Photovoltaic power station hybrid energy storage optimal configuration method and system

Abstract

The invention relates to the field of photovoltaic technology, and discloses a hybrid energy storage optimal configuration method and system for a photovoltaic power station, and the method comprises the steps: obtaining the historical operation data of the photovoltaic power station, and determining a multi-dimensional time sequence vector corresponding to the photovoltaic power station according to the historical operation data; performing clustering processing on the multi-dimensional time sequence vector to obtain typical scene data corresponding to the photovoltaic power station; constructing a target function by taking the maximum annual energy storage net income and the highest photovoltaic consumption rate of the photovoltaic power station as targets, and constructing constraint conditions to obtain a multi-target optimization model; an NSGA-II-GA two-stage collaborative optimization algorithm is adopted, the multi-target optimization model is solved according to the typical scene data and a preset hybrid energy storage charging and discharging strategy, and an optimal operation strategy of photovoltaic power station side hybrid energy storage is obtained; and determining an energy storage configuration scheme according to the optimal operation strategy. The operation strategy can be optimized, and the power generation utilization rate is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of photovoltaic power station, in particular to a photovoltaic power station hybrid energy storage optimization configuration method and system. BACKGROUND

[0002] As an important part of renewable energy, the precise hybrid energy storage optimization configuration technology has become a technical problem to be solved in the field. The existing hybrid energy storage optimization configuration has low accuracy. Therefore, how to improve the accuracy of hybrid energy storage optimization configuration has become a technical problem to be solved by technical personnel in the field. SUMMARY

[0003] Therefore, the embodiments of the present application provide a photovoltaic power station hybrid energy storage optimization configuration method and system.

[0004] In a first aspect, the embodiments of the present application provide a photovoltaic power station hybrid energy storage optimization configuration method, which comprises: obtaining historical operation data of a photovoltaic power station, and determining a multi-dimensional time sequence vector corresponding to the photovoltaic power station according to the historical operation data; performing clustering processing on the multi-dimensional time sequence vector to obtain typical scene data corresponding to the photovoltaic power station; constructing a target function with the maximum annual net benefit of energy storage and the highest photovoltaic consumption rate of the photovoltaic power station as the target, and constructing a constraint condition to obtain a multi-objective optimization model; solving the multi-objective optimization model according to the typical scene data and a preset hybrid energy storage charging and discharging strategy by using a NSGA-II-GA two-stage collaborative optimization algorithm to obtain an optimal operation strategy of hybrid energy storage on the photovoltaic power station side, and determining an energy storage configuration scheme according to the optimal operation strategy.

[0005] In one possible implementation, the clustering processing on the multi-dimensional time sequence vector to obtain the typical scene data corresponding to the photovoltaic power station comprises: dividing the multi-dimensional time sequence vector by time unit to obtain a multi-dimensional time sequence sub-vector set corresponding to each time unit, wherein each multi-dimensional time sequence sub-vector set contains historical operation data of the photovoltaic power station in the corresponding time unit; determining an optimal clustering number corresponding to each multi-dimensional time sequence sub-vector set for each multi-dimensional time sequence sub-vector set; performing clustering on the multi-dimensional time sequence sub-vector set according to the optimal clustering number to obtain the typical scene data corresponding to the photovoltaic power station.

[0006] In one possible implementation, the clustering on the multi-dimensional time sequence sub-vector set according to the optimal clustering number to obtain the typical scene data corresponding to the photovoltaic power station comprises: cluster the multi-dimensional time-series sub-vector sets according to the optimal cluster number to obtain a plurality of cluster clusters, and determine data corresponding to a cluster center of each cluster cluster as initial typical scene data; statistically determine a proportion of an amount of data in each cluster cluster to a total amount of data, and take the proportion as a weight of the corresponding cluster cluster; the total amount of data is a total amount of data of the corresponding multi-dimensional time-series sub-vector set; For each multi-dimensional time-series sub-vector set, the initial typical scene data is calculated according to the weight of each cluster cluster to obtain the typical scene data corresponding to each multi-dimensional time-series sub-vector set; Typical scene data corresponding to all multi-dimensional time-series sub-vector sets is spliced in time sequence to obtain the typical scene data corresponding to the photovoltaic power station.

[0007] In one possible implementation, the target function is constructed with the maximum energy storage annual net benefit and the highest photovoltaic consumption rate of the photovoltaic power station as the target, and the constraint condition is constructed to obtain a multi-objective optimization model, including: The first function is constructed with the maximum energy storage annual net benefit of the photovoltaic power station as the target; The second function is constructed with the highest photovoltaic consumption rate of the photovoltaic power station as the target; The first function and the second function are taken as the target function; The energy balance constraint, the SOC constraint, the power constraint and the charge-discharge mutual exclusion constraint are constructed; The energy balance constraint, the SOC constraint, the power constraint and the charge-discharge mutual exclusion constraint are taken as the constraint condition; The multi-objective optimization model is obtained according to the target function and the constraint condition.

[0008] In one possible implementation, the target function is: ; Wherein, represents the energy storage annual net benefit, represents the annual benefit, represents the annual cost; represents the photovoltaic consumption rate, represents the abandoned light power of the mth month, represents the total annual photovoltaic theoretical output, represents the number of days in the mth month.

[0009] In one possible implementation, the NSGA-II-GA two-stage collaborative optimization algorithm is used to solve the multi-objective optimization model according to the typical scene data and the preset hybrid energy storage charge-discharge strategy, and the optimal operation strategy of the hybrid energy storage on the photovoltaic power station side is obtained, including: The NSGA-II algorithm is adopted to globally solve the multi-objective optimization model according to the typical scene data and a preset hybrid energy storage charging and discharging strategy, so as to obtain the configuration parameters and initial decision variables of the hybrid energy storage on the photovoltaic power station side. Based on the configuration parameters, the genetic algorithm is adopted to optimize the initial decision variables, so as to obtain the optimal operation strategy.

[0010] In a second aspect, the embodiments of the present application provide a photovoltaic power station hybrid energy storage optimization configuration system, the system comprising: A first processing module is configured to acquire historical operation data of a photovoltaic power station, and determine a multi-dimensional time sequence vector corresponding to the photovoltaic power station according to the historical operation data. A second processing module is configured to perform clustering processing on the multi-dimensional time sequence vector, so as to obtain typical scene data corresponding to the photovoltaic power station. A third processing module is configured to construct a target function with the maximum annual net benefit of energy storage and the highest photovoltaic consumption rate of the photovoltaic power station as targets, and construct a constraint condition, so as to obtain a multi-objective optimization model. A fourth processing module is configured to adopt an NSGA-II-GA two-stage collaborative optimization algorithm to solve the multi-objective optimization model according to the typical scene data and a preset hybrid energy storage charging and discharging strategy, so as to obtain an optimal operation strategy of the hybrid energy storage on the photovoltaic power station side, and determine an energy storage configuration scheme according to the optimal operation strategy.

[0011] In a third aspect, the embodiments of the present application provide a computer device, comprising a memory and a processor, which are communicatively connected with each other, and the memory stores computer instructions, and the processor executes the photovoltaic power station hybrid energy storage optimization configuration method of the first aspect or any of the corresponding embodiments thereof by executing the computer instructions.

[0012] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, which stores computer instructions, and the computer instructions are used to make a computer execute the photovoltaic power station hybrid energy storage optimization configuration method of the first aspect or any of the corresponding embodiments thereof.

[0013] In a fifth aspect, the present application provides a computer program product comprising computer instructions for causing a computer to execute the photovoltaic power station hybrid energy storage optimization configuration method of the first aspect or any of the corresponding embodiments thereof.

