Multi-type energy storage multi-time scale cooperative scheduling method and system

Abstract

The invention discloses a multi-type energy storage multi-time scale cooperative scheduling method and system, and the method comprises the steps: obtaining historical data and new energy and load prediction data of multiple days in the future, constructing a multi-day rolling optimization model, carrying out the long-period optimization calculation, obtaining a daily end charge state boundary value of pumped storage, and carrying out the calculation of the daily end charge state boundary value; the boundary condition is used as a day-ahead optimization boundary condition; combining the day-ahead new energy power prediction curve and the net load fluctuation characteristics to construct a multi-type energy storage cooperative scheduling objective function of pumped storage and electrochemical energy storage adapting to different time zone characteristics; establishing a two-stage robust optimization model, and solving to obtain an energy storage charge state robust feasible region of the multi-type energy storage in the scheduling day; updating the uncertainty set by using ultra-short-term prediction data, solving a real-time correction output instruction, and realizing cooperative control of multiple types of energy storage in the robust feasible region; according to the invention, long-time collaboration with multiple types of energy storage resources is realized, and economy and flexibility of system operation are effectively considered.

Description

Technical Field

[0001] This invention relates to the field of energy storage scheduling, and particularly to a method and system for coordinated scheduling of multiple types of energy storage across multiple time scales. Background Technology

[0002] With the continuous advancement of new power system construction and the increasing penetration rate of new energy sources such as wind power and photovoltaics, the randomness and volatility of their output pose a continuous pressure on the power balance and economic operation of the power system. Energy storage systems, as a key means to improve system flexibility and reliability, have been widely introduced into the power grid. These systems encompass various forms, including pumped hydro storage and electrochemical energy storage, with significant differences in technical characteristics such as capacity, response speed, and duration. However, existing energy storage dispatching methods still face the problem of insufficient time-scale matching: traditional day-ahead dispatching models typically only focus on operational optimization for the next 24 hours, resulting in a short window that makes it difficult to coordinate the energy time-shifting demand of long-term storage over multiple days or even weeks. This can easily lead to insufficient energy reserves for long-term storage before consecutive rainy or windless days. Meanwhile, simple multi-day coupling models often have excessively high computational complexity, making it difficult to balance long-term foresight with short-term operational details, and often neglecting the temporal complementarity characteristics of different types of energy storage.

[0003] Furthermore, existing methods have significant limitations in preprocessing and operational strategies. Most studies employ simple smoothing or fixed wind and solar curtailment strategies to handle renewable energy fluctuations, making it difficult to achieve a dynamic balance between high-proportion renewable energy consumption and system economics. Fixed-time-window-based scheduling models cannot coordinate the cross-cycle regulation of long-term energy storage with the rapid response needs of short-term energy storage. Traditional day-ahead-intraday rolling scheduling relies on deterministic point prediction modeling, failing to fully consider the impact of prediction errors on the system's feasible region, resulting in insufficient reserve of energy storage regulation margin. In extreme scenarios such as continuous windless periods or prolonged periods of overcast and rainy weather, the system may face operational risks due to insufficient medium- to long-term regulation capacity or limited short-term regulation space. Summary of the Invention

[0004] The purpose of this invention is to solve the problem of the difficulty in accurately coordinating the operation of multiple types of energy storage across time scales under the high proportion of new energy access; to provide a method and system for coordinating the scheduling of multiple types of energy storage that can integrate multiple time scales, adapt to dynamic scenarios and have strong robustness; to realize the coordination of long-term and multi-type energy storage resources, effectively taking into account the economy and flexibility of system operation.

[0005] To achieve the above objectives, this invention provides a method for coordinated scheduling of multiple types of energy storage across multiple time scales; comprising the following steps:

[0006] Step S1: Obtain historical operating data of the power system and forecast data of new energy sources and loads for the next few days, construct a multi-day rolling optimization model for long-term optimization calculation, and obtain the end-of-day state of charge boundary value of pumped storage as the boundary condition for day-ahead optimization.

[0007] Step S2: Based on the end-of-day state-of-charge boundary value of pumped storage obtained in Step S1, and combined with the day-ahead renewable energy power prediction curve and net load fluctuation characteristics, the scheduling day is divided into renewable energy consumption time zone, auxiliary peak shaving time zone and regular time zone using dynamic programming method. An objective function for the coordinated scheduling of pumped storage and electrochemical energy storage is constructed to adapt to the characteristics of different time zones.

[0008] Step S3: Based on the objective function for coordinated scheduling of multiple types of energy storage adapted to the characteristics of different time zones constructed in Step S2, and considering the uncertainty of new energy prediction error, a two-stage robust optimization model is established to solve for the robust feasible domain of energy storage state of charge of multiple types of energy storage within the scheduling day.

[0009] Step S4: Based on the robust feasible region of the state of charge of the multi-type energy storage obtained in step S3, during the intraday real-time operation phase, the uncertainty set is updated using ultra-short-term forecast data, and the output command is solved in real time to realize the coordinated control of the multi-type energy storage within the robust feasible region.

