\) j and each cluster center in the \((r + 1)\)-th cluster center set \(A\) r+1 , and assign each data object to the class where the cluster center with the closest Euclidean distance is located to obtain \(k\) clusters; use Equation (9) to obtain the average distance \(S\) i between the cluster center \(C\) of the \(i\)-th class and the data objects i in the \(i\)-th class:

[0163]

[0164] In Equation (9), \(K_i\) is the total number of data objects in the \(i\)-th class;

[0165] (2h) Use Equation (10) to obtain the distance \(d\) i between the cluster center \(C\) j of the \(i\)-th class and the cluster center \(C\) i,j of the \(j\)-th class:

[0166] d i,j = dist(\(C\) i , \(C\) j ) (10)

[0167] where \(i = 1, 2, \cdots, k\); \(j = 1, 2, \cdots, k\);

[0168] Use Equation (11) to obtain the \((r)\)-th average maximum inter-class similarity \(AMS\) r :

[0169]

[0170] Meanwhile, judge AMS r <AMS r-1 If it holds, go to step (2i); otherwise, go to step (2l);

[0171] (2i) Update the clustering center C′ using Equation (12) i :

[0172]

[0173] (2j) Calculate the sum of the Euclidean distances between any data object corresponding to a density parameter in the (r + 1)-th candidate point set D r+1 and all updated clustering centers respectively, to obtain a set of the sums of the Euclidean distances between the data objects corresponding to all density parameters and all updated clustering centers. Select the data object corresponding to the maximum value from the set of the sums of the Euclidean distances as the (r + 2)-th clustering center C r+2 and put it into the (r + 1)-th clustering center set A r+1 to obtain the (r + 2)-th clustering center set A r+2 ;

[0174] (2k) Delete the density parameter corresponding to the (r + 2)-th clustering center C r+2 from the (r + 1)-th candidate point set D r+1 to obtain the (r + 2)-th candidate point set D r+2 ;

[0175] (2l) After assigning r + 1 to r, go to step (2f);

[0176] (2m) Let Ns = k, then use the Ns clustering centers corresponding to AMS r-1 as the initial clustering centers of the K-means clustering algorithm to obtain Ns typical daily load fluctuations

[0177] As Figure 2 、 3 shown, step (3) specifically refers to: according to the clustering results of step (2), allocate the daily reference load P i in step (1) to each cluster to obtain the daily reference load P i,k in the planned horizontal year, i = 1, 2,..., Ni; k = 1, 2,..., Ns, and calculate the reference load P k of Ns typical days using the weighted average method, i = 1, 2,..., Ns:

[0178]

[0179] Where: Nk is the number of data objects included in the k-th typical day, k = 1, 2, ..., Ns.

[0180] The specific content of step (4) is as follows: Combining the fluctuating loads of Ns typical days and the reference loads P of Ns typical days obtained in steps (2) and (3), the load P of Ns typical days within the planned horizon is obtained: k k,t :

[0181]

[0182] Where: Nk is the number of data objects included in the k-th typical day, k = 1, 2, ..., Ns; t = 1, 2, ..., Nt;

[0183] Meanwhile, calculate the proportion π(k) of the typical day k within the planned horizon according to formula (14):

[0184]

[0185] Where: Ni represents the number of days in the planned horizon.

[0186] The specific content of step (5) includes the following steps:

[0187] (5a) Establish a hybrid time-scale hydrogen-electricity combined energy storage system planning model for hydrogen energy storage on the annual time scale and electrochemical energy storage on the daily time scale. The objective function of this model is to minimize the comprehensive annualized cost C of the hybrid time-scale hydrogen-electricity combined energy storage system, ann including equipment investment cost C inv , operation and maintenance cost C ope and curtailment penalty cost C pun , and the expression is as follows:

[0188] C ann = C inv + C ope + C pun (15)

[0189] The equipment investment cost C inv is shown in formula (16):

[0190]

[0191] In the formula: x q is the capacity of the q-type equipment, c inv,q is the unit investment cost of the q-type equipment, Nq is the number of types of q-type equipment; r c is the discount rate; N pl is the planned service life of the equipment;

[0192] The operation and maintenance cost of the hybrid time-scale hydrogen-electric energy storage system needs to comprehensively consider the load demand and new energy output under different typical days, as well as the cost generated by energy trading through the external power grid. The operation and maintenance cost C ope As shown in Equation (17):

[0193]

[0194] In the formula: Ns is the number of typical days; is the unit power cost of purchasing electricity from the power grid at time t; is the unit power cost of selling electricity to the power grid at time t; is the 0-1 state variable of purchasing electricity from the power grid at time t. When the value is 0, it means not working; when the value is 1, it means working; is the 0-1 state variable of selling electricity to the power grid at time t. When the value is 0, it means not working; when the value is 1, it means working;

[0195] To rationally configure the wind power and photovoltaic capacity and improve the proportion of new energy utilization, the penalty cost for curtailed wind and curtailed light of the hybrid time-scale hydrogen-electric energy storage system is set. The penalty cost for curtailed wind and curtailed light C pun As shown in Equation (18):

[0196]

[0197] In the formula: is the curtailed wind / light power at time t in typical day k; is the unit cost of curtailed wind / light at time t in typical day k;

[0198] (5b) The constraint conditions of the hybrid time-scale hydrogen-electric energy storage system planning model are as follows:

[0199] Through charge and discharge in different time periods, the electrochemical energy storage can achieve intra-day coordination of electric energy. Its operation needs to meet the relevant constraints of power and energy in different time periods. After linearizing it with the big M method, the constraint conditions of the electrochemical energy storage are obtained as:

[0200]

[0201] In the formula, represents the charging power of the electrochemical energy storage in the t-th time period of typical day k; represents the discharging power of the electrochemical energy storage in the t-th time period of typical day k; represents the rated power of the electrochemical energy storage; represents the 0-1 state variable of charging the electrochemical energy storage in the t-th time period of typical day k. When the value is 0, it means not working; when the value is 1, it means working; A 0-1 state variable representing the discharge of the electrochemical energy storage in the t-th period within a typical daily k. When the value is 0, it means not working; when the value is 1, it means working.

