Multi-type energy storage scheduling method and system based on net load trend-deviation decoupling

Abstract

The application provides a multi-type energy storage scheduling method and system based on net load trend-deviation decoupling. The method first decomposes a low-frequency trend of a day-ahead system net load prediction sequence to obtain a low-frequency trend component, then takes the low-frequency trend component as a day-ahead scheduling target, solves a day-ahead thermal power-pumped storage low-frequency benchmark scheduling model to obtain a low-frequency benchmark scheduling scheme including a pumped storage day-ahead working condition and power, then maps the pumped storage day-ahead working condition and power to an intra-day time scale, and finally takes a deviation of an intra-day real-time net load of the pumped storage from the day-ahead benchmark power as a compensation object, and solves an intra-day electrochemical energy storage-pumped storage collaborative optimization model to obtain an intra-day fine adjustment scheme of the electrochemical energy storage and the pumped storage. The application improves the economy, smoothness and engineering executability of the multi-type energy storage scheduling through a collaborative division of labor mode in which the low-frequency trend is borne by the thermal power and the pumped storage, and the high-frequency deviation is compensated by the electrochemical energy storage.

Description

Technical Field

[0001] This invention belongs to the field of power system dispatching technology, specifically relating to a multi-type energy storage dispatching method and system based on net load trend-deviation decoupling. Background Technology

[0002] Building a new power system with a high proportion of renewable energy has become an inevitable trend for the future. However, the output of renewable energy sources such as wind power and photovoltaics is characterized by randomness, volatility, and prediction errors, resulting in significant multi-timescale fluctuations in the system's net load, which places higher demands on the power system's regulation capabilities. Energy storage is an important means to improve system flexibility and peak-shaving and valley-filling capabilities. Pumped hydro storage has advantages such as large capacity, long lifespan, and suitability for low-frequency energy transfer, but its operating mode switching is constrained by unit operation and it is not suitable for frequently responding to short-term random fluctuations. Electrochemical energy storage has advantages such as fast response speed and high regulation accuracy, but its capacity is limited, its cycle cost is high, and it is difficult to withstand continuous deviations in the long term. Therefore, it is necessary to construct a multi-type energy storage collaborative scheduling method to achieve complementary advantages of different energy storage resources at different time scales.

[0003] Traditional multi-type energy storage coordinated dispatch methods often directly use the original net load or real-time deviation as the unified dispatch object, which can easily lead to unclear task allocation among thermal power units, pumped storage, and electrochemical energy storage. At the same time, existing day-ahead-intraday dispatch methods do not adequately consider the connection between day-ahead planning and intraday execution, which can easily cause pumped storage to frequently switch between power generation, pumping, and shutdown conditions during intraday rolling optimization, or cause electrochemical energy storage to bear an excessive burden of continuous deviation compensation tasks, thereby affecting the economics and engineering feasibility of the dispatch scheme. Summary of the Invention

[0004] The purpose of this invention is to address the aforementioned problems in the existing technology by providing a multi-type energy storage scheduling method and system based on net load trend-deviation decoupling.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows:

[0006] In a first aspect, the present invention provides a multi-type energy storage scheduling method based on net load trend-deviation decoupling, including:

[0007] S1. Obtain the day-ahead system net load prediction sequence under each typical source-load operation scenario;

[0008] S2. Perform low-frequency trend decomposition on the day-ahead system net load forecast sequence to obtain low-frequency trend components;

[0009] S3. Using the low-frequency trend component as the day-ahead scheduling target, solve the day-ahead thermal power-pumped storage low-frequency benchmark scheduling model to obtain a low-frequency benchmark scheduling scheme that includes the day-ahead operating conditions and day-ahead power of pumped storage.

[0010] S4. Map the day-ahead operating conditions of pumped storage to the intraday time scale to obtain the intraday locked operating condition sequence; map the day-ahead power of pumped storage to the intraday time scale to obtain the day-ahead reference power sequence.

[0011] S5. Taking the deviation of the real-time net load of pumped storage within a day from the daytime baseline power as the compensation object, the intraday electrochemical energy storage-pumped storage collaborative optimization model is solved in a rolling manner to obtain the intraday scheduling scheme of electrochemical energy storage and the intraday fine-tuning scheme of pumped storage under various typical source-load operation scenarios. In this intraday optimization model, the actual intraday operation of pumped storage is based on the daytime locked operating condition sequence, coordinating the rapid compensation of electrochemical energy storage and the limited fine-tuning of pumped storage.

[0012] The day-ahead thermal power-pumped storage low-frequency benchmark scheduling model aims to minimize the operating and start-up costs of thermal power units, the day-ahead operating costs of pumped storage, the equivalent costs of day-ahead planning fluctuations, and the costs of power balance and line safety relaxation penalties. Its constraints include system power balance constraints, thermal power unit operating constraints, pumped storage day-ahead operating constraints, and network security constraints.

[0013] The day-ahead planned fluctuation equivalent cost, power balance, and line safety slack penalty cost are calculated based on the following formula:

[0014] ;

[0015] ;

[0016] In the above formula, The equivalent cost of the current planned fluctuations, , These are the ramp smoothing penalty coefficient for thermal power units and the day-ahead baseline power fluctuation penalty coefficient for pumped storage power units, respectively. For the day-ahead time period t, the ramp-up auxiliary variable for thermal power unit i is... For the collection of thermal power units, For the day-ahead time period t, the auxiliary variable is the day-ahead baseline power fluctuation of pumped storage. To offset the cost of power balancing and line safety slack, , These are the power balance relaxation penalty factor and the line safety relaxation penalty factor, respectively. , These represent the positive and negative slack variables of the power balance during the daytime period t, respectively. The day-ahead scheduling interval, For the set of routes, For the daytime period t line The trend of safety relaxation variables, This represents the total number of scheduling periods in the previous day;

[0017] The system power balance constraints include:

[0018] ;

[0019] In the above formula, For the day-ahead period, the output of thermal power unit i is... The net pumped storage power is the power generated during the daytime period t. This represents the low-frequency trend component of the day-ahead period t.

[0020] The intraday electrochemical energy storage-pumped hydro storage collaborative optimization model aims to minimize the residual after compensation, the equivalent cost of electrochemical energy storage degradation, the intraday operating cost of pumped hydro storage, the penalty for pumped hydro storage deviating from the day-ahead plan, and the penalty for terminal SOC. The constraints include intraday deviation balance constraints, electrochemical energy storage constraints, and intraday constraints for pumped hydro storage. The intraday constraints for pumped hydro storage include intraday operating condition locking constraints, intraday fine-tuning boundary constraints, and intraday ramp-up constraints for adjacent periods of actual net power of pumped hydro storage.

[0021] The objective function of the intraday electrochemical energy storage-pumped hydro storage synergistic optimization model includes:

[0022] ;

[0023] In the above formula, The objective function for the day is... , These are the residual balance penalty coefficients, The compensated residual for time period k. , These represent the charging and discharging power of the electrochemical energy storage during time period k. , These are the daily operating cost coefficient for pumped storage and the penalty coefficient for deviation from the daily plan for pumped storage, respectively. , These represent the power generation and pumping power of the pumped storage system during time period k, respectively. Let k be the fine-tuning power of the pumped storage system during time period k. This is the intraday scheduling time interval. For the terminal SOC penalty item of the intraday rolling forecast window, The duration of the intraday rolling forecast window. Numbering of intraday time periods;

[0024] The intraday deviation balance constraint includes:

[0025] ;

[0026] ;

[0027] In the above formula, Intraday period The compensated residual, Intraday period deviation, Intraday period The net power of electrochemical energy storage, Intraday period Fine-tuning power of pumped storage Intraday period Real-time net load, Intraday period The baseline net load is obtained by mapping the low-frequency trend component to the intraday time scale.