[0010] Preferably, step S1, by introducing a multi-day rolling optimization mechanism, coordinates the power management of thermal power unit start-up and shutdown and pumped storage over a longer time dimension. The multi-day rolling mechanism adopts an optimization framework that decouples the prediction time domain from the execution time domain; including: setting the prediction time domain of the multi-day rolling optimization as... , Take 3 to 7 days, that is The execution time is 72–168 hours. , Take 24 hours to predict the time domain. Construct an objective function with the goal of minimizing the total system operating cost:

[0011]

[0012] In the formula, The total operating cost of the system, For time indexing, For thermal power units, It is a pumped storage power station. and It is a collection of thermal power units and pumped storage power stations. For thermal power units The cost of coal, For thermal power units Startup costs, For thermal power units The cost of no-load operation Pumped storage power station The operation and maintenance costs For time period The amount of wind and solar power that has been curtailed. For time period The shear load, and These are their corresponding penalty coefficients. As the terminal deviation penalty factor, For the first The decision-making process at the end of the day accumulates energy. This is the expected terminal water level pre-set based on long-term statistical patterns;

[0013] The constraints that the multi-day rolling optimization needs to meet include:

[0014] System power balance constraints:

[0015]

[0016] In the formula, and Pumped storage power stations The discharge power and charging power, The day-ahead forecast values ​​that contribute to the system's wind and solar power. This is the day-ahead forecast of the total system load;

[0017] Thermal power unit operating constraints:

[0018] Upper and lower limits of thermal power unit output constraints:

[0019]

[0020] In the formula, and thermal power units The lower and upper limits of technical output, For unit start-up and shutdown status variables, when When taking 1, thermal power units exist It is always in a shutdown state, when When the value is 0, the thermal power unit exist It is always powered on;

[0021] Thermal power unit ramping constraints:

[0022]

[0023] In the formula, and thermal power units The limits for uphill and downhill climbing speeds;

[0024] Minimum start-up and shutdown time constraints for thermal power units:

[0025]

[0026] In the formula, and thermal power units During the period The duration of continuous power-on time and the duration of continuous power-off time. and thermal power units Minimum continuous power-on time and minimum continuous downtime;

[0027] Energy coupling constraints of pumped storage across time periods:

[0028] To characterize the dynamic change of water storage capacity in the upper reservoir of a pumped storage power station over continuous time, and to prevent power depletion due to optimization window truncation, the following energy balance equation and reservoir capacity constraints are established:

[0029]

[0030] In the formula, Pumped storage power station exist The energy storage of the upper reservoir during the period, for Energy storage during a certain period; They are respectively Pumping power and power generation during the same period; These are pumping efficiency and power generation efficiency, respectively. , These are the minimum and maximum allowable storage capacity of the upper reservoir, respectively. The duration is the length of the time period. For the first The stored energy at the end of each operating day. The preset target energy threshold at the end of the day; this constraint mandates that the system retain sufficient adjustment capacity at the end of each day to support continuous operation for several more days to come;

[0031] Mutual exclusion constraints for pumped storage operation status:

[0032]

[0033] In the formula: , The units exist The pumping and power generation status for a given time period is represented by 1 for operation and 0 for shutdown. and These are the minimum and maximum power limits for the unit under the corresponding operating conditions.

[0034] Preferably, in step S2, a day-ahead coordinated scheduling model based on dynamic time zone division is constructed. This model is achieved by dynamically dividing the scheduling time zones and constructing an adaptive objective function. The specific process is as follows:

[0035] Given the limitations of traditional static peak-valley time zone division in adapting to the source-load timing characteristics under high-proportion renewable energy access, a dynamic time zone division mechanism based on net load morphology characteristics is introduced.

[0036] Based on the current renewable energy and load forecast data, the system equivalent net load sequence is constructed. Then, considering the amplitude distribution and rate of change characteristics of the net load, a multi-dimensional time zone determination criterion is established, and the 24 time periods in the scheduling cycle are adaptively deconstructed into three types of characteristic time zones: renewable energy consumption, auxiliary peak shaving, and conventional time zones.

[0037] The mathematical criterion for the dynamic time zone division is expressed as follows:

[0038]

[0039] In the formula: for Forecasted system net load for the time period The rate of change of net load over adjacent time periods. and These are the net load thresholds for determining the renewable energy consumption time zone and the auxiliary peak-shaving time zone, respectively. To determine the ramp rate threshold for drastic load fluctuations, , , These represent the time zones for renewable energy consumption, auxiliary peak shaving, and regular time zones, respectively.

[0040] To address the characteristics of different time zones, electrochemical energy storage is incorporated into the optimization framework. Leveraging its fast response speed in conjunction with pumped hydro storage, a day-ahead coordinated scheduling objective function with adaptive weighting coefficients is constructed. These adaptive weighting coefficients are dynamically adjusted based on the characteristic time zone type of the current period. In renewable energy consumption time zones, the proportion of the penalty weight for wind and solar curtailment is increased; in auxiliary peak-shaving time zones, the proportion of the penalty weight for safety margin is increased. The expression is as follows:

[0041]

[0042] In the formula, The total cost of the coordinated scheduling system as of now. , , The adaptive weighting coefficients for the renewable energy consumption time zone, auxiliary peak-shaving time zone, and conventional balance time zone during time period t are determined by the characteristic time zone to which this period belongs. , The system's cost of curtailing wind and solar power. This is a system safety margin penalty item used to penalize the risk of insufficient spinning reserve. This represents the total cost of system collaborative operation up to date. For thermal power units The cost of coal, Pumped storage power station Start-up and shutdown costs For electrochemical energy storage The unit charge / discharge loss coefficient calculated over the entire life cycle. and Electrochemical energy storage The charging and discharging power, It is an electrochemical energy storage system;

[0043] Regarding the value of the weight, when t is in the new energy consumption time zone, set... This significantly increases the penalty weight for wind and solar curtailment, forcing the system to prioritize the use of energy storage for charging and consumption. When t is in the auxiliary peak-shaving time zone, the following settings are made: Increasing the safety margin penalty weight forces energy storage to reserve discharge capacity at its peak. When t is in the normal equilibrium time zone, the following settings are applied. As the dominant weight, , The base value is taken, and the system then reverts to the economy-first mode. This mapping mechanism eliminates the blindness of traditional manual weight assignment and enables the scheduling strategy to switch automatically according to the net load characteristics.