[0014] The technical scheme provided by the present application has the following technical effects: The embodiment of the application provides comprehensive data support for subsequent optimization by acquiring historical operation data and constructing a multi-dimensional time sequence vector, and avoids decision deviation caused by single data; then, typical scene data is obtained by clustering the multi-dimensional time sequence vector, which can effectively reduce the redundancy of original data, reduce the calculation complexity while retaining the core characteristics, make the optimization process more efficient, avoid resource waste and time delay caused by full data operation, and guarantee the representativeness of the scene. The target function and constraint condition are constructed with the maximum annual net benefit of energy storage and the highest photovoltaic consumption rate as the target, the economic and technical targets are considered, and the one-sidedness of single target optimization is avoided. The NSGA-II-GA two-stage collaborative optimization algorithm is adopted, the typical scene data and the charging and discharging strategy are combined to solve the model, the advantages of NSGA-II in searching for multi-objective Pareto optimal solution and the global optimization ability of GA are fully exerted, the optimal operation strategy is accurately output, and finally the determined energy storage configuration scheme can realize accurate matching of energy storage capacity, charging and discharging power and photovoltaic output, load demand, significantly improve the annual net benefit of the system energy storage, maximize the photovoltaic consumption rate, reduce energy waste, and enhance the stability and economy of the photovoltaic power station operation. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the specific embodiments or the related art, the drawings needed to be used in the specific embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0016] Figure 1 is a flowchart of a photovoltaic power station hybrid energy storage optimization configuration method of the embodiment of the application; Figure 2 is a structural block diagram of a photovoltaic power station hybrid energy storage optimization configuration system of the embodiment of the application; Figure 3 is a hardware structure schematic diagram of a computer device of the embodiment of the application. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings of the embodiments of the present application. Obviously, the described embodiments are 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.

[0018] It can be understood that, before using the technical solutions disclosed in the embodiments of the present application, the type, use range, use scenario, etc. of the personal information involved in the present application should be informed to the user and the authorization of the user should be obtained according to relevant laws and regulations.

[0019] The embodiment of the present application provides a photovoltaic power station hybrid energy storage optimization configuration method embodiment. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer device such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that here.

[0020] Figure 1 The flowchart of the photovoltaic power station hybrid energy storage optimization configuration method of the embodiment of the present application is shown in FIG. 1, which includes the following steps: Figure 1 S101: Obtain historical operation data of a photovoltaic power station, and determine a multi-dimensional time sequence vector corresponding to the photovoltaic power station according to the historical operation data.

[0021] In the embodiment of the present application, the historical operation data is the operation data generated in the actual operation process of the photovoltaic power station. The historical operation data can include theoretical output data, actual on-grid power data, and real-time on-grid electricity price data of the photovoltaic power station running in the maximum power tracking (MPPT) mode, and can also include environment-related data such as light intensity, environmental temperature, humidity, etc.

[0022] As an example, the historical operation data of the photovoltaic power station can be collected at a preset period and a preset sampling rate. As an example, the preset period can be in units of years, such as 1 year, and the preset sampling rate can be in units of minutes, such as 15 minutes. Both the short-term fluctuation characteristics of photovoltaic output (such as sudden power change caused by cloud cover) can be accurately captured, and data redundancy caused by too high sampling rate can be avoided, balancing data accuracy and processing efficiency.

[0023] In the embodiment of the present application, the multi-dimensional time sequence vector is used to represent the multi-dimensional time sequence vector of the photovoltaic power station operation characteristics and the power market characteristics. The specific implementation manner of determining the multi-dimensional time sequence vector corresponding to the photovoltaic power station according to the historical operation data is as follows: The historical operation data is preprocessed. The different types of data in the preprocessed historical operation data are integrated in chronological order to obtain a multi-dimensional time sequence vector. The data at each time point corresponds to a time sequence vector. The multi-dimensional time sequence vector represents the dynamic change of the operation data generated in the actual operation process of the photovoltaic power station at different times.

[0024] As an example, the multi-dimensional time sequence vector can be H (​t )=[ P pv ( t ), P grid ( t ), J dj ( t )],in, Ppv(t) This provides theoretical power output data for photovoltaic power plants operating in maximum power point tracking mode. Ppv(t) Let t be the actual grid-connected power of the photovoltaic power station. Jdj(t) Let t be the on-grid electricity price for photovoltaic power.

[0025] This application employs conventional preprocessing methods in the field, such as deduplication, missing data processing, and outlier data processing. For deduplication, direct deletion is used. For missing data, Lagrange interpolation is used to impute individual data points, and B-spline segmented fitting is used to fit and fill in single or multiple missing rows. For outlier data, box plots are used for identification, Gaussian filtering is used to remove them, and then imputation is performed according to the missing data processing method. Through categorized processing, the problems of deduplication, missing data, and outlier data are effectively eliminated, ensuring that the integrity (missing rate reduced to extremely low), accuracy (outlier removal), and uniqueness (no redundancy) of the original data meet the requirements for multidimensional time-series vector construction and cluster analysis.

[0026] S102: Cluster the multi-dimensional time-series vectors to obtain typical scenario data corresponding to photovoltaic power plants.

[0027] In the embodiments of this application, typical scenario data refers to representative data extracted from massive multidimensional time-series vectors through clustering algorithms, which can characterize the monthly or annual operation patterns of photovoltaic power plants and the characteristics of the electricity market. It can replace the original annual data in the optimization calculation, reducing the complexity of the model while retaining the core operating characteristics.

[0028] In this embodiment of the application, when the time unit is a month, the typical scenario data includes: a multi-dimensional time-series vector of typical days corresponding to the best cluster of each month, covering the typical days of that month. Ppv(t) , Ppv(t) , Jdj(t)Change curve. The probability of occurrence of each typical scenario, i.e. the proportion of days in the cluster in the total days of the month. The annual typical scenario set formed by splicing the 12 months of the year, i.e. the typical scenario data corresponding to the photovoltaic power station, can fully reflect the photovoltaic output and price fluctuation law in different seasons throughout the year. In the embodiment of the present application, the multi-dimensional time series vector is clustered to obtain the typical scenario data corresponding to the photovoltaic power station, and the massive time series data is compressed into representative scenarios, which greatly reduces the calculation amount and complexity of the subsequent optimization model and improves the solving efficiency.

[0029] S103: Construct a target function with the maximum annual net benefit of energy storage and the highest photovoltaic consumption rate of the photovoltaic power station as the target, and construct a constraint condition to obtain a multi-objective optimization model.

[0030] S104: Adopt a NSGA-II-GA two-stage collaborative optimization algorithm to solve the multi-objective optimization model according to the typical scenario data and the preset mixed energy storage charging and discharging strategy, and obtain the optimal operation strategy of the mixed energy storage on the photovoltaic power station side. The energy storage configuration scheme is determined according to the optimal operation strategy.

[0031] In the embodiment of the present application, in the operation simulation, the charging strategy preferentially uses photovoltaic abandoned electricity to charge the lithium battery, and if there is still surplus electricity, the sodium battery is charged. The discharging strategy allows the two batteries to discharge at the same time when the electricity price is higher than the peak threshold value, so as to maximize the benefit.

[0032] The preset mixed energy storage charging and discharging strategy specifically includes: Photovoltaic surplus electricity charging: when the photovoltaic power generation power > grid demand, preferentially use photovoltaic abandoned electricity to charge the lithium battery, and after the lithium battery reaches the upper limit of the SOC, the remaining electricity is charged to the sodium battery, maximizing the use of abandoned light resources.

[0033] Low-price grid charging: when the real-time electricity price < grid charging threshold, purchase electricity from the grid to charge the energy storage, and the charging priority is consistent with that of the photovoltaic surplus electricity charging (lithium battery first and sodium battery second), which expands the charging source of the energy storage and increases the arbitrage space.

[0034] High-price energy storage discharging: when the real-time electricity price > peak price threshold, the lithium battery and the sodium battery are discharged at the same time, the power advantages of the two are superimposed, and the electricity selling benefit is maximized at the peak period of the electricity price.