[0202] The energy state levels of the electrochemical energy storage at different times within a typical daily k are expressed as:

[0203]

[0204] In the formula, Represents the energy state level of the electrochemical energy storage in the (t + 1)-th period within a typical daily k; Represents the initial energy state level of the electrochemical energy storage in a typical daily k; η EES Represents the charge-discharge power coefficient of the electrochemical energy storage; Δt E Is the operating time interval of the electrochemical energy storage within a typical daily k; t′ represents the t′-th period, with values ranging from 1 to Nt, and Nt takes 24 h;

[0205] The energy state level of the electrochemical energy storage at the initial state of each day should be equal to the energy state level at the initial state of the next day, that is, the energy state levels of charging and discharging of the electrochemical energy storage device within a typical daily k are equal, expressed as:

[0206]

[0207] The interaction power between the superior power grid and the hybrid time-scale hydrogen-electricity combined energy storage system is less than or equal to the maximum electric power of the power tie line:

[0208]

[0209] In the formula, Is the purchased electric power of the power system in the t-th period within a typical daily k; Is the sold electric power of the power system in the t-th period within a typical daily k; Is the maximum interaction power of the power tie line between the superior power grid and the hydrogen-electricity combined energy storage system; Is a 0-1 state variable for the power system to purchase electricity in the t-th period within a typical daily k. When the value is 0, it means not working; when the value is 1, it means working; Is a 0-1 state variable for the power system to sell electricity in the t-th period within a typical daily k. When the value is 0, it means not working; when the value is 1, it means working;

[0210] The upper and lower limit constraints of the input power of the electrolytic water hydrogen production device in each period are:

[0211]

[0212] In the formula, Represents the input power of the electrolytic water hydrogen production device in the t-th period within a typical daily k; Represents the rated power of the electrolytic water hydrogen production device; Represents the operating state of the electrolytic water hydrogen production device in the t-th period within the typical day k, 1 represents operating, and 0 represents stopped;

[0213] The start-up and shut-down time constraints of the electrolytic water hydrogen production device are expressed as:

[0214]

[0215] In the formula, Represents the operating state of the electrolytic water hydrogen production device in the (t - 1)-th period within the typical day k; Represents the operating state of the electrolytic water hydrogen production device in the k-th period within the typical day k; TO represents the minimum start-up time allowed for the electrolytic water hydrogen production device; TS represents the minimum shut-down time allowed for the electrolytic water hydrogen production device;

[0216] After linearization of the upper and lower limit constraints of the output power of the fuel cell in each period by the big M method, we get:

[0217]

[0218] In the formula, Represents the output power of the fuel cell in the t-th period within the typical day k; Represents the rated power of the fuel cell; Represents the operating state of the fuel cell in the t-th period within the typical day k;

[0219] The continuous start-up and shut-down duration constraints of the fuel cell are:

[0220]

[0221] In the formula, Represents the operating state of the fuel cell in the (t - 1)-th period within the typical day k;

[0222] The energy state levels of the hydrogen storage tank at different times within the typical day k are expressed as:

[0223]

[0224] In the formula, Represents the energy state level of the hydrogen storage tank in the (t + 1)-th period within the typical day k; Represents the initial energy state level of the hydrogen storage tank within the typical day k; η ED Represents the power coefficient of the electrolytic water hydrogen production device; η FC Represents the power coefficient of the fuel cell; Δt H Is the operating time interval of the hydrogen energy storage;

[0225] The hydrogen charging and discharging energy state levels of the hydrogen storage tank are equal within one year, which is expressed as:

[0226]

[0227] During the t-th period of a typical day k, the power balance constraint is satisfied:

[0228]

[0229] In the formula, respectively represent the wind power output and photovoltaic power output during the t-th period of a typical day k.

[0230] As Figure 4 shown, this system includes:

[0231] An electrolytic water hydrogen production device, which uses water as a raw material. Under the action of direct current, water is decomposed into hydrogen and oxygen. The separated hydrogen is controlled by a regulating valve and output, and sent to a hydrogen storage tank;

[0232] An electrochemical energy storage device, which converts electrical energy into chemical energy for storage and converts chemical energy into electrical energy;

[0233] A hydrogen fuel cell, which converts the chemical reaction of hydrogen energy into electrical energy;

[0234] A hydrogen storage tank, which is a pressure vessel made of metal materials such as stainless steel and aluminum alloy and is used to store hydrogen;

[0235] When there is surplus power in the power grid, the electrical energy directly charges the electrochemical energy storage device to convert the electrical energy into chemical energy for storage; or, the electrical energy is converted into hydrogen by the electrolytic water device and stored in the hydrogen storage tank;

[0236] When the power grid lacks power, the electrochemical energy storage device directly converts chemical energy into electrical energy and releases it to the power grid, or the hydrogen storage tank converts hydrogen into electrical energy through a hydrogen fuel cell and releases it to the power grid.