[0044] Regarding constraints, in addition to conventional system balancing and unit operation constraints, differentiated operation constraints for various types of energy storage were constructed, and the inter-day energy boundary passed in step S1 was introduced; the mathematical expressions for the relevant core constraints are as follows:

[0045] Pumped storage constraints:

[0046]

[0047] In the formula, The target electricity consumption at the end of the day, derived from the multi-day rolling optimization in step S1, For the allowed boundary relaxation range;

[0048] Electrochemical energy storage operation constraints:

[0049]

[0050] In the formula, For electrochemical energy storage Percentage of state of charge at time t For its rated capacity, and These are their charging and discharging efficiencies, respectively. and These are the safe upper and lower limits for its state of charge. and This refers to the battery's charge / discharge state. Take 1 as discharge. Set 0 to indicate charging;

[0051] Thermal power output constraints based on stationary unit combinations:

[0052] The real-time output of thermal power units is only allowed to be adjusted within a fixed physical range.

[0053]

[0054] In the formula, Given the known parameters passed in step S1, this constraint ensures that the day-ahead fine-grained scheduling strictly follows the multi-day rolling optimization mechanism;

[0055] System power balance constraints:

[0056] .

[0057] Preferably, in step S3, considering the uncertainty of new energy prediction errors, a two-stage robust optimization model is established. This model constructs an uncertainty set of new energy output and seeks the optimal response strategy under the worst-case scenario using a "pre-scheduling-re-scheduling" framework. The specific process is as follows:

[0058] To quantify the uncertainty of new energy output, a box-shaped uncertainty set is constructed based on historical forecast error data to describe the potential deviation of actual output from the range of the current day forecast value.

[0059] The mathematical expression for the uncertainty set of new energy output is as follows:

[0060]

[0061] In the formula, and Wind power and solar power during different time periods The real-time power output value and the day-ahead forecast value, For time period The maximum permissible range of prediction error fluctuation, i.e., the radius of uncertainty. and These are auxiliary variables used to characterize positive and negative prediction biases; This is the uncertainty budget parameter, used to control the extreme degree of uncertainty scenarios considered. The larger the value, the more conservative the model.

[0062] Based on the aforementioned set of uncertainties, a two-stage robust optimization objective function, Min-Max-Min, is constructed; this function aims to minimize the total system cost under the worst-case uncertainty scenario.

[0063] The mathematical expression for the two-stage robust optimization objective function is as follows:

[0064]

[0065] In the formula, The first phase of day-ahead pre-dispatch decision variables includes the planned output of thermal power units, the energy storage infrastructure charging and discharging plan, and the boundary of its feasible region. Its feasible region is... , The basic operating cost for the first stage is calculated based on the model in step S2; Belongs to the set of uncertainty Any new energy power output scenario, The second-stage real-time rescheduling decision variables include unit output adjustment, energy storage power correction, and load shedding, etc., and its feasible region... Subject to the first phase of decision With uncertain scenarios Common constraints This refers to the penalty costs incurred during the rescheduling phase due to deviation adjustments, such as load shedding costs and emergency energy curtailment costs.

[0066] Unlike fixed trajectories, this method models the state of charge of energy storage as a dynamic band-shaped interval, ensuring that the real-time SOC can be maintained within the interval under any permissible uncertainties, thereby reserving sufficient adjustment capability;

[0067] Constraints of the two-stage robust optimization model:

[0068] Robust operation and boundary connection constraints of pumped storage:

[0069]

[0070] In the formula, to ensure the convergence of optimization in extreme scenarios, a minimal relaxation variable is introduced into the boundary constraints. This is a permissible minimal robust relaxation;

[0071] Electrochemical energy storage robust SOC constraint:

[0072]

[0073] Thermal power output constraints based on stationary unit combinations:

[0074] The real-time output of thermal power units is only allowed to be adjusted within a fixed physical range.

[0075]

[0076] By solving the above model using a column and constraint generation algorithm, robust feasible regions of various energy storage SOCs capable of withstanding the most severe renewable energy fluctuation scenarios can be obtained, along with the corresponding baseline output plan for thermal power generating units; as long as The energy storage state at any given time lies within this interval, for the set of uncertainties. Any perturbation in All systems have feasible control strategies to enable... The state at time 1 remains within the interval of the next time 1, and satisfies the end-of-day boundary constraints.