[0035] In the embodiment of the present application, the operation strategy is used to describe various parameters of the characteristics and operating state of the mixed energy storage system. Specifically, a multi-objective optimization model can be established in combination with the typical scenario data and the preset mixed energy storage charging and discharging strategy to simulate the charging and discharging behavior of the mixed energy storage system under different operating scenarios. The operation strategy of the mixed energy storage system is adjusted through optimization by an algorithm to obtain the optimal operation strategy, so as to improve the benefit of the photovoltaic power station.

[0036] In the embodiments of the present application, the specific implementation manner of determining the energy storage configuration scheme according to the optimal operation strategy can be that the operation strategy is directly taken as the energy storage configuration scheme, and the operation strategy is taken as the configuration scheme of the hybrid energy storage system. In addition, the obtained configuration parameters can also be taken as the configuration scheme of the hybrid energy storage system.

[0037] The core idea of the present application is to form an energy storage configuration scheme suitable for the whole cycle based on historical operation data of a photovoltaic power station, through data processing, scene extraction, model construction and two-stage optimization. First, 1 year or more of 15-minute resolution historical operation data (including light, output, electricity price, etc.) are collected, and after preprocessing (de-duplication, missing data filling, and exception removal), a multi-dimensional time sequence vector is constructed. Then, the improved DTW-Kmeans algorithm is used to divide the subsets by month and dynamically determine the optimal cluster number, select the representative vector of the cluster and generate the monthly typical scene data by weighting, and splice the annual typical scene data. Subsequently, a multi-objective optimization model is constructed with the dual objectives of maximizing the annual net benefit of energy storage and maximizing the photovoltaic power consumption rate, combined with the constraints of energy balance, SOC, and charge-discharge exclusion. Finally, the NSGA-Ⅱ-GA two-stage algorithm is used to optimize the energy storage capacity, power and other configuration parameters by the NSGA-Ⅱ algorithm first, and then optimize the monthly peak-valley electricity price threshold by the GA algorithm based on fixed configuration parameters, to finally obtain a whole-cycle energy storage configuration scheme that takes into account the economy and technology and can be implemented, solving the problem of poor economy of traditional methods.

[0038] In one possible implementation, the multi-dimensional time sequence vector is clustered in S102 to obtain typical scene data corresponding to the photovoltaic power station, including Sa1 to Sa3.

[0039] Sa1, the multi-dimensional time sequence vector is divided by time unit to obtain a set of multi-dimensional time sequence sub-vectors corresponding to each time unit.

[0040] In the embodiments of the present application, each set of multi-dimensional time sequence sub-vectors contains historical operation data of the photovoltaic power station within the corresponding time unit. As an example, the time unit can be a month.

[0041] For example, the multi-dimensional time sequence vector H(t) of the whole year (the collection period is 2 years, divided by natural months) is divided into 12 independent sets of multi-dimensional time sequence sub-vectors, each corresponding to a natural month, with the time unit being a month. Each set of multi-dimensional time sequence sub-vectors contains all 15-minute resolution historical operation data within the corresponding month, ensuring that each set of multi-dimensional time sequence sub-vectors can reflect the unique seasonal characteristics of the month, such as high photovoltaic output in summer, low output in winter, or high electricity price in peak electricity consumption months.

[0042] The embodiments of the application divide the multi-dimensional time sequence vector in time units, improve data pertinence, reduce the data volume of single batch clustering, and improve the accuracy of data clustering.

[0043] Sa2, for each multi-dimensional time sequence sub-vector set, determining the optimal clustering number corresponding to each multi-dimensional time sequence sub-vector set.

[0044] In the embodiments of the application, the Silhouette Score is used as the core evaluation index of clustering effect, which comprehensively considers the cohesion (the average distance of the sample and other samples in the same cluster) and separation (the average distance of the sample and the nearest neighbor cluster sample) of clustering, and the calculation formula is .

[0045] Among them: is the average distance of sample s(i) to all other samples in the same cluster, is the average distance of sample s(i) to all samples in the nearest neighbor cluster.

[0046] The clustering number k is defined as k∈[2, 15], and the Silhouette Score of various clustering schemes is calculated in this range. For each k, after clustering, the mean value of all samples s(i) is calculated, denoted as the overall Silhouette Score S(k): N represents the total number of average distances. The k corresponding to the maximum Silhouette Score is selected as the optimal clustering number The overall Silhouette Score S(k) is the mean value of all samples s(i), and the closer the value is to 1, the better the clustering effect is.

[0047] For each monthly multi-dimensional time sequence sub-vector set, DTW-Kmeans clustering calculation is performed one by one in the range of clustering number k∈[2, 15], and the corresponding overall Silhouette Score S(k) is calculated after each clustering.

[0048] By comparing S(k) corresponding to different k values, the k corresponding to the maximum S(k) is selected as the optimal clustering number k of the monthly sub-vector set, for example, the January sub-vector set may determine k=3 due to small data fluctuations, and the July sub-vector set determines k=5 due to large data fluctuations, ensuring that the clustering number of each month matches the data characteristics.

[0049] ​​In this embodiment, the optimal number of clusters is selected based on the silhouette coefficient, avoiding over-clustering (k too large, scene repetition) or under-clustering (k too small, scene omission) caused by subjectively setting the number of clusters, thus ensuring the scientific nature of the clustering results. The optimal number of clusters is determined separately for different monthly subsets to accommodate the fluctuations in data across months (e.g., stable photovoltaic output in winter, large fluctuations in summer), making the clustering results for each month more closely reflect actual operating characteristics.

[0050] Sa3 clusters the multidimensional time series sub-vectors based on the optimal clustering number to obtain typical scenario data corresponding to photovoltaic power plants.

[0051] In one possible implementation, Sa3 clusters the multidimensional time-series sub-vectors based on the optimal number of clusters to obtain typical scenario data corresponding to photovoltaic power plants, including Sa31 to Sa34.

[0052] Sa31 clusters the multidimensional time series sub-vectors according to the optimal number of clusters to obtain multiple clusters, and determines the data corresponding to the cluster center of each cluster as the initial typical scene data.

[0053] In the embodiments of this application, for each multidimensional time series sub-vector set, an improved DTW-Kmeans algorithm is used (designing a weighted distance function, with weights ω=[0.4, 0.3, 0.3], corresponding to respectively). Ppv(t) , Ppv(t) , Jdj(t) Clustering is performed based on the optimal number of clusters k for that month, dividing all time-series data in the sub-vector set into k non-overlapping clusters.

[0054] Within each cluster, the minimum average DTW distance criterion is used to select the cluster center. The DTW distance between all data points in the cluster and any reference point in the cluster is calculated. The reference point that minimizes the average DTW distance between all data points in the cluster and the reference point is found. The time series data corresponding to the reference point is determined as the initial typical scenario data for that cluster. For example, if a cluster corresponds to a sunny day with high output and high electricity price scenario, its initial typical scenario data is the representative time series curve of that scenario.

[0055] Methods for selecting cluster centers: .

[0056] , representing clusters Cluster centers Represents clusters The total number of data points in the middle, Represents the reference point. This represents the j-th data point. Indicates from The selected d should minimize the average DTW distance, and this selected d is .

[0057] The improved DTW-Kmeans algorithm highlights the core influence of photovoltaic output, on-grid power and electricity price through a weighted distance function, avoids ignoring a certain dimension of data, and ensures that the clustering cluster can accurately reflect the differences of different running scenarios. The clustering center selected by the minimum average DTW distance criterion can represent the overall characteristics of the data in the cluster to the greatest extent, so that the initial typical scene data has strong representativeness and provides high-quality basic data for subsequent weighted calculation.

[0058] Sa32, statistics of the proportion of the data amount in each clustering cluster to the total data amount, the proportion as the weight of the corresponding clustering cluster.