[0028] The pumped storage intraday operating condition locking constraints include:

[0029] ;

[0030] ;

[0031] In the above formula, , Each daytime period The power generation and pumping power of pumped storage hydroelectric power. This is the maximum permissible power of pumped storage. , Each daytime period The power generation state variables and pumping state variables of pumped storage are determined by the intraday locked operating condition sequence;

[0032] The intraday fine-tuning boundary constraints for pumped storage include:

[0033] Under power generation or pumping conditions, , ,in, This is the upper limit of the pumped storage power allowed for minor adjustments within the day. , Each daytime period The actual net power and day-ahead reference net power of pumped storage Determined by the day-ahead reference power sequence;

[0034] The ramp-up constraints for the actual net power output of pumped storage during the day in adjacent time periods include:

[0035] ;

[0036] In the above formula, The allowable ramp rate for the actual net power output of pumped storage per day.

[0037] S2 obtains the low-frequency trend component by performing a low-pass filter on the day-ahead system net load forecast sequence:

[0038] ;

[0039] ;

[0040] In the above formula, , These are the low-frequency trend components for the day-ahead time periods t and t-1, respectively. These are the filter coefficients. The system net load for the day before date t. The time constant of the low-pass filter. This is the day-ahead scheduling time interval.

[0041] The low-pass filter time constant Selected in the following manner:

[0042] Set multiple candidates For each candidate Steps S2-S5 are executed sequentially to obtain the corresponding scheduling schemes. Each scheduling scheme is then comprehensively evaluated based on operational economics, intraday compensation accuracy, day-ahead plan fluctuations, energy storage SOC safety margin, and low-frequency component smoothness. The candidate scheduling scheme with the lowest comprehensive evaluation value is selected. As the final low-pass filter time constant;

[0043] The formula for calculating the comprehensive evaluation value is as follows:

[0044] ;

[0045] ;

[0046] ;

[0047] ;

[0048] ;

[0049] In the above formula, as a candidate The corresponding comprehensive evaluation value, The normalized overall operating cost, This is the normalized compensation error index. This is the normalized day-ahead plan volatility indicator. The normalized energy storage SOC deviation index, This is a normalized low-frequency smoothness index. , , , , They are respectively , , , , The corresponding weights , , , These are the compensation error index, the day-ahead plan fluctuation index, the energy storage SOC deviation index, and the low-frequency smoothness index. The residual after compensation for time period t. This represents the total number of time periods within the day. , The thermal power outputs for time periods t and t-1 are respectively. , These represent the net pumped storage power at time periods t and t-1, respectively. Let SOC be the energy storage state during time period t. This is a reference value for the State of Charge (SOC) of energy storage.

[0050] S4 includes:

[0051] Mapping the daytime pumped storage power generation status, pumping status, and shutdown status to an intraday time scale yields an intraday locked operating condition sequence.

[0052] By mapping the day-ahead pumped storage power generation, pumping power, and net power to an intraday time scale, a day-ahead reference power sequence is obtained.

[0053] Secondly, the present invention provides a multi-type energy storage scheduling system based on net load trend-deviation decoupling, including a day-ahead system net load acquisition module, a net load trend-deviation decoupling module, a day-ahead low-frequency benchmark scheduling module, a day-ahead-intraday mapping module, and an intraday electrochemical energy storage-pumped storage coordinated scheduling module.

[0054] The day-ahead system net load acquisition module is used to acquire the day-ahead system net load prediction sequence under various typical source load operation scenarios.

[0055] The net load trend-deviation decoupling module is used to perform low-frequency trend decomposition on the day-ahead system net load forecast sequence to obtain low-frequency trend components.

[0056] The daytime low-frequency reference scheduling module is used to solve the daytime thermal power-pumped storage low-frequency reference scheduling model with the low-frequency trend component as the daytime scheduling target, and obtain a low-frequency reference scheduling scheme including the daytime operating conditions and daytime power of pumped storage.

[0057] The day-ahead-intraday mapping module is used to map the day-ahead operating conditions of pumped storage to an intraday time scale to obtain an intraday locked operating condition sequence; and to map the day-ahead power of pumped storage to an intraday time scale to obtain a day-ahead reference power sequence.

[0058] The intraday electrochemical energy storage-pumped hydro storage coordinated scheduling module is used to calculate the intraday electrochemical energy storage-pumped hydro storage coordinated optimization model by using the deviation of the intraday real-time net load of pumped hydro storage relative to the day-ahead reference power as the compensation object. This results in intraday scheduling schemes for electrochemical energy storage and intraday fine-tuning schemes for pumped hydro storage under various typical source-load operation scenarios. In this intraday optimization model, the actual intraday operation of pumped hydro storage is based on the day-ahead locked operating condition sequence, coordinating rapid compensation of electrochemical energy storage and limited fine-tuning of pumped hydro storage.

[0059] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0060] 1. This invention discloses a multi-type energy storage scheduling method based on net load trend-deviation decoupling. First, it obtains the day-ahead system net load prediction sequence under various typical source-load operation scenarios, and performs low-frequency trend decomposition on the day-ahead system net load prediction sequence to obtain low-frequency trend components. Then, using the low-frequency trend components as the day-ahead scheduling target, it solves the day-ahead thermal power-pumped storage low-frequency benchmark scheduling model to obtain a low-frequency benchmark scheduling scheme including the day-ahead operating conditions and day-ahead power of pumped storage. Next, it maps the day-ahead operating conditions of pumped storage to the intraday time scale to obtain the intraday locked operating condition sequence, and maps the day-ahead power of pumped storage to the intraday time scale to obtain the day-ahead benchmark power sequence. Finally, it uses the deviation of the intraday real-time net load of pumped storage relative to the day-ahead benchmark power as the compensation object, and continuously solves the intraday electrochemical energy storage-pumped storage collaborative optimization model to obtain the intraday scheduling scheme of electrochemical energy storage and the intraday fine-tuning scheme of pumped storage under various typical source-load operation scenarios. This method uses net load trend-deviation decoupling to determine the task boundaries of different regulation resources, uses the day-ahead low-frequency trend component to determine the pumped storage operating condition, inherits the day-ahead pumped storage operating condition during the intraday phase, does not allow pumped storage to freely switch operating conditions again within a short time scale, and only allows pumped storage to make limited-amplitude fine adjustments under the locked operating condition. At the same time, electrochemical energy storage undertakes rapid deviation compensation, and pumped storage shares the medium and low frequency continuous deviation under the locked operating condition. This can effectively reduce frequent switching of pumped storage in a short time scale and power command step, and improve the economy, smoothness and engineering feasibility of multi-type energy storage scheduling.

[0061] 2. The present invention provides a multi-type energy storage scheduling method based on net load trend-deviation decoupling. When determining the low-pass filter time constant, it comprehensively considers the overall operating cost, compensation error, day-ahead plan fluctuation, energy storage SOC margin, and low-frequency smoothness. It comprehensively evaluates the impact of different low-pass filter time constants from five aspects: economy, accuracy, plan executability, energy storage safety, and filtering effect. This ensures the safe operation of energy storage while taking into account the intraday deviation compensation effect, the smoothness of day-ahead plan, and the overall economic efficiency of system operation. It helps to avoid the problem of poor overall scheduling effect caused by choosing too small or too large a time constant. Attached Figure Description

[0062] Figure 1 This is a schematic diagram of the day-to-day collaborative operation mechanism of the present invention.