[0237] Example 1

[0238] For this example, first input the wind power (photovoltaic power) output and load demand for 8760 hours throughout the year of the predicted planning reference year, and cluster them using this method to obtain the net load demand of a typical day. Two different regions in the south (Region 1) and north (Region 2) of China are selected for demonstration respectively, and the source data of each is as follows: the installed capacities of wind and light are both 200 MW, and the electrical load (including the equivalent hydrogen load) is 300 MW; the maximum allowable power of the power connection line between the superior power grid and the electricity-hydrogen coupling system is 80 MW. Since the loads in these two regions have relatively obvious seasonal characteristics, the clustering results are both 4 categories, and the time series diagrams of wind, light, and load of a typical day in the two regions as shown in Figures 5(a), 5(b), 5(c), 6(a), 6(b), and 6(c) are obtained.

[0239] Meanwhile, substitute its data into the electric-hydrogen coupling system described in this paper for planning and optimization. The specific planning operation parameters are shown in Table (a) and Table (b). Table (a) shows the relevant parameters of each energy storage device (electrochemical energy storage, electrolytic water hydrogen production device, hydrogen storage tank, and fuel cell), such as: unit power / capacity investment cost, service life, discount rate, unit power operation and maintenance cost, unit start-stop cost, and efficiency coefficient, etc.; Table (b) shows the electricity prices for buying and selling electricity from the superior power grid. It can be found that from 9:00 to 11:00 in the morning and from 19:00 to 23:00 in the evening, the buying and selling electricity prices of the superior power grid are the highest. This is because this is the period when user electricity consumption climbs and drops. To maintain the power balance of the power grid, the power system needs to buy and sell electricity from the superior power grid. While from 0:00 to 8:00 in the morning, user electricity consumption is relatively stable, causing less impact on the power grid. Therefore, the buying and selling electricity prices of the superior power grid are the lowest at this time.

[0240] Table (a) Table of relevant parameters of energy storage devices

[0241]

[0242] Table (b) Table of electricity prices of the superior power grid at different time periods

[0243]

[0244] According to the above parameters, this method conducts planning simulation analysis on the MATLAB R2018a platform with YALMIP / CPLEX12.8. The simulation results and analysis under two different scenarios are as follows.

[0245] Analysis of the planning results in Region 1:

[0246] Through MATLAB simulation, the power time series diagrams of each energy storage device can be obtained. Figure 7 is the charge and discharge power time series diagram of the electrochemical energy storage for one year, Figure 8 is the hydrogen charge and discharge power time series diagram of the hydrogen energy storage system for one year, Figure 9 is the energy state level diagram of the hydrogen storage tank at different times of one year, Figure 10 is the power time series diagram of buying and selling electricity from the superior power grid by the electric-hydrogen coupling system within one year. At the same time, the configured power / capacity and total cost of each energy storage device are shown in Table (c).

[0247] Table (c) Summary table of the configured power / capacity and total cost of energy storage in Region 1

[0248]

[0249] By observing the data in Table (c), it can be found that as an energy storage device for long-term energy storage, the configured capacity of the hydrogen storage tank is much larger than that of the electrochemical energy storage device for short-term energy storage. In this scenario, the configured capacity of the hydrogen storage tank is about 1000 times that of the electrochemical energy storage device. Moreover, when the hydrogen storage tank is in the initial state and the electrolyzer is operating at its maximum output power, hydrogen can be continuously filled for 371 hours. When the hydrogen storage tank is in the initial state and the fuel cell is operating at its maximum output power, hydrogen can be continuously discharged for 296 hours. In addition, when all other power sources (i.e., the upstream power grid, wind and solar power generation, and electrochemical energy storage) except hydrogen energy storage in the system are ineffective, the hydrogen energy storage system can continuously supply electrical energy to the load for 80 hours. At this time, it is equivalent to that the hydrogen energy storage system can support the power supply in this area for 80 consecutive hours without any support.

[0250] While the electrochemical energy storage can be continuously charged and discharged for about 2 hours when it is almost out of power or fully charged. It can be seen that the addition of hydrogen energy storage can greatly improve the absorption level of wind and solar energy in the power system. It constructs a long-term and short-term collaborative energy storage mechanism with electrochemical energy storage, making use of both the advantages of rapid charge and discharge of electrochemical energy storage and the characteristics of hydrogen energy storage to achieve long-term energy balance. However, due to the low efficiency coefficients of the electrolyzer and fuel cell in the hydrogen energy storage system, as well as the high upfront investment cost and subsequent operation and maintenance cost, currently, hydrogen energy storage is not the best cost-effective choice for the power grid. But with the improvement of the manufacturing process of electrolyzers and fuel cells and the progress of technology, the development prospect of hydrogen energy storage is still very promising.

[0251] In Figure 7 it can be found that the sum of the charging and discharging energies of the electrochemical energy storage device in each typical day is 0 within a day, that is, the charging and discharging energies within a day are equal. In addition, the duration of each charge and discharge of the electrochemical energy storage device is relatively short, which conforms to the positioning of short-term energy storage in this article.

[0252] In Figure 8 it can be found that the sum of the hydrogen charging and discharging powers of the hydrogen storage tank in each typical day is not 0. This is because the weight coefficient is not multiplied at this time. After multiplying the weight coefficient, the sum of the hydrogen charging and discharging energies of the hydrogen storage tank within a year is 0, as Figure 9 shown, that is, within a year, the hydrogen charging and discharging energies of the hydrogen storage tank are equal. Although the energy of each hydrogen charge and discharge is different, the total within a year is 0. In addition, the hydrogen energy storage system can maintain a relatively long time for hydrogen charging and discharging, forming a long-term and short-term collaborative energy storage mechanism with electrochemical energy storage.