[0059] In the embodiment of the application, the total data amount is the total data amount corresponding to the multi-dimensional time sequence sub-vector set.

[0060] In the embodiment of the application, for each multi-dimensional time sequence sub-vector set, the number of time sequence data points contained in each clustering cluster (i.e. the number of 15-minute resolution data in the cluster, or the corresponding number of days, converted according to 96 data points per day) is counted.

[0061] The data amount of each clustering cluster is divided by the total data amount of the monthly multi-dimensional time sequence sub-vector set (the total number of 15-minute resolution data in the month), and the obtained proportion is the weight of the clustering cluster. For example, the total data amount of a monthly multi-dimensional time sequence sub-vector set corresponds to 30 days, and a clustering cluster contains 15 days of data, so the weight of the cluster is 15 / 30=0.5, representing the occurrence probability of the scene in the month is 50%.

[0062] The weight of the embodiment of the application is directly related to the actual occurrence frequency of the clustering cluster, so that the subsequent typical scene data can reflect the occurrence probability of different scenes, avoid treating low-frequency scenes and high-frequency scenes equally, and improve the actual reference value of the typical scene. The weight calculation method based on data amount proportion is simple and intuitive, and is related to the actual running days, so as to ensure that the weight result can truly reflect the distribution rule of the scene in the monthly period.

[0063] Sa33, for each multi-dimensional time sequence sub-vector set, the initial typical scene data is weighted and calculated according to the weight of each clustering cluster, to obtain the typical scene data corresponding to each multi-dimensional time sequence sub-vector set.

[0064] In the embodiments of the present application, for each set of multi-dimensional time sequence sub-vectors, the initial typical scene data of each cluster is weighted and summed with the weight of each cluster as the coefficient, and the formula is C_typical.m =∑(weight.m × C_initial.m), where C_typical.m is the final typical scene data (in the form of a multi-dimensional time sequence vector) of the month, weight.m is the weight of the mth cluster, and C_initial.m is the initial typical scene data of the mth cluster.

[0065] In the embodiments of the present application, by weighted fusion, the scene features of multiple clusters in the same month are integrated into a single typical scene data, which not only retains the core features of each scene, but also avoids the complexity of multi-scene parallel computing, greatly simplifying the input data of the subsequent optimization model. Weighted calculation makes the features of high-frequency scenes account for a higher proportion in the monthly typical scene, ensuring that the monthly typical scene can truly reflect the mainstream running state of the month and improving the adaptability of the optimization scheme to the actual scene of the month.

[0066] In the embodiments of the present application, the typical scene data obtained after weighting is: . Wherein, represents the typical scene data obtained after weighting, which is presented in the form of a matrix and integrates the typical daily features of different clusters, and can more comprehensively represent the running situation of the corresponding time period. is the number of days in the mth cluster, which reflects the frequency of the scene corresponding to the cluster in reality. is the total number of days in the month, which is used to calculate the proportion of the number of days in the mth cluster to the total number of days in the month, and the proportion can reflect the relative importance of the cluster scene. is the initial typical scene data of the mth cluster. By weighting and summing the typical daily time sequence of each cluster according to the proportion of the number of days in the corresponding cluster to the total number of days in the month, a comprehensive and more representative weighted typical daily multi-dimensional time sequence, i.e., typical scene data, is obtained, which can be used for subsequent analysis and optimization of related systems such as photovoltaic power stations, etc., and can highlight the influence of more frequently occurring scenes while retaining different scene features, thereby improving the accuracy and practicality of the analysis.

[0067] Sa34, the typical scene data corresponding to all sets of multi-dimensional time sequence sub-vectors is spliced in time sequence to obtain the typical scene data of the photovoltaic power station.

[0068] In the embodiment of the present application, the typical scene data corresponding to the 12 monthly sub-vector sets can be spliced in the order of natural months (January, February, …, December) to form a continuous multi-dimensional time sequence vector covering the whole year. In the splicing process, it is necessary to ensure that the time resolution (15 minutes) of each monthly typical scene data is consistent, and there is no overlap and no missing in adjacent month data, and finally form the annual typical scene data of 12x30x96=34560 data points (calculated according to 30 days per month, 96 15-minute data points per day) covering the whole year. The photovoltaic output, on-grid power, and electricity price characteristics of each season throughout the year are fully covered.

[0069] In the embodiment of the present application, the monthly typical scene is integrated into the annual typical scene in time sequence, so that the typical scene data can reflect the annual operation rule, and provide complete input basis for subsequent full-cycle optimization model, avoiding optimization limited to a single month. The amount of annual typical scene data after splicing is only 1 / 730 of the original 2-year data (about 525600 data points), which greatly compresses the data amount under the premise of retaining the core characteristics, and significantly improves the solution efficiency of the subsequent multi-objective optimization model.

[0070] In one possible implementation, the target function is constructed with the maximum energy storage annual net income of the photovoltaic power station and the highest photovoltaic consumption rate as the target in S103, and the constraint condition is constructed to obtain a multi-objective optimization model, including: The first function is constructed with the maximum energy storage annual net income of the photovoltaic power station as the target.

[0071] The second function is constructed with the highest photovoltaic consumption rate of the photovoltaic power station as the target.

[0072] The first function and the second function are taken as the target function.

[0073] The energy balance constraint, the SOC constraint, the power constraint, and the charge-discharge mutual exclusion constraint are constructed.

[0074] The energy balance constraint, the SOC constraint, the power constraint, and the charge-discharge mutual exclusion constraint are taken as the constraint condition.

[0075] The multi-objective optimization model is obtained according to the target function and the constraint condition.

[0076] In the embodiment of the present application, the target function is: .

[0077] Wherein, represents the energy storage annual net income, represents the annual income, represents the annual cost. represents the photovoltaic consumption rate, represents the abandoned light power of the mth month, denotes the total annual photovoltaic theoretical output, denotes the number of days in the mth time unit (month). Among them, the annualized income mainly comes from photovoltaic consumption income and power market arbitrage income, and the annualized cost includes the annualized conversion of the initial investment cost and the operation and maintenance cost of the energy storage system. A double-objective optimization model can be used to balance between economy and technology, and to avoid one-sidedness that may be caused by single-objective optimization.

[0078] In the embodiments of the present application, the cost of the hybrid energy storage system mainly consists of the initial investment cost and the operation and maintenance cost. The initial investment cost covers the one-time expenditure in the construction stage, and the operation and maintenance cost involves the continuous expenses during the system operation. Specifically, the cost structure includes the following key parts: cell cost (purchase cost of battery core components), system cost (equipment and integration cost except for the cell), EPC cost (engineering, design and construction cost), and operation and maintenance cost (maintenance and energy consumption expenditure during operation).

[0079] In the application of configuring the hybrid energy storage system (lithium battery and sodium battery) in the photovoltaic power station, the operation mode of the energy storage system has an important influence on its economic benefit and photovoltaic consumption capacity, two operation modes, i.e. the mode of not allowing to purchase power from the grid and the mode of allowing to purchase power from the grid, are designed, and the income difference of the two modes is compared through optimization and simulation. In the mode of not allowing to purchase power from the grid, the energy storage system only relies on photovoltaic abandoned power to charge, and the income comes from photovoltaic consumption and discharge in high price period. In the mode of allowing to purchase power from the grid, the function of grid charging in low price period is added, which further improves the arbitrage space.

[0080] Not allowing to purchase power from the grid: when the energy storage system only relies on photovoltaic abandoned power to charge, the income only comes from photovoltaic consumption and discharge in high price period, the mode of not allowing to purchase power from the grid is adopted. This case is usually applicable to the scene with low dependence on the grid and the photovoltaic abandoned power can meet the charging demand of the energy storage.