[0077] Preferably, in step S4, based on the robust feasible region of the energy storage state of charge obtained in step S3, during the intraday real-time operation phase, the uncertainty set is updated using ultra-short-term forecast data, and the specific process for solving the real-time corrected output command is as follows:

[0078] During the intraday real-time operation phase, a Model Predictive Control (MPC) strategy is employed, with a rolling execution time window set as follows: The time resolution is ;

[0079] Based on the latest ultra-short-term renewable energy power forecast data Update the uncertain input state of the system; at this point, the core objective of optimization is to track the real-time power imbalance of the system with the least adjustment cost while satisfying the safety boundary determined a day before.

[0080] The objective function for intraday real-time rolling correction is constructed, and its mathematical expression is as follows:

[0081]

[0082] In the formula, For the current moment, For the predicted time within the scrolling window, This represents the output adjustment of thermal power units relative to the day-ahead baseline plan. To adjust the cost coefficient, Representing pumped hydro storage and electrochemical energy storage, This refers to the power deviation of various types of energy storage relative to the day-ahead pre-dispatch command. This is the corresponding deviation penalty coefficient. For real-time power imbalance residuals, This is the real-time power imbalance penalty factor;

[0083] Regarding constraints, a robust feasible region constraint for the energy storage charge state is introduced; in order to ensure the robustness of the day-ahead scheduling strategy, the energy storage energy state in real-time operation is forcibly constrained within the envelope range calculated in step S3.

[0084] Real-time collaborative operation constraints:

[0085]

[0086] In the formula, and These represent the actual energy states of pumped hydro storage and electrochemical energy storage during real-time operation, respectively. and These are the lower and upper boundaries of the robust feasible region of the energy storage charged state obtained in step S3, respectively. The real-time ramp rate limitation for electrochemical energy storage The real-time ramp rate limit for pumped storage;

[0087] Terminal state guidance constraints:

[0088]

[0089]

[0090] In the formula, The rolling prediction time domain length for model predictive control; This is the end time of the prediction window; Penalty cost for deviation of terminal state;

[0091] By solving the above model, the system generates the current time. The final power command is issued and distributed to each energy storage power station for execution. This strategy utilizes the high ramp rate of electrochemical energy storage to prioritize smoothing out high-frequency and small-amplitude power fluctuations, and utilizes the large capacity characteristics of pumped hydro storage to bear continuous energy imbalances within the range allowed by the robust feasible domain, thereby achieving closed-loop collaborative control of multiple types of energy storage on the entire time scale.

[0092] This invention also provides a multi-type energy storage multi-timescale collaborative scheduling system, including a method for implementing the above-described multi-type energy storage multi-timescale collaborative scheduling method, and further comprising:

[0093] The multi-day rolling optimization module is responsible for acquiring historical operating data of the power system and forecasting data of new energy sources and loads for the next few days; it performs multi-day rolling optimization calculations that decouple the prediction time domain from the execution time domain, comprehensively considering the coal cost of thermal power units, start-up and shutdown costs, and operation and maintenance costs of pumped storage, and outputs the end-of-day state-of-charge boundary value of pumped storage under the conditions of satisfying system power balance constraints, thermal power unit ramp-up and start-up and shutdown time constraints, and pumped storage cross-time period energy coupling constraints;

[0094] The day-ahead dynamic zone scheduling module, based on the day-ahead renewable energy power prediction curve and net load fluctuation characteristics output by the multi-day rolling optimization module, calculates the system net load sequence and its rate of change characteristics. With the goal of adapting to the source-load time sequence characteristics, the scheduling day is adaptively divided into renewable energy consumption time zone, auxiliary peak shaving time zone and regular time zone according to the preset net load amplitude and ramp rate threshold. A multi-type energy storage collaborative scheduling objective function containing adaptive weight coefficients is constructed.

[0095] The two-stage robust optimization module is based on the collaborative scheduling objective function constructed by the day-ahead dynamic partition scheduling module, and constructs a set of uncertainties in new energy output by combining historical prediction error data; a two-stage robust optimization model of Min-Max-Min is established and solved by column and constraint generation algorithm. Under the condition of satisfying the system operation safety constraints in the worst scenario, it outputs the robust feasible domain of energy storage state of charge and the benchmark output plan of thermal power units in the scheduling day for multiple types of energy storage.

[0096] The intraday real-time rolling correction module, based on the robust feasible region of the energy storage state of charge output by the two-stage robust optimization module, updates the system uncertainty state using ultra-short-term renewable energy power prediction data; adopts a model predictive control strategy, with the goal of minimizing real-time power imbalance and adjustment costs, and solves and outputs the real-time corrected output command of each energy storage power station under the condition that the real-time energy storage state does not exceed the robust feasible region.

[0097] The present invention also provides a storage medium storing a program, which, when executed by a processor, implements the above-described method for coordinated scheduling of multiple types of energy storage across multiple time scales.

[0098] The present invention also provides a computing device, including a processor and a memory for storing processor-executable programs, wherein when the processor executes the program stored in the memory, it implements the above-described multi-type energy storage multi-timescale collaborative scheduling method.

[0099] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0100] 1. This invention breaks through the limitations of a single time scale window, maximizing the cross-day energy value of long-term energy storage. Existing technologies are usually limited to a 24-hour day-ahead optimization window, which makes long-term energy storage such as pumped hydro storage prone to "short-sighted decision-making" and unable to balance the power over a longer period. This invention, through the multi-day rolling optimization model constructed in step S1, extends the decision window to 3-7 days, pre-locking the end-of-day charge state boundary of long-term energy storage before day-ahead scheduling. This mechanism effectively overcomes the truncation effect of the time window, ensuring that long-term energy storage still retains sufficient cross-day regulation power under extreme weather conditions such as continuous windless or long-term rainy weather, significantly improving the system's long-term supply guarantee capability.