[0253] At the same time, it can also be seen that the hydrogen energy storage system stores the remaining energy of the power grid in summer and autumn, and converts hydrogen energy into electrical energy and sends it to the power grid when the wind and solar power generation are insufficient to meet the load demand in spring and winter. The detailed analysis is as follows:

[0254] (1) In spring, from 1:00 to 6:00, it is the cheapest time of the day to purchase electricity from the superior power grid, and the load demand is small. At this time, the load demand can be met by purchasing electricity from the superior power grid. Therefore, the excess wind power output can be stored through the hydrogen energy storage system. From 8:00 to 10:00, 14:00, and from 18:00 to 22:00, it is more expensive to purchase electricity from the superior power grid. Although there is wind and light output, the load demand is large and it is not enough to meet the load demand. Therefore, to ensure cost minimization, while purchasing a part of electricity from the superior power grid, the hydrogen energy stored in the hydrogen energy storage system is converted into electric energy and sent to the power grid to ensure the safe and stable operation of the power grid;

[0255] (2) In summer, at 1:00 to 3:00, 6:00, and 23:00, it is the cheapest time of the day to purchase electricity from the superior power grid, and the load demand is small. At this time, the load demand can be met by purchasing electricity from the superior power grid. Therefore, the excess wind power output can be stored through the hydrogen energy storage system. From 9:00 to 11:00 and from 18:00 to 20:00, it is more expensive to purchase electricity from the superior power grid. Although there is a certain amount of wind and light output but it is not enough to meet the load demand. Therefore, while purchasing a part of electricity from the superior power grid, hydrogen is released from the hydrogen storage tank and converted into electric energy through a fuel cell and transmitted to the power grid to ensure the power and electricity balance of the power grid;

[0256] (3) In autumn, from 23:00 to 6:00, it is the cheapest time of the day to purchase electricity from the superior power grid, and the load demand is small. At this time, the load demand can be met by purchasing electricity from the superior power grid. Therefore, the excess wind power output can be stored through the hydrogen energy storage system. At 22:00, it is more expensive to purchase electricity from the superior power grid. Although there is a certain amount of wind and light output but it is not enough to meet the load demand. Therefore, while purchasing a part of electricity from the superior power grid, hydrogen is released from the hydrogen storage tank and converted into electric energy through a fuel cell and transmitted to the power grid to ensure the power and electricity balance of the power grid;

[0257] (4) In winter, from 0:00 to 3:00, it is the cheapest time of the day to purchase electricity from the superior power grid, and the load demand is small. At this time, the load demand can be met by purchasing electricity from the superior power grid. Therefore, the excess wind power output can be stored through the hydrogen energy storage system. From 4:00 to 10:00 and from 17:00 to 21:00, although it is more expensive to purchase electricity from the superior power grid at this time, the wind and light output is not much and is not enough to meet the load demand. Therefore, while purchasing the maximum amount of electricity from the superior power grid, hydrogen also needs to be released from the hydrogen storage tank to obtain electric energy through the fuel of the fuel cell. From 11:00 to 16:00 and from 22:00 to 23:00, while purchasing the maximum allowable amount of electricity from the superior power grid, there is still a surplus of wind and light output after meeting the load demand. Therefore, energy can be stored in the hydrogen storage tank.

[0258] exist Figure 9 It can be found that in a year, the hydrogen energy storage system releases hydrogen in spring and winter and converts it into electricity through fuel cells to supply the grid to meet the load demand, while in summer and autumn, the water electrolysis hydrogen production device electrolyzes water to convert electricity into hydrogen energy and stores it in the hydrogen storage tank. The energy of the hydrogen storage tank at the beginning of the year is equal to the energy of the hydrogen storage tank at the end of the year. This is because the sum of the energy of charging and discharging hydrogen in the hydrogen storage tank in a year is 0;

[0259] In addition, the present invention extends the energy state level in the hydrogen storage tank on four typical days to the energy state level in the hydrogen storage tank within a year, so the four amplified curves are the change amplitudes of the energy state level in the hydrogen storage tank on four typical days, that is, they represent the change amplitudes of the energy state level in the hydrogen storage tank under different seasons. At the same time, it can be found that the energy state level curve of a typical day in spring has a downward trend, and after extending to a season, its energy state level curve also has a downward trend, and the same is true for winter; while in summer, its energy state level curve has an upward trend, and the same is true for autumn. This is because the change amplitude of the energy state level of a typical day represents the change amplitude of the energy state level of each day in a season, and the change amplitude of each day is consistent, so the two have the same change trend.

[0260] It can also be found that the rising and falling rates of each season are inconsistent, because the energy of hydrogen charging and discharging is different every day in each season; in addition, from the turning point of the curve, it can be clearly seen that the summer in a certain area in the south occupies a longer time in a year, while spring, autumn and winter are evenly distributed, and the sum of the proportions of the three is consistent with the proportion of summer.

[0261] Figure 10 In the process, since wind and solar power output accounts for 82% of the total source output, it can be sold to the upper-level grid when there is excess wind and solar power output, and sold to the upper-level grid when there is insufficient wind and solar power output. The stable and reliable operation of the power grid must be guaranteed at all times, and the amount of electricity purchased and sold in each period depends on the wind and solar power output, load power, electrochemical energy storage power, hydrogen energy storage power and the constraints of each device.

[0262] Analysis of results for Region 2:

[0263] Through MATLAB simulation, the power timing diagram of each energy storage device can be obtained. Figure 11 This is the charging and discharging power timing diagram of electrochemical energy storage in one year. Figure 12 This is the power sequence diagram of hydrogen charging and discharging in a hydrogen energy storage system within one year. Figure 13 This is the energy status level diagram of the hydrogen storage tank at different times of the year. Figure 14 This is the power sequence diagram of the electricity-hydrogen coupling system purchasing and selling electricity to the upper grid within one year. At the same time, the configuration capacity and total cost of each energy storage device are shown in Table (d).