[0063] Figure 2 This is a schematic diagram of the IEEE 30-node system topology in Example 1.

[0064] Figure 3 This is a schematic diagram of the overall operating framework of Example 1.

[0065] Figure 4 This is a schematic diagram of typical source-load operation scenarios in spring and summer in Example 1.

[0066] Figure 5 This is a schematic diagram of typical source-load operation scenarios in autumn and winter in Example 1.

[0067] Figure 6 This is a schematic diagram showing the contribution decomposition of intraday electrochemical energy storage compensation and pumped storage fine-tuning in Example 2.

[0068] Figure 7 This is a structural block diagram of the system described in Example 3. Detailed Implementation

[0069] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings.

[0070] This invention proposes a multi-type energy storage scheduling method based on net load trend-deviation decoupling. This method uses net load trend-deviation decoupling to determine the task boundaries of different regulation resources; it uses the day-ahead low-frequency trend component to determine the power output of thermal power plants and the power generation, pumping, and shutdown conditions of pumped storage; it inherits the day-ahead pumped storage operating conditions during the intraday phase, disallowing pumped storage to freely switch operating conditions within a short timescale; under locked operating conditions, only limited-amplitude fine-tuning of pumped storage is allowed, and adjacent time-period ramping constraints are applied to the actual net power of pumped storage during the intraday period; simultaneously, electrochemical energy storage undertakes rapid deviation compensation, while pumped storage shares the burden of continuous low- and medium-frequency deviations under locked operating conditions. That is, the day-ahead model is responsible for determining the structured low-frequency baseline plan, and the intraday model is responsible for compensating for real-time deviations without disrupting the day-ahead pumped storage operating conditions. These mechanisms together form a technical solution for day-ahead-intraday coordinated scheduling of multiple types of energy storage. The day-ahead-intraday coordinated scheduling operation mechanism is as follows: Figure 1 As shown.

[0071] The method described in this invention can achieve the following four technical effects:

[0072] 1. By decoupling the generation of typical scenarios from historical source load data with the net load trend-deviation, the scheduling model can adapt to changes in new energy output and load demand under different typical operating conditions in spring, summer, autumn, and winter.

[0073] 2. By decoupling the net load trend and deviation, the low-frequency trend and high-frequency deviation are separated, so that thermal power units and pumped storage mainly undertake the low-frequency benchmark scheduling task, and electrochemical energy storage mainly undertakes the high-frequency deviation compensation task, thus avoiding the functional overlap of different energy storage resources.

[0074] 3. The pumped storage power generation, pumping, and shutdown conditions are determined in the day-ahead phase. These conditions are inherited in the intraday phase with only limited minor adjustments allowed. At the same time, the actual net power of the pumped storage is subject to ramp-up constraints for adjacent time periods. This can avoid frequent switching of operating conditions in the short-timescale rolling optimization of pumped storage, reduce the mechanical execution risk caused by step jumps in power commands across hours, and improve the engineering feasibility of the optimization results.

[0075] 4. By complementing and synergizing electrochemical energy storage and pumped storage, electrochemical energy storage can quickly respond to short-term deviations, while pumped storage can share the burden of continuous deviations under locked operating conditions. Compared with no intraday compensation or a single energy storage compensation method, it can reduce the residual error after compensation. Compared with the conventional day-ahead-intraday method that directly tracks the original net load, it can reduce the day-ahead planning fluctuations of thermal power and pumped storage, and improve the overall operating economy.

[0076] In this invention, the functional positioning of different regulatory resources is shown in Table 1:

[0077] Table 1 Functional positioning of different regulatory resources

[0078] .

[0079] Example 1:

[0080] This embodiment implements a multi-type energy storage scheduling method based on net load trend-deviation decoupling in the MATPOWER case30 system, such as... Figure 2 As shown, the system comprises 30 nodes and 41 lines. Thermal power units are connected to nodes 1, 2, and 22; pumped storage is connected to node 4; electrochemical energy storage is connected to node 15; photovoltaic power is connected to nodes 10 and 24; and wind power is connected to nodes 15 and 29. The overall operational framework of this method is described in [reference needed]. Figure 3 The specific process includes:

[0081] 1. Based on historical data of annual renewable energy output and load demand in a certain region, construct a typical source-load operation scenario. The specific steps are as follows:

[0082] 1.1 Obtain historical data on wind power, photovoltaic output and load demand for a certain area for 8760 hours throughout the year. Construct daily feature samples for wind power, photovoltaic and load according to a 24-hour period. Each daily feature matrix consists of 3 variables and 24 time points, forming a 3×24 daily feature matrix. Expand it into a 72-dimensional daily feature vector in chronological order to obtain 365 daily feature samples for the whole year.

[0083] 1.2. The daily characteristic data of wind power, photovoltaic and load are normalized to the maximum value, and physical quantities of different dimensions are uniformly mapped to the [0,1] interval to eliminate the influence of numerical magnitude differences on subsequent clustering distance calculation.

[0084] 1.3. In view of the seasonal differences in source and load characteristics, the above daily feature data are divided into four seasonal scene subsets: spring, summer, autumn and winter, according to seasonal characteristics. Each subset contains 92, 92, 91 and 90 high-dimensional daily feature vectors, respectively.

[0085] 1.4. The K-Medoids clustering method is used to cluster the high-dimensional daily feature vectors of each scene subset, and the real samples that minimize the Euclidean distance within each subset are extracted as typical daily curves to obtain four types of typical source-load operation scenarios for spring, summer, autumn and winter.

[0086] 1.5 Preferably, a high-fluctuation weighted typical scenario can be constructed based on the verification target as the final typical source-load operation scenario.

[0087] To verify the scheduling effect under conditions of high renewable energy fluctuations and high regulation pressure, a high-fluctuation weighted typical scenario is constructed. In this embodiment, the weights of four typical scenarios—spring, summer, autumn, and winter—are set to 0.05, 0.80, 0.10, and 0.05, respectively. This scenario is not the annual average operating scenario, but rather a representative operating condition characterized by strong renewable energy fluctuations and prominent regulation demands.

[0088] For any source load time series among load, wind power output, and photovoltaic output, its high-fluctuation weighted typical scenario sequence can be represented as:

[0089] ;

[0090] In the above formula, This represents the sequence value of source load type m in a typical high-fluctuation weighted scenario for time period t. , , , These represent load, wind power output, and photovoltaic power output, respectively. , , , These represent the sequence values ​​of source load type m for time period t under four typical scenarios: spring, summer, autumn, and winter. , , , The weights are for typical scenarios in spring, summer, autumn, and winter, respectively.

[0091] The weights of typical scenarios can be determined based on the intensity of renewable energy fluctuations, peak-to-valley load differences, the strength of regulation demand, or research and verification objectives, and are not limited to fixed values. When the research object is the coordinated operation of multiple types of energy storage under high renewable energy fluctuations and high regulation pressure, the weights of summer or other high-fluctuation seasons can be increased, while retaining some source-load characteristics of other seasons. This weighted typical scenario is not an annual average scenario, but is used to construct representative operating conditions under high regulation pressure to verify the effectiveness of trend-deviation decoupling, pumped storage operating condition locking, and multi-type energy storage coordinated compensation mechanisms.

[0092] The schematic diagrams of the typical source-load operation scenarios obtained in this embodiment are as follows: Figures 4-5 As shown.