[0101] 2. This invention breaks the rigid pattern of fixed-time scheduling, significantly improving the system's adaptability to the temporal characteristics of source and load. Addressing the issue of rapid switching between "consumption" and "supply guarantee" in system operation due to high-frequency fluctuations in renewable energy output, this invention, through the dynamic time zone division mechanism proposed in step S2, adaptively divides the scheduling day into renewable energy consumption time zones, auxiliary peak-shaving time zones, and regular time zones using the net load change rate characteristics. Combined with an adaptive weighted objective function, this allows various types of energy storage to switch operating strategies "according to the time," such as prioritizing charging in the consumption time zone and prioritizing discharging in the peak-shaving time zone, thereby ensuring a high proportion of renewable energy consumption while balancing the economy and safety of system operation.

[0102] 3. This invention constructs a "pre-scheduling + re-scheduling" architecture based on the robust feasible region of SOC, fundamentally enhancing the anti-disturbance capability of the scheduling scheme. Addressing the problem of insufficient adjustment margin in traditional deterministic scheduling when facing large deviations in real-time renewable energy forecasts, this invention, through the two-stage robust optimization model established in step S3, does not directly solve for a single fixed output curve, but instead solves for the robust feasible region of energy storage charge state covering all adverse scenarios. This feasible region reserves a definite safety adjustment boundary for intraday real-time operation, effectively solving the risk of energy storage exceeding limits due to forecast errors, and ensuring the physical consistency between day-ahead planning and real-time operation.

[0103] 4. This invention achieves synergy among multiple types of energy storage resources, effectively balancing the economy and flexibility of system operation. Through the real-time rolling correction strategy in step S4, this invention achieves complementary advantages between pumped hydro storage and electrochemical energy storage within a unified optimization framework. Utilizing multi-day rolling and robust feasible region constraints, it leverages the advantages of pumped hydro storage—large capacity and long duration—to handle long-term energy imbalances; while utilizing real-time MPC control, it leverages the advantages of electrochemical energy storage—fast response and strong ramp-up capability—to smooth high-frequency power fluctuations. This hierarchical synergistic mechanism not only improves the overall regulation quality of the system but also effectively reduces frequent deep charge-discharge cycles of electrochemical energy storage, helping to extend equipment lifespan. Attached Figure Description

[0104] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0105] Figure 1 This is a flowchart of the steps of a multi-type energy storage multi-timescale collaborative scheduling method provided by the present invention;

[0106] Figure 2 This is a schematic diagram of the dynamic time zone division principle provided by the present invention;

[0107] Figure 3 This is a diagram of the multi-time-scale collaborative scheduling framework provided by the present invention;

[0108] Figure 4 This is a structural block diagram of a multi-type energy storage multi-timescale collaborative scheduling system provided by the present invention. Detailed implementation

[0109] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are one embodiment of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0110] Example 1

[0111] Please refer to Figure 1 This embodiment discloses a multi-type energy storage multi-timescale collaborative scheduling method, the specific details of which are as follows:

[0112] Step S1: Obtain historical operating data of the power system and forecast data of new energy sources and loads for the next few days, construct a multi-day rolling optimization model for long-term optimization calculations, and obtain the end-of-day state-of-charge boundary values ​​of pumped storage as the boundary conditions for day-ahead optimization; details are as follows:

[0113] Step S1, by introducing a multi-day rolling optimization mechanism, coordinates the power management of thermal power unit start-up and shutdown and pumped storage over a longer time dimension. This multi-day rolling mechanism employs an optimization framework that decouples the prediction time domain from the execution time domain, specifically including:

[0114] The prediction time domain for multi-day rolling optimization is set as follows: , Take 3 to 7 days, that is The execution time is 72–168 hours. , Take 24 hours to predict the time domain. Construct an objective function with the goal of minimizing the total system operating cost:

[0115]

[0116] In the formula, The total operating cost of the system, For time indexing, For thermal power units, It is a pumped storage power station. and It is a collection of thermal power units and pumped storage power stations. For thermal power units The cost of coal, For thermal power units Startup costs, For thermal power units The cost of no-load operation Pumped storage power station The operation and maintenance costs For time period The amount of wind and solar power that has been curtailed. For time period The shear load, and These are their corresponding penalty coefficients. As the terminal deviation penalty factor, For the first The decision-making process at the end of the day accumulates energy. This is the expected terminal water level pre-set based on long-term statistical patterns.

[0117] The constraints that the multi-day rolling optimization needs to meet include:

[0118] System power balance constraints:

[0119]

[0120] In the formula, and Pumped storage power stations The discharge power and charging power, The day-ahead forecast values ​​that contribute to the system's wind and solar power. This is the day-ahead forecast of the total system load.