[0081] Allowing to purchase power from the grid: when the function of grid charging in low price period is added to further improve the arbitrage space, the mode of allowing to purchase power from the grid is adopted. Generally, it is used in the case that the photovoltaic abandoned power is insufficient, but the income obtained by discharging in high price period after purchasing power from the grid in low price period is higher.

[0082] The detailed calculation formula of the income is as follows: The income of the photovoltaic power station configured with the hybrid energy storage system in the mode of purchasing power from the grid: .

[0083] The income of the photovoltaic power station configured with the hybrid energy storage system in the mode of not purchasing power from the grid: .

[0084] wherein: Ppeakis the peak time electricity price (RMB / MWh), i.e. the price of selling electricity by the energy storage system at high electricity price, Qdischargeis the discharging electricity quantity of the lithium battery energy storage system in period t (MWh), ηis the charging and discharging efficiency of the lithium battery energy storage system, Pvalleyis the valley electricity price (RMB / MWh), i.e. the price of selling electricity by the energy storage system at valley electricity price, Qsis the purchasing electricity quantity of the lithium battery energy storage system from the power grid in period t (MWh). The same applies to the sodium battery energy storage system, which is not described here again.

[0085] Annualized income: .

[0086] Net present value of the annualized income: .

[0087] Capital recovery factor of the annualized income: .

[0088] Annualized cost: .

[0089] Net present value of the annualized cost: .

[0090] Capital recovery factor of the annualized cost: .

[0091] Self-funding: .

[0092] Loan amount: .

[0093] wherein: Cis the annualized cost, Tis the design life / lifetime, rdis the discount rate, ris the loan interest rate, fis the self-funding ratio, lis the loan ratio, Ccellis the total cost of lithium battery cells (RMB / MWh), Csystemis the lithium battery system cost (including battery, battery box, PCS, etc.) (RMB / MWh), CEPCis the lithium battery EPC cost (RMB / MWh), Coperationis the lithium battery operation and maintenance cost (RMB / MWh / year), Ccellis the total cost of sodium battery cells (RMB / MWh), Csystemis the sodium battery system cost (including battery, battery box, PCS, etc.) (RMB / MWh), CEPCis the sodium battery EPC cost (RMB / MWh), is the operation and maintenance cost for sodium electricity (Yuan / MWh / year). is the number of days per time unit (month), is the capital recovery factor, is the net present value, is the own funds, is the loan amount, is the maximum capacity of the sodium battery, is the maximum capacity of the lithium battery.

[0094] As an example, in the case of a hybrid energy storage system consisting of lithium batteries and sodium batteries, the constraint conditions specifically include the following: The energy balance constraint is: The lithium energy balance constraint is: .

[0095] The sodium energy balance constraint is: .

[0096] wherein, represents the lithium battery energy storage system power at time t, represents the lithium battery energy storage system power at time t-1, represents the lithium battery charging power at time t-1, represents the lithium battery discharging power at time t-1, represents the lithium battery charging efficiency, represents the lithium battery discharging efficiency, represents the sodium battery energy storage system power at time t, represents the sodium battery energy storage system power at time t-1, represents the sodium battery charging power at time t-1, represents the sodium battery discharging power at time t-1, represents the sodium battery charging efficiency, represents the sodium battery discharging efficiency.

[0097] The SOC constraint is: The lithium battery SOC constraint is: , .

[0098] The sodium battery SOC constraint is: , .

[0099] wherein, is the state of charge of the lithium battery energy storage system at time t, is the power of the lithium battery energy storage system at time t, is the maximum power of the lithium battery energy storage system at time t, SoS Li battery energy storage system at time t, SoS Li battery energy storage system at time t,

[0100] The power constraint is: The power constraint of the lithium battery is:

[0101] The power constraint of the sodium battery is:

[0102] wherein, Pmax is the maximum charge-discharge power of the lithium battery, Pmax is the maximum charge-discharge power of the lithium battery, Pcharge is the charging power of the lithium battery at time t, Pmax is the maximum charge-discharge power of the lithium battery,

[0103] The charge-discharge mutual exclusion constraint is: The charge-discharge mutual exclusion constraint of the lithium battery is:

[0104] The charge-discharge mutual exclusion constraint of the sodium battery is:

[0105] wherein, Pcharge is the charging power of the lithium battery at time t, Pdischarge is the discharging power of the lithium battery at time t, and the same applies to the sodium battery.

[0106] In one possible implementation, the NSGA-II-GA two-stage collaborative optimization algorithm in S104 is used to solve the multi-objective optimization model according to the typical scenario data and the preset hybrid energy storage charge-discharge strategy, to obtain the optimal operation strategy of the hybrid energy storage on the side of the photovoltaic power station, including Sb1 to Sb2.

[0107] Sb1, the NSGA-II algorithm is used to globally solve the multi-objective optimization model according to the typical scenario data and the preset hybrid energy storage charge-discharge strategy, to obtain the configuration parameters and initial decision variables of the hybrid energy storage on the side of the photovoltaic power station.

[0108] ​​​​​​In the embodiments of the present application, the decision variables can include a peak price threshold and a grid charging threshold. The peak price threshold determines the triggering condition for discharging of the hybrid energy storage system in a high electricity price period. The grid charging threshold determines the triggering condition for charging in a low electricity price period in the mode of allowing electricity purchase from the grid. In the case of typical scenario data being monthly typical scenario data, the decision variables can be monthly decision variables. The configuration parameters can include hardware parameters such as the capacity and power of the lithium battery and the sodium battery.

[0109] In the embodiments of the present application, the core parameters of the NSGA-II algorithm are first initialized, and reasonable population size, maximum iteration number, and crossover and mutation probability are set in combination with the dimension of the typical scenario data, to ensure that the algorithm takes into account both search efficiency and population diversity. Then, the configuration parameters and the decision variables of the photovoltaic power station are real-coded to form chromosomes and generate an initial population. Subsequently, taking the typical scenario data as input, the annual average net income and the photovoltaic consumption rate of each population individual are calculated according to the preset hybrid energy storage charging and discharging strategy (photovoltaic excess electricity preferentially charges the lithium battery, the lithium battery charges the sodium battery when the SOC upper limit is reached, both batteries are discharged when the electricity price is higher than the peak price threshold, and electricity is purchased from the grid to store energy when the electricity price is lower than the grid charging threshold). The individual is then subjected to constraint checking, and feasible solutions that satisfy the energy balance, SOC, power, and charging and discharging exclusion constraints are selected. Then, the Pareto level of the feasible solutions is divided by fast non-dominated sorting, the individual crowding degree is calculated to ensure population diversity, and the population is updated through selection, crossover, and mutation operations. The target calculation, constraint checking, and population updating process are repeated until the maximum iteration number is reached, and finally the configuration parameters and the initial decision variables of the hybrid energy storage of the photovoltaic power station are extracted from the optimal non-dominated solutions. The optimal non-dominated solution corresponds to the maximum annual net income of the energy storage of the photovoltaic power station and the highest photovoltaic consumption rate. The present application can take the maximum annual net income of the energy storage and the highest photovoltaic consumption rate as the optimization target, simulate the charging and discharging process under different configuration parameters and decision variables based on the typical scenario data and in combination with the preset hybrid energy storage charging and discharging strategy, obtain the corresponding annual net income and photovoltaic consumption rate, and finally extract the parameters meeting the double optimization targets to obtain the configuration parameters and the initial decision variables under the constraint conditions of energy balance, SOC, power, and charging and discharging exclusion. The different configuration parameters and decision variables used in the simulation can be preset values.

[0110] Sb2, based on the configuration parameters, optimizes the initial decision variables by using a genetic algorithm to obtain an optimal operation strategy.