[0264] Table (d) Summary Table of Energy Storage Configuration Power / Capacity and Total Cost in Region 2

[0265]

[0266]

[0267] Observing the data in Table (d), it can be found that as an energy storage device for long - time - scale energy storage, the configured capacity of the hydrogen storage tank is much larger than that of the electrochemical energy storage device for short - time - scale energy storage. In this scenario, the configured capacity of the hydrogen storage tank is about 1000 times that of the electrochemical energy storage device. Moreover, when the hydrogen storage tank is in the initial state and the electrolyzer is at the maximum output power, it can continuously charge hydrogen for 515 hours. When the hydrogen storage tank is in the initial state and the fuel cell is at the maximum output power, it can continuously discharge hydrogen for 185 hours. In addition, when all other power sources (i.e., the upstream power grid, wind and solar power output, and electrochemical energy storage) in the system except for hydrogen energy storage are invalid, the hydrogen energy storage system can continuously supply electrical energy to the load for 38 hours. At this time, it is equivalent to that the hydrogen energy storage system can support the power supply in this area continuously for 38 hours without any support.

[0268] The electrochemical energy storage can be continuously charged and discharged for about 2 hours when it is almost out of power or fully charged. Thus, it can be seen that the addition of hydrogen energy storage can greatly improve the wind and solar power consumption level in the power system. It constructs a long - term and short - term collaborative energy storage mechanism with the electrochemical energy storage, making use of both the advantage of the rapid charge and discharge of the electrochemical energy storage and the characteristic of the hydrogen energy storage to achieve long - time - scale power balance. However, because the efficiency coefficients of the electrolyzer and fuel cell in the hydrogen energy storage system are not high, and the upfront investment cost and later operation and maintenance cost are relatively high, currently, hydrogen energy storage is not the best cost - effective choice for the power grid. But with the improvement of the manufacturing process of the electrolyzer and fuel cell and the progress of technology, the development prospect of hydrogen energy storage is still very optimistic.

[0269] Same Figure 7 as Figure 11 in the electrochemical energy storage, the charge and discharge energy in a day are equal, and its maximum power for each charge and discharge is 7.5 MW, which is equal to its configured power and meets its operation constraints. Therefore, its charge and discharge energy in a year are also equal.

[0270] In Figure 12Overall, within one year, the charging and discharging energy of the hydrogen storage tank (i.e., multiplied by the corresponding weight coefficient) is equal, although the hydrogen charging and discharging amounts on each typical day are not equal; in addition, it can also be seen that the hydrogen energy stored in autumn and winter can supply the hydrogen energy consumed in spring and summer; at the same time, it can also be found that the sum of the areas corresponding to the curve and the coordinate axes is 0, because it satisfies the condition that the charging and discharging energy of the hydrogen energy storage system is equal within one year of the operation cycle. The detailed analysis is as follows:

[0271] (1) In spring, from 1:00 to 6:00, it is the cheapest time to purchase electricity from the superior power grid in a day, and the load demand is small. At this time, the load demand can be met by purchasing electricity from the superior power grid. Therefore, the excess wind power output can be stored through the hydrogen energy storage system. At 8:00, from 21:00 to 22:00, it is more expensive to purchase electricity from the superior power grid at this time. Although there is wind and light output, the load demand is large and it is not enough to meet the load demand. Therefore, to ensure the minimum cost, while purchasing a part of electricity from the superior power grid, the hydrogen energy stored in the hydrogen energy storage system is converted into electricity and sent to the power grid to ensure the safe and stable operation of the power grid;

[0272] (2) In summer, from 23:00 to 6:00, it is cheaper to purchase electricity from the superior power grid and the load demand is small. The load demand can be met by purchasing electricity from the superior power grid. Therefore, the excess wind power output can be stored through the hydrogen energy storage system. From 17:00 to 22:00, it is more expensive to purchase electricity from the superior power grid at this time. Although there is wind and light output, the load demand is large and it is not enough to meet the load demand. Therefore, to ensure the minimum cost, while purchasing a part of electricity from the superior power grid, the hydrogen energy stored in the hydrogen energy storage system is converted into electricity and sent to the power grid to ensure the safe and stable operation of the power grid;

[0273] (3) In autumn, from 23:00 to 6:00, purchasing electricity from the superior power grid can meet the load demand, and the cost is relatively small compared to sending electricity through a fuel cell. Therefore, the wind energy can be stored through the hydrogen energy storage system. From 21:00 to 22:00, it is more expensive to purchase electricity from the superior power grid. Although there is wind and light output but it is not enough to meet the load demand. Therefore, while purchasing electricity from the superior power grid, the hydrogen in the hydrogen energy storage system needs to be converted into electricity and sent to the power grid through the fuel cell;

[0274] (4) In winter, from 23:00 to 6:00, purchasing electricity from the superior power grid can meet the load demand. Therefore, the excess wind and light output can be used to electrolyze water in the electrolytic water hydrogen production device of the hydrogen energy storage system to produce hydrogen and store it in the hydrogen storage tank for future use. At 10:00 and 19:00, purchasing electricity from the superior power grid is not enough to meet the load demand. Therefore, it is necessary to burn hydrogen in the fuel cell of the hydrogen energy storage system to convert it into electricity and send it to the power grid to ensure the safe and stable operation of the power grid.

[0275] InFigure 13 It can be found that, overall, the hydrogen energy storage system releases hydrogen in spring and summer, which is converted into electrical energy through a fuel cell to supply the power grid to meet the load demand. While in autumn and winter, the electrolyzer electrolyzes water to convert electrical energy into hydrogen energy and stores it in the hydrogen storage tank. And the energy of the hydrogen storage tank at the beginning of the year is equal to that at the end of the year, because the sum of the hydrogen charging and discharging energy in the hydrogen storage tank within a year is 0.