[0121] Thermal power unit operating constraints:

[0122] Upper and lower limits of thermal power unit output constraints:

[0123]

[0124] In the formula, and thermal power units The lower and upper limits of technical output, For unit start-up and shutdown status variables, when When taking 1, thermal power units exist It is always in a shutdown state, when When the value is 0, the thermal power unit exist It is always powered on;

[0125] Thermal power unit ramping constraints:

[0126]

[0127] In the formula, and thermal power units The limits for uphill and downhill climbing speeds;

[0128] Minimum start-up and shutdown time constraints for thermal power units:

[0129]

[0130]

[0131] In the formula, and thermal power units During the period The duration of continuous power-on time and the duration of continuous power-off time. and thermal power units Minimum continuous power-on time and minimum continuous downtime;

[0132] Energy coupling constraints of pumped storage across time periods:

[0133] To characterize the dynamic change of water storage capacity in the upper reservoir of a pumped storage power station over continuous time, and to prevent power depletion due to optimization window truncation, the following energy balance equation and reservoir capacity constraints are established:

[0134]

[0135] In the formula, Pumped storage power station exist The energy storage of the upper reservoir during the period, for Energy storage during a certain period; They are respectively Pumping power and power generation during the same period; These are pumping efficiency and power generation efficiency, respectively. , These are the minimum and maximum allowable storage capacity of the upper reservoir, respectively. The duration is the length of the time period. For the first The stored energy at the end of each operating day. The preset target energy threshold at the end of the day; this constraint mandates that the system retain sufficient adjustment capacity at the end of each day to support continuous operation for several more days to come;

[0136] Mutual exclusion constraints for pumped storage operation status:

[0137]

[0138] In the formula: , The units exist The pumping and power generation status for a given time period is represented by 1 for operation and 0 for shutdown. and These are the minimum and maximum power limits for the unit under the corresponding operating conditions.

[0139] Step S2: Based on the end-of-day state-of-charge boundary value of pumped hydro storage obtained in Step S1, and combined with the day-ahead renewable energy power forecast curve and net load fluctuation characteristics, a dynamic programming method is used to divide the scheduling day into renewable energy consumption time zones, auxiliary peak-shaving time zones, and conventional time zones. An objective function for the coordinated scheduling of multiple types of energy storage, including pumped hydro storage and electrochemical energy storage, is constructed to adapt to the characteristics of different time zones; the details are as follows:

[0140] Step S2 involves constructing a day-ahead collaborative scheduling model based on dynamic time zone division. This model is achieved by dynamically dividing scheduling time zones and constructing an adaptive objective function. The specific process is as follows:

[0141] Given the limitations of traditional static peak-valley time zone division in adapting to the source-load timing characteristics under high-proportion renewable energy access, a dynamic time zone division mechanism based on net load morphology characteristics is introduced.

[0142] Based on the current renewable energy and load forecast data, the system equivalent net load sequence is constructed. Then, considering the amplitude distribution and rate of change characteristics of the net load, a multi-dimensional time zone determination criterion is established, and the 24 time periods in the scheduling cycle are adaptively deconstructed into three types of characteristic time zones: renewable energy consumption, auxiliary peak shaving, and conventional time zones.

[0143] The mathematical criterion for the dynamic time zone division is expressed as follows:

[0144]

[0145] In the formula: for Forecasted system net load for the time period The rate of change of net load over adjacent time periods. and These are the net load thresholds for determining the renewable energy consumption time zone and the auxiliary peak-shaving time zone, respectively. To determine the ramp rate threshold for drastic load fluctuations, , , These represent the time zones for renewable energy consumption, auxiliary peak shaving, and regular time zones, respectively.

[0146] To address the characteristics of different time zones, electrochemical energy storage is incorporated into the optimization framework. Leveraging its fast response speed in conjunction with pumped hydro storage, a day-ahead coordinated scheduling objective function with adaptive weighting coefficients is constructed. These adaptive weighting coefficients are dynamically adjusted based on the characteristic time zone type of the current period. In renewable energy consumption time zones, the proportion of the penalty weight for wind and solar curtailment is increased; in auxiliary peak-shaving time zones, the proportion of the penalty weight for safety margin is increased. The expression is as follows:

[0147]

[0148] In the formula, The total cost of the coordinated scheduling system as of now. , , The adaptive weighting coefficients for the renewable energy consumption time zone, auxiliary peak-shaving time zone, and conventional balance time zone during time period t are determined by the characteristic time zone to which this period belongs. , The system's cost of curtailing wind and solar power. This is a system safety margin penalty item used to penalize the risk of insufficient spinning reserve. This represents the total cost of system collaborative operation up to date. For thermal power units The cost of coal, Pumped storage power station Start-up and shutdown costs For electrochemical energy storage The unit charge / discharge loss coefficient calculated over the entire life cycle. and Electrochemical energy storage The charging and discharging power, It is an electrochemical energy storage system;

[0149] Regarding the value of the weight, when t is in the new energy consumption time zone, set... This significantly increases the penalty weight for wind and solar curtailment, forcing the system to prioritize the use of energy storage for charging and consumption. When t is in the auxiliary peak-shaving time zone, the following settings are made: Increasing the safety margin penalty weight forces energy storage to reserve discharge capacity at its peak. When t is in the normal equilibrium time zone, the following settings are applied. As the dominant weight, , The base value is taken, and the system then reverts to the economy-first mode. This mapping mechanism eliminates the blindness of traditional manual weight assignment and enables the scheduling strategy to switch automatically according to the net load characteristics.