[0111] In the embodiments of the present application, first, the hardware configuration parameters such as the lithium battery capacity, the sodium battery capacity, the lithium battery power and the sodium battery power determined by the first-stage NSGA-II algorithm are fixed to ensure that the subsequent optimization is only around the operation strategy, avoiding the influence of hardware parameter changes on the strategy adaptability. Then, the optimization object of the genetic algorithm is the initial decision variable, i.e. the peak price threshold and the grid charging threshold, which directly determines the triggering time of the energy storage system charging and discharging and is the key factor affecting the monthly income.

[0112] Subsequently, the typical scene data of each month is used as the input data of the genetic algorithm to ensure that the optimization process fits the actual operation characteristics of the month. In terms of the dimension of the monthly data and the characteristics of the optimization target, reasonable population size (such as 50-100 individuals), maximum iteration number (such as 30-50 generations) and crossover probability (such as 0.7-0.9), mutation probability (such as 0.01-0.05) are set to balance the search efficiency and optimization accuracy of the algorithm.

[0113] In the target function calculation link, according to the preset hybrid energy storage charging and discharging strategy (photovoltaic excess electricity priority charging lithium battery, lithium battery reaching the upper limit of SOC charging sodium battery, when the electricity price is higher than the peak price threshold, the double batteries are discharged at the same time, and when the electricity price is lower than the grid charging threshold, the energy storage is purchased from the grid), the charging and discharging process of each population individual (corresponding to a set of peak price threshold and grid charging threshold) in the typical scene of the month is simulated, and its monthly income is calculated; at the same time, combined with the photovoltaic power station operation data, the monthly photovoltaic consumption rate corresponding to the individual is calculated, forming a single-objective optimization direction with the core of maximizing the monthly income and taking into account the photovoltaic consumption rate (since the hardware parameters are fixed, the focus is on the improvement of the income of the operation strategy).

[0114] Then, the population individuals are checked for constraints to verify whether they meet the energy balance constraint (the change in the amount of electricity of the lithium battery and the sodium battery needs to meet the relationship between the charging amount, the discharging amount and the efficiency), the SOC constraint (the SOC of the lithium battery and the sodium battery needs to be maintained between the set minimum and maximum values), the power constraint (the charging and discharging power does not exceed the maximum allowed power of the device) and the charging and discharging exclusion constraint (the battery cannot be charged and discharged at the same time), and the individuals that do not meet the constraints are punished by a penalty function to reduce their fitness, ensuring the feasibility of the optimization result.

[0115] The population is updated by selection (e.g., roulette wheel selection or tournament selection), crossover (e.g., real number crossover), and mutation (e.g., uniform mutation) operations, and the target function calculation and constraint checking steps are iteratively performed until a maximum number of iterations is reached. Finally, the individual with the highest monthly revenue and the highest photovoltaic consumption rate is selected from the optimal population after iteration, and the corresponding peak price threshold and grid charging threshold are extracted as the optimal decision variable for the month. The above process is repeated to optimize the decision variable for 12 months, and the optimal peak price threshold and grid charging threshold for 12 months are integrated to form an optimal operation strategy for the photovoltaic power station side hybrid energy storage system that adapts to different seasonal operating characteristics throughout the year.

[0116] In the embodiments of the present application, the energy storage configuration scheme is applied to the whole cycle operation scenario of the photovoltaic power station. Based on the typical scenarios extracted from the historical operation data of the photovoltaic power station (such as illumination, output, electricity price, etc.), and in combination with the preset hybrid energy storage charging and discharging strategy (photovoltaic excess electricity priority charging lithium battery, grid low price electricity charging, lithium / sodium battery discharging at high electricity price peak), the configuration parameters of the photovoltaic power station side hybrid energy storage system and the optimized charging and discharging thresholds determined by the NSGA-II-GA two-stage algorithm can be directly used for the actual deployment and scheduling of the photovoltaic power station energy storage system. In daily operation, the charging and discharging operations are dynamically performed according to the real-time output, on-grid power and electricity price, thereby reducing light abandonment and improving arbitrage income.

[0117] In the embodiments of the present application, the NSGA-II-GA two-stage collaborative optimization algorithm includes: (1) Two-stage collaborative mechanism: In the first stage, the NSGA-II multi-objective optimization algorithm is used to perform global energy storage system configuration optimization to determine the hardware parameters such as energy storage capacity and power, i.e., configuration parameters. In the second stage, based on the fixed hardware configuration, the genetic algorithm is used to finely optimize the operation strategy for each month. This hierarchical optimization strategy ensures global optimality and achieves seasonal adaptability of the operation strategy.

[0118] (2) The NSGA-II algorithm is specially adapted and designed for the characteristics of the photovoltaic power station hybrid energy storage system configuration. In the coding scheme, four-dimensional hybrid coding is used: lithium battery capacity, sodium battery capacity, peak price threshold, and grid charging threshold, among which the energy storage capacity and the charging and discharging threshold are coded as continuous real numbers. In the design of the objective function, the algorithm optimizes both economic and technical objectives. The economic objective aims to maximize net investment returns, and the technical objective aims to maximize the photovoltaic consumption rate. In the design of the constraint conditions, a hierarchical strategy is adopted. For hard constraints such as device physical limitations and charging and discharging thresholds, constraint checking and penalty functions are used to ensure feasibility. For soft constraints such as investment recovery period requirements and light abandonment rate control objectives, an adaptive penalty mechanism is used to guide the search process to converge to the ideal region.

[0119] (3) In the optimization design of the hybrid energy storage system of the photovoltaic power station, fine adjustment of the operation strategy is a key link to improve economic benefits and photovoltaic consumption capacity. Based on the energy storage capacity and power configuration determined by the first-stage NSGA-II algorithm, the second-stage genetic algorithm is used to optimize the operation strategy of the energy storage system in each month, and the peak price threshold and the grid charging threshold are adjusted to maximize the monthly income. The genetic algorithm is an optimization method based on the principles of natural selection and genetic evolution. By simulating the iterative evolution of the population, it can search for the optimal solution in a complex nonlinear solution space, and is particularly suitable for the monthly optimization demand of the energy storage system operation strategy.

[0120] In the optimization process, the GA algorithm optimizes the operation strategy of 12 months independently based on the typical day data of each month. The decision variables include the peak price threshold (which determines the trigger condition for discharging the energy storage system during the high electricity price period) and the grid charging threshold (which determines the trigger condition for charging during the low electricity price period in the mode of purchasing electricity from the grid). The monthly optimization process includes: 1) fixing the energy storage capacity and power configuration obtained in the first stage; 2) optimizing the strategy based on the typical day data of the month; 3) outputting the optimal charging and discharging thresholds and the expected daily income of the month.

[0121] In the embodiment of the present application, the current relevant data of the photovoltaic power station can be obtained, which can include the current photovoltaic power generation actual output data, actual grid-connected power data, real-time grid-connected electricity price data, and the current SOC data, remaining capacity data, and current charging and discharging power data of the hybrid energy storage system, etc., to provide real-time data support for subsequent scene judgment and operation decision.

[0122] As an example, the charging scene judgment can be performed based on the current relevant data.

[0123] The charging scene is divided into two categories: photovoltaic surplus electricity charging and grid low price charging. The judgment needs to prioritize the use of photovoltaic surplus electricity, and then consider grid electricity purchase: Photovoltaic surplus electricity charging scene: when the real-time photovoltaic power generation actual output > actual grid-connected power (i.e. there is photovoltaic waste electricity), and the hybrid energy storage system has not reached the SOC upper limit (the remaining capacity is sufficient), the scene is triggered. At this time, the photovoltaic waste electricity is used to charge the lithium battery first, and when the lithium battery reaches the SOC upper limit, if there is still surplus waste electricity, the sodium battery is charged, to maximize the use of waste light resources and avoid energy waste.