[0276] In addition, in the present invention, the energy state levels in the hydrogen storage tanks on four typical days are extended to the energy state levels in the hydrogen storage tank within a year. Therefore, the amplified four curves represent the changing trends of the energy state levels in the hydrogen storage tanks on four typical days, that is, they respectively represent the changing trends of the energy state levels in the hydrogen storage tank in different seasons.

[0277] Meanwhile, it can also be found that the energy state level curve on a typical day in spring has a downward trend, and after being extended to a season, its energy state level curve also has a downward trend, and the same is true for summer; while in autumn, its energy state level curve has an upward trend, and the energy state level curve in winter also has an upward trend. This is because the change range of the energy state level on a typical day represents the change range of the energy state level of each day within a season, and the change range of each day is the same. Therefore, they have the same changing trend, and the sum of the change ranges of the energy state levels of each day is the change range of the energy state level in the hydrogen storage tank within a year.

[0278] It can also be found that the rising and falling rates in each season are inconsistent, because the hydrogen charging and discharging energy of each day in each season is different; in addition, from the turning points of the curves, it can be clearly seen that the four seasons of spring, summer, autumn and winter in a certain northern region are evenly distributed throughout the year, and the weight coefficients are all equal.

[0279] Figure 14 Among them, since the wind and light output accounts for 82% of the total source output, it can be sold to the superior power grid when the wind and light output is excessive, and purchased from the superior power grid when the wind and light output is insufficient. At any time, the stable and reliable operation of the power grid must be ensured, and the amount of electricity purchased and sold at each time period depends on the wind and light output, load power, electrochemical energy storage power, hydrogen energy storage power and the constraint conditions of each device.

[0280] To sum up, in view of the multi-time scale characteristics of the flexibility demand in the power system with a high proportion of renewable energy, the present invention constructs a hybrid time scale hydrogen-electricity combined energy storage system, and establishes a planning model of the hybrid time scale hydrogen-electricity combined energy storage system considering technical economy. Compared with the existing single time scale energy storage system planning method, the factors considered in the planning are closer to the actual operation, and the planning results can significantly improve the technical and economic benefits of the planning.

Claims

1. A planning method for a hydrogen-electric hybrid energy storage system with a hybrid time scale, characterized in that: The method includes the following steps in sequence: (1) Decompose the daily net load net_P for the planned reference years arranged in chronological order i,t into daily fluctuating load and daily reference load P i , where i represents the date serial number within the planned reference year, i = 1, 2, …, Ni, and Ni represents the number of days in the planned reference year; t represents the time period serial number for each day within the planned reference year, t = 1, 2, ..., Nt, and Nt represents the number of time periods for each day within the planned reference year. (2) For the daily fluctuating load obtained in step (1) Use the K-means clustering algorithm improved based on sensitive cluster centers to obtain Ns typical daily fluctuating loads within the planning horizon (3) For the daily reference load P obtained in step (1) i , the weighted average method is used to obtain the reference loads P of Ns typical days within the planned horizon k ; (4) Combine the Ns typical daily fluctuating loads obtained in steps (2) and (3) and the Ns typical daily reference loads P k , and obtain the Ns typical daily loads P k,t within the planned horizon. Meanwhile, calculate the proportion π(k) of the typical day k within the planned horizon; (5) Based on the data obtained in step (4), establish a hybrid time-scale hydrogen-electricity combined energy storage system planning model for hydrogen energy storage on an annual time scale and electrochemical energy storage on a daily time scale; the objective function of this model is to minimize the comprehensive annualized cost C of the hybrid time-scale hydrogen-electricity combined energy storage system ann , including the equipment investment cost C inv , operation and maintenance cost C ope , and curtailment penalty cost C pun , and the expression is as follows: C ann = C inv + C ope + C pun (15) Equipment investment cost C inv As shown in Equation (16): Where: x q is the capacity of equipment of type q, c inv,q is the unit investment cost of equipment of type q, and Nq is the number of types of equipment of type q; r c is the discount rate; N pl is the planned service life of the equipment; Operation and maintenance cost C ope As shown in Equation (17): Where: Ns is the number of typical days; is the unit power cost of purchasing electricity from the power grid at time t; is the unit power cost of selling electricity to the power grid at time t; is the 0-1 state variable of purchasing electricity from the power grid at time t. When the value is 0, it means not working; when the value is 1, it means working; is the 0-1 state variable of selling electricity to the power grid at time t. When the value is 0, it means not working; when the value is 1, it means working; Penalty cost C for curtailed wind and solar power pun As shown in Equation (18): Where: is the curtailment wind / solar power at time t within the typical day k; is the unit curtailment wind / solar cost at time t within the typical day k; (6) Solve using a mixed-integer non-linear programming solver to obtain the optimal planned capacity of each device in the hybrid time-scale hydrogen-electricity combined energy storage system.