[0093] 2. Obtain the day-ahead system net load prediction sequence under each typical source load operation scenario.

[0094] For node j, the node's net load is expressed as:

[0095] ;

[0096] In the above formula, The net load at node j during the day-ahead period t. For the load at node j in the day-ahead time period t, , These represent the photovoltaic and wind power injected at node j during the daytime period t, respectively.

[0097] The system net load is the sum of the net loads of each node:

[0098] ;

[0099] In the above formula, The system net load for the day before date t.

[0100] Based on the above definition, the uncertainty of source load can be uniformly transformed into the equivalent change in system demand, providing a unified object for low-frequency benchmark planning and intraday deviation compensation.

[0101] 3. Low-pass filtering is applied to the day-ahead net load forecast sequence to achieve net load trend-deviation decoupling, resulting in low-frequency trend components and high-frequency deviation components.

[0102] The low-pass filter coefficients and their decomposition relationships are as follows:

[0103] ;

[0104] ;

[0105] ;

[0106] In the above formula, , These are the low-frequency trend components for the day-ahead time periods t and t-1, respectively. These are the filter coefficients. The system net load for the day before date t. The time constant of the low-pass filter. The day-ahead scheduling interval, The high-frequency deviation component is used for day-ahead thermal power-pumped storage benchmark dispatch, while the high-frequency deviation component is used for intraday electrochemical energy storage and pumped storage fine-tuning compensation.

[0107] Preferably, to reduce the phase lag caused by unidirectional low-pass filtering, a forward-reverse zero-phase-shift filter is used. First, the day-ahead forecast net load is low-pass filtered in forward time sequence. Then, the forward filtering result is reversed and filtered again. Finally, the original time sequence is restored to obtain the final low-frequency component, which is:

[0108] ;

[0109] In the above formula, , These represent the forward filtering result for the time period t before the day, and the low-frequency components restored to the original time sequence after reverse filtering. The system net load for the day before date t.

[0110] The above treatment retains the low-frequency smoothing effect while avoiding the misalignment of peak and valley periods, ensuring that the peak shaving and valley filling actions of pumped storage are consistent with the net load trend.

[0111] 4. For any typical source-load operation scenario, construct the corresponding day-ahead thermal power-pumped storage low-frequency benchmark scheduling model.

[0112] The day-ahead model uses low-frequency trend components as the day-ahead scheduling objective. Decision variables include the start-up / shutdown status and output of thermal power units, the generation / pumping / shutdown status and power of pumped storage hydropower, the state of charge (SOC) of pumped storage, node-injected power, line power flow, and necessary safety slack variables. Its objective function aims to comprehensively minimize the operating and start-up / shutdown costs of thermal power units, the day-ahead operating costs of pumped storage hydropower, the equivalent cost of day-ahead planning fluctuations, and the penalty costs for power balance and line safety slack, while satisfying system power balance, thermal power unit operation, day-ahead operation of pumped storage hydropower, and network security constraints. Specific objective functions include:

[0113] ;

[0114] ;

[0115] ;

[0116] ;

[0117] ;

[0118] In the above formula, For overall cost, For the operation and start-up / shutdown costs of thermal power units, , , These are the output cost coefficient, no-load cost coefficient, and start-up cost coefficient of thermal power unit i, respectively. The day-ahead operating cost of pumped storage, For the day-ahead period, the output of thermal power unit i is... , These are the operating status variables and startup variables of thermal power unit i during the daytime period t, respectively. This represents the day-ahead operating cost coefficient for pumped storage. , These represent the power generation and pumping power of the pumped storage system during the daytime period t, respectively. The equivalent cost of the current planned fluctuations, , These are the ramp smoothing penalty coefficient for thermal power units and the day-ahead baseline power fluctuation penalty coefficient for pumped storage power units, respectively. For the day-ahead time period t, the ramp-up auxiliary variable for thermal power unit i is... For the collection of thermal power units, For the day-ahead time period t, the auxiliary variable is the day-ahead baseline power fluctuation of pumped storage. To offset the cost of power balancing and line safety slack, , These are the power balance relaxation penalty factor and the line safety relaxation penalty factor, respectively. , These represent the positive and negative slack variables of the power balance during the daytime period t, respectively. The day-ahead scheduling interval, For the set of routes, For the daytime period t line Safety slack variables, This represents the total number of scheduling periods in the previous day;

[0119] Among them, auxiliary variables for climbing slope of thermal power units Auxiliary variable for day-ahead reference power fluctuation of pumped storage units, used to characterize the output variation of thermal power units between adjacent day-ahead periods. This is used to characterize the variation of the day-ahead baseline net power of pumped storage between adjacent day-ahead periods. To transform the absolute value form of power change into a linear constraint, the above auxiliary variables can be made to satisfy:

[0120] .

[0121] Based on the above constraints It can be used to represent the absolute value of the change in power output of a thermal power unit between adjacent time periods. It can be used to represent the absolute value of the change in the day-ahead benchmark net power of pumped storage between adjacent time periods, thereby penalizing the thermal power ramp-up and pumped storage benchmark power fluctuation in the day-ahead objective function.

[0122] Through the above objective function, the day-ahead model can track low-frequency net load while comprehensively considering thermal power operation, thermal power start-up and shutdown, pumped storage operation, day-ahead planning smoothness, and network security feasibility.

[0123] The constraints of the current thermal power-pumped storage low-frequency benchmark dispatch model include the following categories:

[0124] (1) System power balance constraint. This constraint is used to ensure that the output of thermal power and the net output of pumped storage power jointly track the low-frequency net load demand in each day-ahead period. Since the low-frequency net load has been obtained by subtracting the output of new energy sources such as wind power and photovoltaic power from the load and through trend-deviation decoupling, the new energy output and load demand items are no longer listed separately in this constraint:

[0125] ;

[0126] In the above formula, For the day-ahead period, the output of thermal power unit i is... Let be the net pumped storage power during the day-ahead period t, and , Let be the low-frequency trend component of the day-ahead period t. The left side of this constraint represents the deviation between day-ahead supply and low-frequency net load, while the right side uses positive and negative slack variables to characterize the direction of the deviation, and a large penalty coefficient is set in the objective function.

[0127] When power balance is described using nodal injection, the net injected power at each node can be expressed as:

[0128] ;

[0129] And satisfy:

[0130] ;

[0131] In the above formula, This represents the net injected power at node i during the day-ahead time period t. For the day-ahead period, the output of thermal power unit g is t. , These represent the power generation and pumping power of the pumped storage system at node i during the daytime period t, respectively. The low-frequency net load of node i during the day-ahead period t can be obtained from the day-ahead net load sequence of node i through trend-deviation decoupling, or it can be obtained by allocating the system's low-frequency net load to each node according to the proportion of each node's day-ahead net load to the system's day-ahead net load. Let i be the set of thermal power units at node i. It is a set of nodes.

[0132] The injection method of this node is consistent with the low-frequency net load scheduling caliber and provides input for subsequent PTDF line power flow calculation.

[0133] (2) Operating constraints of thermal power units. These constraints include upper and lower limits of output, start-up and shutdown logic, minimum start-up and shutdown time, ramp-up constraints, and standby constraints.