[0150] Regarding constraints, in addition to conventional system balancing and unit operation constraints, differentiated operation constraints for various types of energy storage were constructed, and the inter-day energy boundary passed in step S1 was introduced; the mathematical expressions for the relevant core constraints are as follows:

[0151] Pumped storage constraints:

[0152]

[0153] In the formula, The target electricity consumption at the end of the day, derived from the multi-day rolling optimization in step S1, For the allowed boundary relaxation range;

[0154] Electrochemical energy storage operation constraints:

[0155]

[0156] In the formula, For electrochemical energy storage Percentage of state of charge at time t For its rated capacity, and These are their charging and discharging efficiencies, respectively. and These are the safe upper and lower limits for its state of charge. and This refers to the battery's charge / discharge state. Take 1 as discharge. Set 0 to indicate charging;

[0157] Thermal power output constraints based on stationary unit combinations:

[0158] The real-time output of thermal power units is only allowed to be adjusted within a fixed physical range.

[0159]

[0160] In the formula, Given the known parameters passed in step S1, this constraint ensures that the day-ahead fine-grained scheduling strictly follows the multi-day rolling optimization mechanism;

[0161] System power balance constraints:

[0162] .

[0163] Step S3: Based on the objective function for coordinated scheduling of multiple types of energy storage adapted to the characteristics of different time zones constructed in Step S2, and considering the uncertainty of new energy prediction error, a two-stage robust optimization model is established to solve for the robust feasible domain of energy storage state of charge of multiple types of energy storage within the scheduling day.

[0164] In step S3, considering the uncertainty of new energy prediction errors, a two-stage robust optimization model is established. This model constructs an uncertainty set of new energy output and seeks the optimal response strategy under the worst-case scenario using a "pre-scheduling-rescheduling" framework. The specific process is as follows:

[0165] To quantify the uncertainty of new energy output, a box-shaped uncertainty set is constructed based on historical forecast error data to describe the potential deviation of actual output from the range of the current day forecast value.

[0166] The mathematical expression for the uncertainty set of new energy output is as follows:

[0167]

[0168] In the formula, and Wind power and solar power during different time periods The real-time power output value and the day-ahead forecast value, For time period The maximum permissible range of prediction error fluctuation, i.e., the radius of uncertainty. and These are auxiliary variables used to characterize positive and negative prediction biases; This is the uncertainty budget parameter, used to control the extreme degree of uncertainty scenarios considered. The larger the value, the more conservative the model.

[0169] Based on the aforementioned set of uncertainties, a two-stage robust optimization objective function, Min-Max-Min, is constructed; this function aims to minimize the total system cost under the worst-case uncertainty scenario.

[0170] The mathematical expression for the two-stage robust optimization objective function is as follows:

[0171]

[0172] In the formula, The first phase of day-ahead pre-dispatch decision variables includes the planned output of thermal power units, the energy storage infrastructure charging and discharging plan, and the boundary of its feasible region. Its feasible region is... , The basic operating cost for the first stage is calculated based on the model in step S2; Belongs to the set of uncertainty Any new energy power output scenario, The second-stage real-time rescheduling decision variables include unit output adjustment, energy storage power correction, and load shedding, etc., and its feasible region... Subject to the first phase of decision With uncertain scenarios Common constraints This refers to the penalty costs incurred during the rescheduling phase due to deviation adjustments, such as load shedding costs and emergency energy curtailment costs.

[0173] Unlike fixed trajectories, this method models the state of charge of energy storage as a dynamic band-shaped interval, ensuring that the real-time SOC can be maintained within the interval under any permissible uncertainties, thereby reserving sufficient adjustment capability;

[0174] Constraints of the two-stage robust optimization model:

[0175] Robust operation and boundary connection constraints of pumped storage:

[0176]

[0177] In the formula, to ensure the convergence of optimization in extreme scenarios, a minimal relaxation variable is introduced into the boundary constraints. This is a permissible minimal robust relaxation;

[0178] Electrochemical energy storage robust SOC constraint:

[0179]

[0180]

[0181] Thermal power output constraints based on stationary unit combinations:

[0182] The real-time output of thermal power units is only allowed to be adjusted within a fixed physical range.

[0183]

[0184] By solving the above model using a column and constraint generation algorithm, robust feasible regions of various energy storage SOCs capable of withstanding the most severe renewable energy fluctuation scenarios can be obtained, along with the corresponding baseline output plan for thermal power generating units; as long as The energy storage state at any given time lies within this interval, for the set of uncertainties. Any perturbation in All systems have feasible control strategies to enable... The state at time 1 remains within the interval of the next time 1, and satisfies the end-of-day boundary constraints.

[0185] Step S4: Based on the robust feasible region of the state of charge of the multi-type energy storage obtained in step S3, during the intraday real-time operation phase, the uncertainty set is updated using ultra-short-term forecast data, and the output command is solved in real time to realize the coordinated control of the multi-type energy storage within the robust feasible region.

[0186] In step S4, based on the robust feasible region of the energy storage state of charge obtained in step S3, during the intraday real-time operation phase, the uncertainty set is updated using ultra-short-term forecast data, and the specific process for solving the real-time corrected output command is as follows:

[0187] This embodiment employs Model Predictive Control (MPC) during the intraday real-time operation phase, setting a rolling execution time window as follows: The time resolution is ;

[0188] Based on the latest ultra-short-term renewable energy power forecast data Update the uncertain input state of the system; at this point, the core objective of optimization is to track the real-time power imbalance of the system with the least adjustment cost while satisfying the safety boundary determined a day before.