2. The method for planning a hydrogen-electric hybrid energy storage system with a mixed time scale according to claim 1, wherein: The specific content of step (1) is as follows: Obtain the renewable energy power generation newP at the t-th time period of the i-th day within the planned horizontal year i,t , the electrical load P' t,t , and the hydrogen load H i,t . Convert the hydrogen load into an electrical load according to formula (1) to obtain the comprehensive electricity consumption load P at the t-th time period of the i-th day within the planned horizontal year i,t : P i,t = k h2p H i,t + P' i,t (1) where k h2p is the hydrogen-electric conversion coefficient; Obtain the net load net_P for the planned horizontal year i,t , as shown in Equation (2): net_P i,t = newP i,t - P i,t (2) Decompose the net load net_P in the planned horizontal year i,t into the intra-day fluctuating load and the intra-day reference load P i : Let the intra-day fluctuating load in the first time period of each day within the planning horizon be the intra-day reference load P i , that is 3. The method for planning a hydrogen-electric hybrid energy storage system with a mixed time scale according to claim 1, wherein: The specific steps of step (2) include the following: (2a) Take the intra-day fluctuating load in the planned horizontal year as the sample data set, denoted as P = {P1, P2,..., P Ni}; where Calculate the Euclidean distance dist(P i and P j ) between any two data objects P i and P j : Calculate the average distance MeanDist between data objects in the sample data set P using Equation (5): In the formula, is the number of combinations of taking 2 data objects from Ni data objects; (2b) With P i as the center, the area with the average distance MeanDist as the radius is called the neighborhood of P i . At the same time, calculate the density parameter density(P i ) as shown in Equation (6): i ​ In formula (6), u(dist(P i ,P j )) represents a function, and there is: Obtain the density parameters of all Ni data objects; add the top M data objects with the largest density parameters to the candidate point set D, D = {D i | density(P i )}, i = 1, 2,..., M, where M is 0.8Ni; (2c) Define the loop variable r and initialize r = 1; define the number of clusters as k c , and initialize k c = 1; define the initial maximum similarity between clusters as AMS r-1 , and initialize AMS r-1 = 0; define the set of cluster centers as A r-1 , and initialize it as an empty set; use the alternative point set D described in step (2b) as the (r - 1)-th alternative point set D r-1 ; (2d) Select two data objects \(P\) and \(P\) with the largest corresponding density parameters from the \((r - 1)\)-th alternative point set \(D\). r-1 Take the density parameters of each as weights, and obtain the \(r\)-th clustering center \(C\) by weighted summation according to formula (8). max，1 max，2 r :​​ C r = density(P max，1 )·P max，1 + density(P max，2 )·P max，2 (8) Add C r to the clustering center set A r-1 to obtain the r-th clustering center set A r ; meanwhile, delete the density parameters corresponding to the two data objects P max，1 and P max，2 from the alternative point set D r-1 to obtain the r-th alternative point set D r ; (2e) Select a sample \(D\) from the alternative point set \(D\). r such that the corresponding \(P\) i is the farthest from the cluster center \(C\) i in \(A\) r , and use this \(P\) r as the \((r + 1)\)-th cluster center \(C\) i , add it to the set of cluster centers to obtain the \((r + 1)\)-th set of cluster centers \(A\) r+1 ; at the same time, delete the density parameter corresponding to this data object \(P\) r+1 from the alternative point set \(D\) i to obtain the \((r + 1)\)-th alternative point set \(D\) r ; r+1 ​ (2f) Assign k + 1 to k; (2g) Calculate the data object P corresponding to the remaining data objects in the (r + 1)-th candidate point set D r+1 and the Euclidean distances between each cluster center in the (r + 1)-th cluster center set A j and each data object, and assign each data object to the class where the cluster center with the closest Euclidean distance is located, so as to obtain k clusters; use Equation (9) to obtain the cluster center C of the i-th class in the k clusters r+1 and the average distance s i between the cluster center C of the i-th class and the data object P i j in the i-th class i : In Equation (9), Ki is the total number of data objects in the i-th class; (2h) Obtain the cluster center C of the i-th class using Equation (10) i and the cluster center C of the j-th class j The distance d between them i,j : d i,j = dist(C i , C j ) (10) In the formula, i = 1, 2, …, k; j = 1, 2, …, k; The r-th average maximum similarity between classes AMS is obtained by using Equation (11). r : Meanwhile, determine AMS r <AMS r-1 to check if it holds. If it does, go to step (2i); otherwise, go to step (2l); (2i) Update the cluster center C′ using Equation (12) i :[[-END]] (2j) Calculate the sum of the Euclidean distances between any data object corresponding to a density parameter in the (r + 1)-th candidate point set D r+1 and all updated cluster centers respectively, to obtain a set of the sums of the Euclidean distances between the data objects corresponding to all density parameters and all updated cluster centers. Select the data object corresponding to the maximum value from the set of the sums of the Euclidean distances as the (r + 2)-th cluster center C r+2 and put it into the (r + 1)-th cluster center set A r+1 to obtain the (r + 2)-th cluster center set A r+2 ; (2k) Remove the density parameter corresponding to the (r + 2)-th clustering center C r+2 from the (r + 1)-th set of candidate points D r+1 to obtain the (r + 2)-th set of candidate points D r+2 ; (2l) After assigning r + 1 to r, go to step (2f); (2m) Let Ns = k, then the Ns cluster centers corresponding to AMS r-1 are used as the initial cluster centers of the K-means clustering algorithm, and Ns typical daily load fluctuations are obtained 4. The method for planning a hydrogen-electric hybrid energy storage system with a mixed time scale according to claim 1, wherein: The specific content of step (3) is as follows: According to the clustering result of step (2), the daily reference load P in step (1) i is allocated to each cluster to obtain the daily reference load P of the planned horizontal year i,k , where i = 1, 2,..., Ni; k = 1, 2,..., Ns. The weighted average method is used to calculate the reference load P of Ns typical days k , where i = 1, 2,..., Ns: Where: Nk is the number of data objects included in the k-th typical day, k = 1, 2, ..., Ns.