[0124] Grid low price charging scene: when the real-time grid-connected electricity price < the corresponding grid charging threshold of the month, and the hybrid energy storage system has not reached the SOC upper limit, the scene is triggered. At this time, the grid electricity is purchased to charge the energy storage system, and the charging priority is consistent with the photovoltaic surplus electricity charging (lithium first and sodium second). Through the arbitrage of storing electricity during the low electricity price period and discharging during the high electricity price period, the energy storage income space is expanded.

[0125] As an example, the discharge scenario judgment can be made based on the current relevant data.

[0126] The discharge scenario is triggered only under the condition of high price discharge of energy storage: when the real-time on-grid electricity price > the peak threshold corresponding to the month, and the SOC of the hybrid energy storage system is higher than the minimum threshold (there is dischargeable capacity), the scenario is triggered. At this time, lithium batteries and sodium batteries are allowed to discharge simultaneously, and the power advantages of the two are superimposed to maximize the electricity selling revenue during the peak period of electricity price, while the power demand of the power grid can be supplemented.

[0127] As an example, the abandoned electricity scenario judgment can be made based on the current relevant data.

[0128] The abandoned electricity scenario occurs only when there is no charging demand or no charging: when the actual output of real-time photovoltaic power generation > the actual on-grid power (there is photovoltaic abandoned electricity), but the hybrid energy storage system has reached the upper limit of the SOC (there is no remaining capacity to be charged), and the real-time on-grid electricity price does not reach the charging threshold of the power grid (there is no need to purchase electricity from the power grid, and there is no discharge revenue space), the excess photovoltaic power cannot be stored or consumed by the energy storage system, and finally abandoned electricity is formed.

[0129] As an example, the judgment of whether to purchase electricity from the power grid can be made based on the current relevant data.

[0130] Whether to purchase electricity from the power grid depends only on the triggering condition of the low-price charging scenario of the power grid: only when the real-time on-grid electricity price < the charging threshold of the power grid corresponding to the month, and the hybrid energy storage system has remaining capacity (chargeable), electricity is purchased from the power grid; if the real-time electricity price ≥ the charging threshold of the power grid, or the energy storage system is full (no remaining capacity), the operation of purchasing electricity from the power grid is not triggered, so as to avoid the loss of revenue due to the high cost of purchasing electricity.

[0131] The present application can calculate the economic evaluation index after obtaining the energy storage configuration scheme, and verify the feasibility of the energy storage configuration scheme through the economic evaluation index.

[0132] As an example, the economic evaluation index can include annualized net income and return on investment.

[0133] In the case that the annualized net income is greater than 0 and the return on investment is greater than or equal to a preset reference value, the energy storage configuration scheme is determined to be feasible, and in the case that the annualized net income is less than or equal to 0 or the return on investment is less than the preset reference value, the energy storage configuration scheme is determined to be unfeasible and needs to be adjusted.

[0134] The annualized net income is defined as the annualized income minus the annualized cost: The net income is positive, indicating that the project is economically feasible, and the larger the net income, the better the economy of the project.

[0135] The return on investment (ROI) reflects the efficiency of the investment: .

[0136] It should be noted that the contents not described in detail in the specification of the present application are the known technology of those skilled in the art.

[0137] In the embodiment, a photovoltaic power station hybrid energy storage optimization configuration system is also provided, which is used to realize the above-mentioned embodiments and preferred embodiments, and has been described above. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the system described in the following embodiments is preferably implemented in software, hardware, or a combination of software and hardware is also possible and is contemplated.

[0138] Figure 2 is a structural block diagram of the photovoltaic power station hybrid energy storage optimization configuration system of the embodiment of the present application.

[0139] The embodiment provides a photovoltaic power station hybrid energy storage optimization configuration system, as shown in Figure 2 , comprising: A first processing module 11 is configured to acquire historical operation data of the hybrid energy storage system, and determine a multi-dimensional time sequence vector corresponding to the hybrid energy storage system according to the historical operation data.

[0140] A second processing module 12 is configured to perform clustering processing on the multi-dimensional time sequence vector to obtain typical scenario data corresponding to the hybrid energy storage system.

[0141] A third processing module 13 is configured to construct a target function with the maximum annual net income of the hybrid energy storage system and the highest photovoltaic consumption rate as targets, and construct a constraint condition to obtain a multi-objective optimization model.

[0142] A fourth processing module 14 is configured to solve the multi-objective optimization model according to the typical scenario data and a preset hybrid energy storage charging and discharging strategy by using an NSGA-II-GA two-stage collaborative optimization algorithm to obtain optimal operation parameters of the hybrid energy storage system, and determine an optimization configuration scheme according to the optimal operation parameters.

[0143] In one possible implementation, the second processing module 12 comprises: A first processing unit is configured to divide the multi-dimensional time sequence vector into a plurality of multi-dimensional time sequence sub-vector sets in time units, wherein each multi-dimensional time sequence sub-vector set contains historical operation data of the photovoltaic power station in a corresponding time unit.

[0144] A second processing unit is configured to determine an optimal clustering number corresponding to each multi-dimensional time sequence sub-vector set for each multi-dimensional time sequence sub-vector set.

[0145] The third processing unit is configured to cluster the set of multi-dimensional time-series sub-vectors according to the optimal number of clusters to obtain the typical scene data corresponding to the photovoltaic power station.

[0146] In a possible implementation, the third processing unit is specifically configured to cluster the set of multi-dimensional time-series sub-vectors according to the optimal number of clusters to obtain a plurality of clustering clusters, and determine data corresponding to a clustering center of each clustering cluster as initial typical scene data.

[0147] The proportion of the amount of data in each clustering cluster in the total amount of data is calculated, and the proportion is taken as the weight of the corresponding clustering cluster. The total amount of data is the total amount of data of the corresponding set of multi-dimensional time-series sub-vectors.

[0148] For each set of multi-dimensional time-series sub-vectors, the initial typical scene data is calculated by weighting according to the weight of each clustering cluster, to obtain the typical scene data corresponding to each set of multi-dimensional time-series sub-vectors.

[0149] The typical scene data corresponding to all sets of multi-dimensional time-series sub-vectors is spliced in time sequence to obtain the typical scene data corresponding to the photovoltaic power station.

[0150] In a possible implementation, the third processing module 13 is specifically configured to construct a first function with the maximum annual net benefit of the energy storage of the photovoltaic power station as a target.

[0151] A second function is constructed with the highest photovoltaic consumption rate of the photovoltaic power station as a target.

[0152] The first function and the second function are taken as objective functions.

[0153] An energy balance constraint, an SOC constraint, a power constraint, and a charge-discharge mutual exclusion constraint are constructed.

[0154] The energy balance constraint, the SOC constraint, the power constraint, and the charge-discharge mutual exclusion constraint are taken as constraint conditions.

[0155] A multi-objective optimization model is obtained according to the objective functions and the constraint conditions.

[0156] In a possible implementation, the objective functions are as follows: .

[0157] wherein, represents the annual net benefit of the energy storage, represents the annual benefit, represents the annual cost. represents the photovoltaic consumption rate, represents the abandoned light power in the mth month, represents the total annual photovoltaic theoretical output, represents the number of days in the mth month.

[0158] In one possible implementation, the fourth processing module 14 is specifically used to employ the NSGA-II algorithm to globally solve the multi-objective optimization model based on typical scenario data and a preset hybrid energy storage charging and discharging strategy, thereby obtaining the configuration parameters and initial decision variables of the hybrid energy storage on the photovoltaic power station side.

[0159] Based on the configuration parameters, a genetic algorithm is used to optimize the initial decision variables to obtain the optimal operating strategy.

[0160] In this embodiment, the photovoltaic power plant hybrid energy storage optimized configuration system is presented in the form of functional units. Here, a unit refers to an ASIC circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0161] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0162] This application also provides a computer device having the above-described features. Figure 2 The photovoltaic power station shows a hybrid energy storage optimization configuration system.