[0134] Output upper and lower limit constraints can be expressed as:

[0135] ;

[0136] Start-stop logic constraints can be expressed as:

[0137] ;

[0138] The minimum startup time and minimum downtime constraints can be expressed as:

[0139] ;

[0140] The climbing constraint can be expressed as:

[0141] ;

[0142] The spare constraint can be expressed as:

[0143] ;

[0144] In the above formula, , These are the minimum and maximum outputs of thermal power unit i, respectively. Let i be the shutdown variable for thermal power unit i during the day-ahead period t. For time period The operating state variables of thermal power unit i , These represent the minimum start-up time and minimum shutdown time for thermal power unit i, respectively. , These represent the upward and downward climbing capabilities of thermal power unit i, respectively. , These represent the allowable output variation boundaries for thermal power unit i during startup and shutdown, respectively. This represents the required reserve capacity for the day-ahead period t.

[0145] With the above constraints, thermal power units can not only undertake low-frequency reference output, but also meet the operational requirements of start-up, shutdown, ramping and standby.

[0146] (3) Day-ahead operating constraints for pumped storage. These constraints include power generation / pumping mutual exclusion, power boundary, SOC recursion, upper and lower limits of SOC, and end-of-day recovery constraints.

[0147] The mutual exclusion constraint between power generation and pumping is as follows:

[0148] ;

[0149] The power boundary constraints are:

[0150] ;

[0151] The SOC recursive constraint is:

[0152] ;

[0153] SOC upper and lower limits and end-of-day recovery constraints can be expressed as:

[0154] ;

[0155] In the above formula, , These represent the power generation and pumping power of the pumped storage system during the daytime period t, respectively. The maximum allowable power of pumped storage, i.e., the rated power. , These are the power generation state variables and pumping state variables for pumped storage during the daytime period t, respectively. For the day-ahead period t, the SOC of pumped storage is... , These are pumping efficiency and power generation efficiency, respectively. This refers to the equivalent energy capacity of pumped hydro storage. , These are the lower and upper limits of the State of Charge (SOC) for pumped storage, respectively. , The pumped storage SOCs are the initial and final time periods, respectively.

[0156] The above constraints ensure that the pumped storage day-ahead plan meets the requirements of power generation / pumping mutual exclusion, power boundary and energy sustainability.

[0157] (4) Network security constraints

[0158] To ensure that the day-ahead dispatching scheme meets the operational feasibility requirements of the power grid, the PTDF linear power flow model is used to describe the line power flow. The day-ahead time period t for each line... trend It can be represented as:

[0159] ;

[0160] In the above formula, Net power injection for node i to the line Power distribution factor of power flow This represents the net injected power at node i during the day-ahead period t.

[0161] The line capacity constraint is:

[0162] ;

[0163] In the above formula, For the line Maximum transmission capacity , Line t for the daytime period The current flow safety positive and negative slack variables are used, and a large penalty coefficient is set in the objective function.

[0164] By using power balance constraints and PTDF line power flow constraints, it can be ensured that the day-ahead low-frequency reference dispatch scheme not only meets energy balance but also meets the safety requirements of power grid lines.

[0165] 5. Using the low-frequency trend component as the day-ahead scheduling input, call the mixed integer linear programming solver to solve the day-ahead thermal power-pumped storage low-frequency benchmark scheduling model to obtain the low-frequency benchmark scheduling scheme. The low-frequency benchmark scheduling scheme includes the day-ahead thermal power unit start-up and shutdown status, thermal power unit output, pumped storage power generation / pumping / shutdown conditions, pumped storage power generation, pumping power and net power, SOC trajectory and line power flow results.

[0166] 6. Map the daytime operating conditions of pumped storage (including daytime power generation, pumping, and shutdown states) to a 15-minute intraday timescale using a hold-type mapping to obtain an intraday locked operating condition sequence (including intraday power generation, pumping, and shutdown state variables); map the daytime power of pumped storage (including daytime power generation, pumping power, and net power) to a 15-minute intraday timescale using a hold-type mapping to obtain a daytime reference power sequence (including daytime reference power generation, pumping power, and net power). The mapping relationship is as follows:

[0167] ;

[0168] In the above formula, , These are the day-ahead pumped storage power generation state variables and the pumping state variables, respectively. , Each daytime period The power generation state variables and pumping state variables of pumped storage. This formula shows that the daily pumped storage state inherits the daily plan and does not switch freely.

[0169] Among them, the actual net power of pumped storage during the day and the reference net power before the day are defined as follows:

[0170] ;

[0171] The intraday fine-tuning power of pumped storage is defined as follows:

[0172] ;

[0173] In the above formula, , Each daytime period The actual net power and day-ahead reference net power of pumped storage , Each daytime period The power generation and pumping power of pumped storage hydroelectric power. , Each daytime period The day-ahead baseline generating capacity and pumping capacity of pumped storage, Intraday period Fine-tuning power of pumped storage.

[0174] Therefore, in subsequent fine-tuning of boundary constraints This indicates the deviation of the actual net power of pumped storage within a day from the baseline net power of the day before.

[0175] During the intraday phase, the actual operation of pumped storage is based on the day-ahead locked operating conditions; the intraday model does not freely re-determine the pumped storage's power generation, pumping, or shutdown status. The corresponding relationship between the locked operating conditions and the actual intraday power output can be expressed as:

[0176] ;

[0177] ;

[0178] ;

[0179] In the above formula, This is the maximum allowable power for pumped storage.

[0180] The aforementioned relationships constitute the fundamental mechanism for the day-ahead and intraday linkage of pumped storage, used to clarify the correspondence between the day-ahead locked operating condition, the intraday actual net power, and the intraday fine-tuning power. Specific fine-tuning boundaries, SOC safety constraints, and actual net power ramp-up constraints are further applied in the intraday rolling optimization model.

[0181] 7. Establish a synergistic optimization model for intraday electrochemical energy storage and pumped hydro storage.

[0182] This intraday model uses the deviation of real-time net load from the day-ahead low-frequency benchmark as the compensation object. Based on the day-ahead operating condition lock of pumped storage, it coordinates rapid compensation by electrochemical energy storage and limited fine-tuning by pumped storage. The objective function of the intraday model comprehensively considers the residual penalty cost after compensation, the equivalent cost of electrochemical energy storage degradation, the intraday operating cost of pumped storage, the penalty cost of pumped storage deviating from the day-ahead plan, and the terminal SOC penalty. It adopts an intraday rolling execution rule, with each intraday rolling optimization occurring within an intraday rolling forecast window (starting from the current time period and including a total of...). The control sequence is solved within a time period, but only the control inputs for the first time period within the current rolling window are actually executed. The electrochemical storage SOC, pumped storage SOC, and the latest net load deviation after execution are used as the initial state for the next rolling window. The objective function can be specifically expressed as:

[0183] ;

[0184] In the above formula, The objective function for the day is... , These are the residual balance penalty coefficients, The compensated residual for time period k. , These represent the charging and discharging power of the electrochemical energy storage during time period k. , These are the daily operating cost coefficient for pumped storage and the penalty coefficient for deviation from the daily plan for pumped storage, respectively. , These represent the power generation and pumping power of the pumped storage system during time period k, respectively. Let k be the fine-tuning power of the pumped storage system during time period k. This is the intraday scheduling time interval. For the terminal SOC penalty item of the intraday rolling forecast window, The duration of the intraday rolling forecast window. Numbering of intraday time periods;

[0185] The constraints of intraday models include the following categories:

[0186] (1) Intraday Deviation Balance Constraint

[0187] ;

[0188] ;

[0189] ;

[0190] In the above formula, Intraday period The compensated residual, Intraday period The deviation represents the difference between the real-time net load and the day-ahead low-frequency reference. Intraday period The net power of electrochemical energy storage, Intraday period The intraday fine-tuning power of pumped storage, Intraday period Real-time net load, Intraday period The baseline net load is obtained by mapping the low-frequency trend component to the intraday time scale. , Each daytime period The charging and discharging power of electrochemical energy storage;

[0191] (2) Electrochemical energy storage constraints

[0192] Electrochemical energy storage charging and discharging mutual exclusion constraints. Electrochemical energy storage can only be charged, discharged, or left idle within the same time period of the day, and its constraints are as follows:

[0193] ;

[0194] In the above formula, , Each daytime period The charge and discharge states of electrochemical energy storage , Indicates charging. Indicates discharge. This represents the maximum power of electrochemical energy storage.