5. The method for planning a hybrid time-scale hydrogen-electricity combined energy storage system according to claim 1, wherein: The specific content of step (4) is as follows: combining the Ns typical daily fluctuating loads obtained in steps (2) and (3) and the Ns typical daily reference loads P k , and calculating the Ns typical daily loads P k,t within the planned horizontal year: Where: Nk is the number of data objects included in the k-th typical day, k = 1, 2, ..., Ns; t = 1, 2, ..., Nt; At the same time, calculate the proportion π(k) of the k-th typical day within the planning horizon according to formula (14): Where: Ni represents the number of days in the planning horizon.

6. The method for planning a hydrogen-electric hybrid energy storage system with a mixed time scale according to claim 1, wherein: The constraint conditions of the planning model of the hybrid time-scale hydrogen-electricity combined energy storage system are as follows: The constraint conditions of the electrochemical energy storage are: In the formula, represents the charging power of the electrochemical energy storage in the t-th time period within the typical day k; represents the discharging power of the electrochemical energy storage in the t-th time period within the typical day k; represents the rated power of the electrochemical energy storage; represents the 0-1 state variable of the charging of the electrochemical energy storage in the t-th time period within the typical day k. When the value is 0, it means not working, and when the value is 1, it means working; represents the 0-1 state variable of the discharging of the electrochemical energy storage in the t-th time period within the typical day k. When the value is 0, it means not working, and when the value is 1, it means working; The energy state level of the electrochemical energy storage at different time periods within the k-th typical day is expressed as: In the formula, represents the energy state level of the electrochemical energy storage in the (t + 1)-th period within the typical day k; represents the initial energy state level of the electrochemical energy storage within the typical day k; η EES represents the charge-discharge power coefficient of the electrochemical energy storage; Δt E is the operating time interval of the electrochemical energy storage within the typical day k; t′ represents the t′-th period, with values ranging from 1 to Nt, and Nt is taken as 24 h; The energy state level of the electrochemical energy storage at the initial state of each day should be equal to the energy state level at the initial state of the next day, that is, the energy state levels of charging and discharging of the electrochemical energy storage device within the k-th typical day are equal, expressed as: The interaction power between the superior power grid and the hybrid time-scale hydrogen-electricity combined energy storage system is less than or equal to the maximum electric power of the power tie line: In the formula, is the purchased power of the power system during the t-th period within the typical day k; is the sold power of the power system during the t-th period within the typical day k; is the maximum interactive power of the power connection line between the superior power grid and the hydrogen-electricity combined energy storage system; is the 0-1 state variable of power purchase of the power system during the t-th period within the typical day k. When the value is 0, it means not working; when the value is 1, it means working; is the 0-1 state variable of power sale of the power system during the t-th period within the typical day k. When the value is 0, it means not working; when the value is 1, it means working; The upper and lower limit constraints of the input power of the electrolytic water hydrogen production device at each time period are: Wherein, represents the input power of the electrolytic water hydrogen production device in the t-th period within the typical day k; represents the rated power of the electrolytic water hydrogen production device; represents the operating state of the electrolytic water hydrogen production device in the t-th period within the typical day k, 1 represents operation, and 0 represents stop; The start-up and shut-down time constraints of the electrolytic water hydrogen production device are expressed as: wherein represents the operating state of the electrolytic water hydrogen production device in the (t - 1)-th period within the typical day k; represents the operating state of the electrolytic water hydrogen production device in the k-th period within the typical day k; TO represents the minimum allowable startup time of the electrolytic water hydrogen production device; TS represents the minimum allowable shutdown time of the electrolytic water hydrogen production device; The upper and lower limit constraints of the output power of the fuel cell at each time period after linearization by the big M method are: Wherein, represents the output power of the fuel cell in the t-th period within the typical day k; represents the rated power of the fuel cell; represents the operating state of the fuel cell in the t-th period within the typical day k; The continuous start-up and shut-down duration constraints of the fuel cell are: In the formula, represents the operating state of the fuel cell in the (t-1)-th period within the typical day k; The energy state level of the hydrogen storage tank at different time periods within the k-th typical day is expressed as: In the formula, represents the energy state level of the hydrogen storage tank in the (t + 1)-th time period within the typical day k; represents the initial energy state level of the hydrogen storage tank within the typical day k; η ED represents the power coefficient of the water electrolysis hydrogen production device; η FC represents the power coefficient of the fuel cell; Δt H is the operating time interval of the hydrogen energy storage; The energy state levels of hydrogen charging and discharging of the hydrogen storage tank within one year are equal, expressed as: The power balance constraint is satisfied at the t-th time period within the k-th typical day: In the formula, respectively represent the wind power output and photovoltaic output in the t-th period of the k-th typical day.

7. A system for implementing the planning method of the hybrid time-scale hydrogen-electric combined energy storage system according to any one of claims 1 to 6, characterized in that: Including: An electrolytic water hydrogen production device, which uses water as a raw material. Under the action of direct current, water is decomposed into hydrogen and oxygen. The separated hydrogen is controlled by a regulating valve and output and sent to a hydrogen storage tank; An electrochemical energy storage device, which stores electrical energy by converting it into chemical energy and converts chemical energy into electrical energy; A hydrogen fuel cell, which converts electrical energy through the chemical reaction of hydrogen energy; A hydrogen storage tank, which is a pressure vessel made of metal materials such as stainless steel and aluminum alloy and is used to store hydrogen; When the power grid has surplus power, the electrical energy directly charges the electrochemical energy storage device to store the electrical energy by converting it into chemical energy; or, the electrical energy is converted into hydrogen by an electrolytic water device and stored in the hydrogen storage tank; When the power grid has insufficient power, the electrochemical energy storage device directly converts chemical energy into electrical energy and releases it to the power grid, or the hydrogen storage tank converts hydrogen into electrical energy through a hydrogen fuel cell and releases it to the power grid.

Images