[0163] Please see Figure 3 , Figure 3 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of this application, such as... Figure 3 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 3 Take a processor 10 as an example.

[0164] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0165] The memory 20 stores instructions executable by the at least one processor 10 to cause the at least one processor 10 to perform the methods illustrated by the above embodiments.

[0166] The memory 20 can include a program storage area and a data storage area. The program storage area can store an operating system and applications required by at least one function. The data storage area can store data created by the use of the computer device according to the presentation of the applet landing page, and the like. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some alternative embodiments, the memory 20 can optionally include memory that is remotely located with respect to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0167] The memory 20 can include a volatile memory, such as a random access memory. The memory can also include a non-volatile memory, such as a flash memory, a hard disk, or a solid-state disk. The memory 20 can also include a combination of the above-mentioned kinds of memories.

[0168] The computer device also includes a communication interface 30 for the computer device to communicate with other devices or communication networks.

[0169] The embodiments of the present application also provide a computer readable storage medium. The above-mentioned methods according to the embodiments of the present application can be implemented in hardware, firmware, or recorded in a storage medium, or implemented as computer code originally stored in a remote storage medium or a non-transitory machine readable storage medium and downloaded to a local storage medium, so that the methods described herein can be processed by such software on a storage medium using a general-purpose computer, a special-purpose processor, or programmable or special-purpose hardware. The storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid-state disk, etc. Further, the storage medium can also include a combination of the above-mentioned kinds of memories. It can be understood that the computer, the processor, the microprocessor controller, or the programmable hardware includes a storage component that can store or receive software or computer code, when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the methods illustrated by the above embodiments are implemented.

[0170] Part of the present application can be applied as a computer program product, for example, computer program instructions, when executed by a computer, through the operation of the computer, can invoke or provide methods and / or technical solutions according to the present application. Those skilled in the art should understand that the form of computer program instructions in computer readable medium includes but is not limited to source files, executable files, installation package files and the like, and accordingly, the way of computer program instructions executed by computer includes but is not limited to: the computer directly executes the instructions, or the computer compiles the instructions and then executes the corresponding compiled program, or the computer reads and executes the instructions, or the computer reads and installs the instructions and then executes the corresponding installed program. Here, the computer readable medium can be any available computer readable storage medium or communication medium accessible to the computer.

[0171] Although the embodiments of the present application are described in conjunction with the drawings, various modifications and changes can be made by those skilled in the art without departing from the spirit and scope of the present application, and such modifications and changes fall within the scope defined by the appended claims.

Claims

1. A method for optimizing the configuration of hybrid energy storage in a photovoltaic power plant, characterized in that, The method includes: Acquire historical operating data of a photovoltaic power station, and determine the multi-dimensional time-series vector corresponding to the photovoltaic power station based on the historical operating data; Clustering is performed on the multidimensional time-series vector to obtain typical scenario data corresponding to the photovoltaic power station; An objective function is constructed with the goals of maximizing the annualized net energy storage revenue and the highest photovoltaic absorption rate of the photovoltaic power station, and constraints are established to obtain a multi-objective optimization model. The NSGA-Ⅱ-GA two-stage collaborative optimization algorithm is adopted to solve the multi-objective optimization model based on the typical scenario data and the preset hybrid energy storage charging and discharging strategy, so as to obtain the optimal operation strategy of hybrid energy storage on the photovoltaic power station side; and the energy storage configuration scheme is determined based on the optimal operation strategy.

2. The method according to claim 1, characterized in that, The clustering process of the multidimensional time-series vector to obtain typical scenario data corresponding to the photovoltaic power station includes: The multidimensional time series vector is divided into time units to obtain a multidimensional time series sub-vector set corresponding to each time unit; wherein, each multidimensional time series sub-vector set contains the historical operation data of the photovoltaic power station within the corresponding time unit; For each multidimensional time series sub-vector set, determine the optimal number of clusters corresponding to each multidimensional time series sub-vector set; Clustering the multidimensional time-series sub-vector set based on the optimal clustering number yields typical scenario data corresponding to the photovoltaic power station.

3. The method according to claim 2, characterized in that, The process of clustering the multidimensional time-series sub-vector set according to the optimal clustering number to obtain typical scenario data corresponding to the photovoltaic power station includes: The multidimensional time series sub-vector set is clustered according to the optimal number of clusters to obtain multiple clusters, and the data corresponding to the cluster center of each cluster is determined as the initial typical scene data. The proportion of data in each cluster to the total data is calculated, and this proportion is used as the weight of the corresponding cluster; the total data is the total data of the corresponding multidimensional time series sub-vector set. For each multi-dimensional time series sub-vector set, the initial typical scene data is weighted according to the weight of each cluster to obtain the typical scene data corresponding to each multi-dimensional time series sub-vector set; The typical scene data corresponding to all multi-dimensional time-series sub-vector sets are concatenated in chronological order to obtain the typical scene data corresponding to the photovoltaic power station.

4. The method according to claim 1, characterized in that, The objective function is constructed with the goals of maximizing the annualized net energy storage revenue and the highest photovoltaic grid integration rate of the photovoltaic power station, and constraints are established to obtain a multi-objective optimization model, including: A first function is constructed with the objective of maximizing the annualized net energy storage benefit of the photovoltaic power station. A second function is constructed with the objective of maximizing the photovoltaic absorption rate of the photovoltaic power station. The first function and the second function are used as the target function; Construct energy balance constraints, SOC constraints, power constraints, and charge / discharge mutual exclusion constraints; The energy balance constraint, the SOC constraint, the power constraint, and the charge / discharge mutual exclusion constraint are used as the constraint conditions. The multi-objective optimization model is obtained based on the objective function and the constraints.

5. The method according to claim 4, characterized in that, The objective function is: ； in, This indicates the annualized net return on energy storage. This represents the annualized return. Indicates annualized cost; Indicates the photovoltaic power absorption rate. This represents the amount of solar power wasted in month m. This represents the total theoretical output of photovoltaic power throughout the year. This represents the number of days in the m-th month.

6. The method according to claim 1, characterized in that, The NSGA-II-GA two-stage collaborative optimization algorithm is used to solve the multi-objective optimization model based on the typical scenario data and the preset hybrid energy storage charging and discharging strategy, thereby obtaining the optimal operation strategy for the hybrid energy storage on the photovoltaic power station side, including: The NSGA-II algorithm is used to solve the multi-objective optimization model globally based on the typical scenario data and the preset hybrid energy storage charging and discharging strategy, so as to obtain the configuration parameters and initial decision variables of hybrid energy storage on the photovoltaic power station side. Based on the configuration parameters, a genetic algorithm is used to optimize the initial decision variables to obtain the optimal operating strategy.

7. A hybrid energy storage optimized configuration system for photovoltaic power plants, characterized in that, The system includes: The first processing module is used to acquire historical operating data of the photovoltaic power station and determine the multi-dimensional time series vector corresponding to the photovoltaic power station based on the historical operating data. The second processing module is used to perform clustering processing on the multidimensional time-series vector to obtain typical scene data corresponding to the photovoltaic power station. The third processing module is used to construct an objective function with the goal of maximizing the annualized net energy storage income and the highest photovoltaic absorption rate of the photovoltaic power station, and to construct constraints to obtain a multi-objective optimization model. The fourth processing module is used to solve the multi-objective optimization model based on the typical scenario data and the preset hybrid energy storage charging and discharging strategy using the NSGA-Ⅱ-GA two-stage collaborative optimization algorithm, to obtain the optimal operation strategy of hybrid energy storage on the photovoltaic power station side; and to determine the energy storage configuration scheme based on the optimal operation strategy.

8. A computer device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method of any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the method of any one of claims 1 to 6.

10. A computer program product, characterized in that, Includes computer instructions for causing a computer to perform the method of any one of claims 1 to 6.

Images