[0195] Since the intraday model contains the aforementioned 0-1 state variables and quadratic objective terms, the intraday rolling optimization model in this embodiment belongs to the mixed integer quadratic programming predictive control model.

[0196] Electrochemical energy storage SOC constraint. The recursive relationship for electrochemical energy storage SOC is:

[0197] ;

[0198] The upper and lower limits of SOC are:

[0199] ;

[0200] In the above formula, Intraday period Electrochemical energy storage SOC , These refer to the electrochemical energy storage charging efficiency and discharging efficiency, respectively. For electrochemical energy storage capacity, , These are the lower and upper limits of the state of charge (SOC) for electrochemical energy storage, respectively.

[0201] (3) Daily constraints of pumped storage

[0202] Pumped storage hydroelectric power generation and pumping status are locked within the day. The power generation and pumping status of pumped storage hydroelectric power generation are inherited from the previous day's status and do not switch freely during the day's rolling process. The following constraints apply:

[0203] ;

[0204] ;

[0205] .

[0206] This constraint is the core constraint that ensures the feasibility of pumped storage projects in this invention.

[0207] Pumped storage intraday fine-tuning boundary constraints. To prevent pumped storage from deviating too much from the day-ahead plan, fine-tuning power boundaries are set:

[0208] Under power generation conditions, The generator will remain in operation throughout the day, with only limited adjustments allowed to the power generation capacity.

[0209] Under pumping conditions, The pumping operation will continue throughout the day, with only limited adjustments to the pumping power permitted.

[0210] During shutdown conditions, The pumped storage system does not operate.

[0211] in, , This is the upper limit of the pumped storage power allowed for minor adjustments within the day. , Each daytime period The actual net power and day-ahead reference net power of pumped storage Determined from the day-ahead reference power sequence.

[0212] Ramp-up constraints on pumped storage's intraday actual net power during adjacent time periods. To address potential command jumps between adjacent time periods of the day-ahead baseline plan, especially at hourly boundaries, the intraday ramp-up constraints are applied directly to the pumped storage's intraday actual net power, rather than just to the fine-tuned power relative to the day-ahead plan. The constraints are as follows:

[0213] ;

[0214] In the above formula, This represents the allowable ramp rate for the actual net power output of pumped storage units within a day. This constraint directly limits the variation in the actual net power output issued to the pumped storage units, reducing the impact of hydraulic transition processes and mechanical execution risks caused by step commands across hours.

[0215] Pumped storage intraday SOC constraints. The recursive relationship for pumped storage intraday SOC is as follows:

[0216] ;

[0217] ;

[0218] In the above formula, Intraday period Pumped storage (SOC).

[0219] The above constraints ensure that pumped storage still meets the energy state safety boundary under intraday locked operating conditions and limited fine-tuning conditions.

[0220] The intraday model of this invention does not directly use the parameter of "high-frequency deviation component," but rather reflects it through the "deviation between real-time net load and the day-ahead low-frequency benchmark." Since the day-ahead low-frequency benchmark is formed by low-frequency trend components, the deviation of real-time net load relative to this benchmark is the high-frequency fluctuation and prediction error that needs to be compensated for during the intraday phase, which is subsequently compensated by electrochemical energy storage and pumped hydro storage fine-tuning.

[0221] 8. Read the electrochemical energy storage SOC, pumped storage SOC, pumped storage day-ahead locked condition, day-ahead reference power, and real-time net load deviation at the initial moment of intraday rolling optimization, and set the rolling prediction window length. Within the current rolling window, based on the deviation between the real-time net load and the day-ahead low-frequency reference, call the mixed integer quadratic programming solver to solve the intraday electrochemical energy storage-pumped storage co-optimization model, and obtain the intraday scheduling scheme for electrochemical energy storage within the window, including electrochemical energy storage charging power, discharging power, pumped storage intraday fine-tuning power, post-compensation residual, and energy storage SOC prediction sequence.

[0222] In this embodiment, the rated power of pumped storage is 23.70MW, the equivalent energy capacity is 106.35MWh, the power generation efficiency and pumping efficiency are both 0.75, the SOC range is [0.12, 1.00], and the initial SOC is 0.50; the daily fine-tuning ratio of pumped storage is 0.65, therefore the upper limit of the daily allowable fine-tuning power is... The allowable ramp rate for the actual net daily power output of pumped storage is 15.405 MW. .

[0223] The rated power of the electrochemical energy storage is 8.15MW, the rated capacity is 16.20MWh, the charging efficiency and discharging efficiency are both 0.90, the SOC range is [0.12, 0.88], and the initial SOC is 0.50.

[0224] 9. Execute the intraday dispatch plan for the current period, that is, issue electrochemical energy storage charging and discharging commands and fine-tuning commands under the pumped storage lockout condition for the current period.

[0225] 10. Update the energy storage status and proceed to the next intraday rolling forecast window.

[0226] Based on the executed electrochemical energy storage charging and discharging power, pumped storage power generation / pumping power, real-time net load, and measurement feedback, update the electrochemical energy storage SOC, pumped storage SOC, compensated residual, and the initial state of the model for the next time period; then move the rolling prediction window forward by one intraday period and repeat step 7 until the intraday scheduling cycle of this typical source-load operation scenario ends.

[0227] 11. Summarize and output the coordinated scheduling results. For each typical source-load operation scenario, output the daily output plan of thermal power, the daily generation / pumping / shutdown conditions of pumped storage, the daily fine-tuning power of pumped storage, the charging and discharging power of electrochemical energy storage, the residual after compensation, the SOC trajectory of electrochemical energy storage, the SOC trajectory of pumped storage, the line power flow safety verification results, and the comprehensive operating cost.

[0228] Example 2:

[0229] The overall steps are the same as in Example 1, except that:

[0230] To determine the low-pass filter time constant This embodiment sets multiple candidate time constants and performs day-to-day collaborative scheduling calculations under the same typical source-load operation scenario, equipment parameters, and scheduling constraints. Specifically,

[0231] For each candidate The process proceeds sequentially through steps 3-10 of Example 1 (first, performing forward-reverse low-pass filtering on the day-ahead predicted net load to obtain low-frequency trend components and high-frequency deviation components; then forming a day-ahead baseline scheduling plan based on the low-frequency trend components, and performing rolling corrections based on real-time net load deviations during the intraday phase), resulting in the corresponding scheduling scheme. A comprehensive evaluation of each scheduling scheme is then conducted based on operational economics, intraday compensation accuracy, day-ahead plan fluctuations, energy storage SOC safety margin, and low-frequency component smoothness. The candidate scheduling scheme corresponding to the one with the lowest comprehensive evaluation value is selected. This serves as the final low-pass filter time constant. Among them,

[0232] Since the dimensions and numerical ranges of the various evaluation indicators are different, each indicator needs to be normalized first, and then the comprehensive evaluation value is calculated based on the following formula:

[0233] ;

[0234] ;

[0235] ;

[0236] ;

[0237] ;

[0238] In the above formula, as a candidate The corresponding comprehensive evaluation value, The smaller the value, the better. The better the overall performance in terms of system operating costs, compensation errors, plan fluctuations, and energy storage status deviations, the better. The normalized comprehensive operating cost comprises two parts: day-ahead dispatch cost and intraday rolling correction cost. Day-ahead dispatch cost includes the operation and start-up / shutdown of thermal power units, the day-ahead operating cost of pumped storage, and the day-ahead plan fluctuation penalty cost. Intraday rolling correction cost includes the equivalent cost of electrochemical energy storage degradation, the intraday operating cost of pumped storage, the pumped storage deviation penalty cost, and the compensation residual penalty cost. This is the normalized compensation error index. This is the normalized day-ahead plan volatility indicator. The normalized energy storage SOC deviation index, This is a normalized low-frequency smoothness index. , , , , They are respectively , , , , The corresponding weights , , , These are the compensation error index, the day-ahead plan fluctuation index, the energy storage SOC deviation index, and the low-frequency smoothness index. The residual after compensation for time period t. This represents the total number of time periods within the day. , The thermal power outputs for time periods t and t-1 are respectively. , These represent the net pumped storage power at time periods t and t-1, respectively. Let SOC be the energy storage state during time period t. The SOC reference value for energy storage can be the midpoint between the upper and lower limits of the energy storage SOC.

[0239] This embodiment Select 0.65h, the day-ahead scheduling time interval. Take 1h, corresponding to the filter coefficients .

[0240] In the intraday scheduling scheme obtained in this embodiment, the output contributions of electrochemical energy storage and pumped hydro storage are as follows: Figure 6 As shown.

[0241] To examine the effectiveness of the method described in this embodiment, five sets of control cases were set up, including no intraday compensation, conventional day-to-day method, electrochemical energy storage compensation only, pumped storage fine-tuning only, and the method of this embodiment. The functional configuration of each case is shown in Table 2:

[0242] Table 2 Functional configuration of the five sets of examples

[0243] .

[0244] The comprehensive operating cost, compensated RMSE, maximum residual, thermal power ramp rate, and pumped storage day-ahead ramp rate for each calculation example are statistically analyzed. The results are shown in Table 3.

[0245] Table 3. Costs (RMB), Compensation Errors, and Daily Planned Ramp-up Indicators for Each Calculation Example in Typical Scenarios

[0246] .

[0247] The SOC range of energy storage, the number of simultaneous charging and discharging operations of electrochemical energy storage, the number of violations of pumped storage operating conditions, and the number of violations of pumped storage ramp-up operations were statistically analyzed for each example. The results are shown in Table 4.

[0248] Table 4 shows the energy storage SOC range, the number of simultaneous charging and discharging cycles of electrochemical energy storage, and the number of violations for each calculation example.

[0249] .

[0250] As shown in Tables 3 and 4, the method described in this invention can meet the requirements of energy storage SOC boundary, pumped storage operating condition locking, and intraday actual net power ramping while reducing overall operating costs and day-ahead planned ramping, thus verifying the feasibility of the method described in this invention.

[0251] Example 3:

[0252] See Figure 7 A multi-type energy storage dispatching system based on net load trend-deviation decoupling includes a day-ahead system net load acquisition module, a net load trend-deviation decoupling module, a day-ahead low-frequency benchmark dispatching module, a day-ahead-intraday mapping module, and an intraday electrochemical energy storage-pumped storage collaborative dispatching module.

[0253] The day-ahead system net load acquisition module is used to acquire the day-ahead system net load prediction sequence under various typical source load operation scenarios. For the specific execution method of this module, please refer to steps 1-2 of Embodiment 1.

[0254] The net load trend-deviation decoupling module is used to perform low-frequency trend decomposition on the day-ahead system net load forecast sequence to obtain low-frequency trend components. For the specific execution method of this module, please refer to step 3 of Example 1.

[0255] The day-ahead low-frequency reference scheduling module is used to solve the day-ahead thermal power-pumped storage low-frequency reference scheduling model with the low-frequency trend component as the day-ahead scheduling target, and obtain a low-frequency reference scheduling scheme including the day-ahead operating conditions and day-ahead power of pumped storage. For the specific execution method of this module, please refer to steps 4-5 of Example 1.

[0256] The day-ahead-intraday mapping module is used to map the day-ahead operating conditions of pumped storage to an intraday time scale to obtain an intraday locked operating condition sequence; and to map the day-ahead power of pumped storage to an intraday time scale to obtain a day-ahead reference power sequence. For the specific execution method of this module, please refer to step 6 of Embodiment 1.

[0257] The intraday electrochemical energy storage-pumped hydro storage coordinated scheduling module is used to calculate the intraday electrochemical energy storage-pumped hydro storage coordinated optimization model by using the deviation of the intraday real-time net load of pumped hydro storage relative to the day-ahead reference power as the compensation object, and to obtain the intraday scheduling scheme of electrochemical energy storage and the intraday fine-tuning scheme of pumped hydro storage under various typical source-load operation scenarios. For the specific execution method of this module, please refer to steps 7-10 of Example 1.

Claims

1. A multi-type energy storage scheduling method based on net load trend-deviation decoupling, characterized in that: The method includes: S1. Obtain the day-ahead system net load prediction sequence under each typical source-load operation scenario; S2. Perform low-frequency trend decomposition on the day-ahead system net load forecast sequence to obtain low-frequency trend components; S3. Using the low-frequency trend component as the day-ahead scheduling target, solve the day-ahead thermal power-pumped storage low-frequency benchmark scheduling model to obtain a low-frequency benchmark scheduling scheme that includes the day-ahead operating conditions and day-ahead power of pumped storage. S4. Map the day-ahead operating conditions of pumped storage to the intraday time scale to obtain the intraday locked operating condition sequence; map the day-ahead power of pumped storage to the intraday time scale to obtain the day-ahead reference power sequence. S5. Taking the deviation of the real-time net load of pumped storage within a day from the daytime baseline power as the compensation object, the intraday electrochemical energy storage-pumped storage collaborative optimization model is solved in a rolling manner to obtain the intraday scheduling scheme of electrochemical energy storage and the intraday fine-tuning scheme of pumped storage under various typical source-load operation scenarios. In this intraday optimization model, the actual intraday operation of pumped storage is based on the daytime locked operating condition sequence, coordinating the rapid compensation of electrochemical energy storage and the limited fine-tuning of pumped storage.

2. The multi-type energy storage scheduling method based on net load trend-deviation decoupling according to claim 1, characterized in that: The day-ahead thermal power-pumped storage low-frequency benchmark scheduling model aims to minimize the operating and start-up costs of thermal power units, the day-ahead operating costs of pumped storage, the equivalent costs of day-ahead planning fluctuations, and the costs of power balance and line safety relaxation penalties. Its constraints include system power balance constraints, thermal power unit operating constraints, pumped storage day-ahead operating constraints, and network security constraints.

3. The multi-type energy storage scheduling method based on net load trend-deviation decoupling according to claim 2, characterized in that: The day-ahead planned fluctuation equivalent cost, power balance, and line safety slack penalty cost are calculated based on the following formula: ； ； In the above formula, The equivalent cost of the current planned fluctuations, , These are the ramp smoothing penalty coefficient for thermal power units and the day-ahead baseline power fluctuation penalty coefficient for pumped storage power units, respectively. For the day-ahead time period t, the ramp-up auxiliary variable for thermal power unit i is... For the collection of thermal power units, For the day-ahead time period t, the auxiliary variable is the day-ahead baseline power fluctuation of pumped storage. To offset the cost of power balancing and line safety slack, , These are the power balance relaxation penalty factor and the line safety relaxation penalty factor, respectively. , These represent the positive and negative slack variables of the power balance during the daytime period t, respectively. The day-ahead scheduling interval, For the set of routes, For the daytime period t line The trend of safety relaxation variables, This represents the total number of scheduling periods in the previous day; The system power balance constraints include: ； In the above formula, For the day-ahead period, the output of thermal power unit i is... The net pumped storage power is the power generated during the daytime period t. This represents the low-frequency trend component of the day-ahead period t.

4. The multi-type energy storage scheduling method based on net load trend-deviation decoupling according to claim 1, characterized in that: The intraday electrochemical energy storage-pumped hydro storage collaborative optimization model aims to minimize the residual after compensation, the equivalent cost of electrochemical energy storage degradation, the intraday operating cost of pumped hydro storage, the penalty for pumped hydro storage deviating from the day-ahead plan, and the penalty for terminal SOC. The constraints include intraday deviation balance constraints, electrochemical energy storage constraints, and intraday constraints for pumped hydro storage. The intraday constraints for pumped hydro storage include intraday operating condition locking constraints, intraday fine-tuning boundary constraints, and intraday ramp-up constraints for adjacent periods of actual net power of pumped hydro storage.

5. A multi-type energy storage scheduling method based on net load trend-deviation decoupling according to claim 4, characterized in that: The objective function of the intraday electrochemical energy storage-pumped hydro storage synergistic optimization model includes: ； In the above formula, The objective function for the day is... , These are the residual balance penalty coefficients, The compensated residual for time period k. , These represent the charging and discharging power of the electrochemical energy storage during time period k. , These are the daily operating cost coefficient for pumped storage and the penalty coefficient for deviation from the daily plan for pumped storage, respectively. , These represent the power generation and pumping power of the pumped storage system during time period k, respectively. Let k be the fine-tuning power of the pumped storage system during time period k. This is the intraday scheduling time interval. For the terminal SOC penalty item of the intraday rolling forecast window, The duration of the intraday rolling forecast window. Numbering of intraday time periods; The intraday deviation balance constraint includes: ； ； In the above formula, Intraday period The compensated residual, Intraday period deviation, Intraday period The net power of electrochemical energy storage, Intraday period Fine-tuning power of pumped storage Intraday period Real-time net load, Intraday period The baseline net load is obtained by mapping the low-frequency trend component to the intraday time scale.

6. A multi-type energy storage scheduling method based on net load trend-deviation decoupling according to claim 4, characterized in that: The pumped storage intraday operating condition locking constraints include: ； ； In the above formula, , Each daytime period The power generation and pumping power of pumped storage hydroelectric power. This is the maximum permissible power of pumped storage. , Each daytime period The power generation state variables and pumping state variables of pumped storage are determined by the intraday locked operating condition sequence; The intraday fine-tuning boundary constraints for pumped storage include: Under power generation or pumping conditions, , ,in, This is the upper limit of the pumped storage power allowed for minor adjustments within the day. , Each daytime period The actual net power and day-ahead reference net power of pumped storage Determined by the day-ahead reference power sequence; The ramp-up constraints for the actual net power output of pumped storage during the day in adjacent time periods include: ； In the above formula, The allowable ramp rate for the actual net power output of pumped storage per day.

7. A multi-type energy storage scheduling method based on net load trend-deviation decoupling according to claim 1, characterized in that: S2 obtains the low-frequency trend component by performing a low-pass filter on the day-ahead system net load forecast sequence: ； ； In the above formula, , These are the low-frequency trend components for the day-ahead time periods t and t-1, respectively. These are the filter coefficients. The system net load for the day before date t. The time constant of the low-pass filter. This is the day-ahead scheduling time interval.

8. A multi-type energy storage scheduling method based on net load trend-deviation decoupling according to claim 7, characterized in that: The low-pass filter time constant Selected in the following manner: Set multiple candidates For each candidate Steps S2-S5 are executed sequentially to obtain the corresponding scheduling schemes. Each scheduling scheme is then comprehensively evaluated based on operational economics, intraday compensation accuracy, day-ahead plan fluctuations, energy storage SOC safety margin, and low-frequency component smoothness. The candidate scheduling scheme with the lowest comprehensive evaluation value is selected. As the final low-pass filter time constant; The formula for calculating the comprehensive evaluation value is as follows: ； ； ； ； ； In the above formula, as a candidate The corresponding comprehensive evaluation value, The normalized overall operating cost, This is the normalized compensation error index. This is the normalized day-ahead plan volatility indicator. The normalized energy storage SOC deviation index, This is a normalized low-frequency smoothness index. , , , , They are respectively , , , , The corresponding weights , , , These are the compensation error index, the day-ahead plan fluctuation index, the energy storage SOC deviation index, and the low-frequency smoothness index. The residual after compensation for time period t. This represents the total number of time periods within the day. , The thermal power outputs for time periods t and t-1 are respectively. , These represent the net pumped storage power at time periods t and t-1, respectively. Let SOC be the energy storage state during time period t. This is a reference value for the State of Charge (SOC) of energy storage.

9. A multi-type energy storage scheduling method based on net load trend-deviation decoupling according to claim 1, characterized in that: S4 includes: Mapping the daytime pumped storage power generation status, pumping status, and shutdown status to an intraday time scale yields an intraday locked operating condition sequence. By mapping the day-ahead pumped storage power generation, pumping power, and net power to an intraday time scale, a day-ahead reference power sequence is obtained.

10. A multi-type energy storage dispatch system based on net load trend-deviation decoupling, characterized in that: The system includes a day-ahead net load acquisition module, a net load trend-deviation decoupling module, a day-ahead low-frequency benchmark scheduling module, a day-ahead-intraday mapping module, and an intraday electrochemical energy storage-pumped storage coordinated scheduling module. The day-ahead system net load acquisition module is used to acquire the day-ahead system net load prediction sequence under various typical source load operation scenarios. The net load trend-deviation decoupling module is used to perform low-frequency trend decomposition on the day-ahead system net load forecast sequence to obtain low-frequency trend components. The daytime low-frequency reference scheduling module is used to solve the daytime thermal power-pumped storage low-frequency reference scheduling model with the low-frequency trend component as the daytime scheduling target, and obtain a low-frequency reference scheduling scheme including the daytime operating conditions and daytime power of pumped storage. The day-to-day mapping module is used to map the day-to-day operating conditions of pumped storage to an intraday time scale to obtain an intraday locked operating condition sequence. Mapping the daytime power of pumped storage to an intraday timescale yields the daytime baseline power sequence. The intraday electrochemical energy storage-pumped hydro storage coordinated scheduling module is used to calculate the intraday electrochemical energy storage-pumped hydro storage coordinated optimization model by using the deviation of the intraday real-time net load of pumped hydro storage relative to the day-ahead reference power as the compensation object. This results in intraday scheduling schemes for electrochemical energy storage and intraday fine-tuning schemes for pumped hydro storage under various typical source-load operation scenarios. In this intraday optimization model, the actual intraday operation of pumped hydro storage is based on the day-ahead locked operating condition sequence, coordinating rapid compensation of electrochemical energy storage and limited fine-tuning of pumped hydro storage.

Images