[0189] The objective function for intraday real-time rolling correction is constructed, and its mathematical expression is as follows:

[0190]

[0191] In the formula, For the current moment, For the predicted time within the scrolling window, This represents the output adjustment of thermal power units relative to the day-ahead baseline plan. To adjust the cost coefficient, Representing pumped hydro storage and electrochemical energy storage, This refers to the power deviation of various types of energy storage relative to the day-ahead pre-dispatch command. This is the corresponding deviation penalty coefficient. For real-time power imbalance residuals, This is the real-time power imbalance penalty factor;

[0192] Regarding constraints, a robust feasible region constraint for the energy storage charge state is introduced; in order to ensure the robustness of the day-ahead scheduling strategy, the energy storage energy state in real-time operation is forcibly constrained within the envelope range calculated in step S3.

[0193] Real-time collaborative operation constraints:

[0194]

[0195] In the formula, and These represent the actual energy states of pumped hydro storage and electrochemical energy storage during real-time operation, respectively. and These are the lower and upper boundaries of the robust feasible region of the energy storage charged state obtained in step S3, respectively. The real-time ramp rate limitation for electrochemical energy storage The real-time ramp rate limit for pumped storage;

[0196] Terminal state guidance constraints:

[0197]

[0198]

[0199] In the formula, The rolling prediction time domain length for model predictive control; This is the end time of the prediction window; Penalty cost for deviation of terminal state;

[0200] By solving the above model, the system generates the current time. The final power command is issued and distributed to each energy storage power station for execution. This strategy utilizes the high ramp rate of electrochemical energy storage to prioritize smoothing out high-frequency and small-amplitude power fluctuations, and utilizes the large capacity characteristics of pumped hydro storage to bear continuous energy imbalances within the range allowed by the robust feasible domain, thereby achieving closed-loop collaborative control of multiple types of energy storage on the entire time scale.

[0201] In terms of conventional power sources, to verify the synergistic effect of units with different regulation capabilities, we selected five representative thermal power units in the case study, covering typical configurations ranging from 1000MW baseload units to 300MW flexible units. The specific technical parameters of each thermal power unit are shown in Table 1.

[0202] Table 1. Technical Parameters of Conventional Thermal Power Units

[0203] Unit number Rated capacity (MW) Minimum technical output (MW) Climbing rate (MW / min) Marginal cost coefficient (yuan / MW) G1 1000 500 15 285 G2 600 300 10 305 G3 600 300 10 305 G4 300 120 8 340.5 G5 300 90 12 450

[0204] The specific energy storage configuration is shown in Table 2: The system is configured with a pumped-storage power station participating in the multi-day rolling optimization of step S1, utilizing its large capacity characteristics to smooth out long-cycle fluctuations. Simultaneously, electrochemical energy storage is configured to participate in day-ahead and intraday dispatching. Furthermore, this embodiment sets the adjustment cost of electrochemical energy storage based on actual market data to ensure the economic realism of the dispatching.

[0205] Table 2. Parameter Table for Collaborative Configuration of Multiple Energy Storage Types

[0206] Energy storage type Rated power (MW) Rated capacity (MWh) Charge and discharge efficiency Climbing rate (MW / min) Adjusting cost / life loss factor Pumped storage 300 1800 75% 20 0 Electrochemical energy storage 100 200 90% 100 50 yuan / MW

[0207] The simulation experiment used load curves and wind and solar power output forecast data for a typical day. The system's peak load was approximately 2,100 MW, and it exhibited a high wind and solar power penetration rate. Based on the dynamic time zone division mechanism proposed in this invention, the system first adaptively divides the scheduling day into a new energy consumption time zone, an auxiliary peak-shaving time zone, and a regular time zone according to the statistical characteristics of the net load forecast curve, such as... Figure 2 As shown, for example, during the midday period when photovoltaic output surges and net load drops significantly, the system automatically identifies the "new energy consumption time zone" and guides pumped storage and electrochemical energy storage to charge in tandem by dynamically adjusting the weight of the objective function, effectively alleviating the pressure of curtailment of solar power.

[0208] Simulation results show that, compared with the traditional deterministic scheduling mode with fixed time periods, the method proposed in this invention achieves significant improvements in both economy and safety. In terms of economic indicators, thanks to the rapid smoothing of high-frequency fluctuations by electrochemical energy storage, the frequent ramp-up and start-stop cycles of high-cost thermal power units (such as G5) are significantly reduced, effectively controlling the total daily operating cost of the system. Regarding absorption indicators, the dynamic zoning strategy reduces the amount of renewable energy curtailed by approximately 18.5%, significantly improving the system's capacity to accommodate renewable energy. Meanwhile, as... Figure 3 As shown, under the dynamic feasible region constraint generated by the two-stage robust optimization, the evolution curves of the state of charge (SOC) of various types of energy storage strictly follow the robust envelope, thereby ensuring that it is always within the physical safety range of 0.1 to 0.9. Moreover, at the end of the scheduling cycle, it successfully returns to the target level set by the multi-day rolling optimization, verifying that the present invention effectively takes into account the continuity of cross-day regulation capability while ensuring the intraday operation safety of the system.

[0209] Example 2

[0210] This embodiment discloses a multi-type energy storage multi-timescale collaborative scheduling system based on dynamic time zone division, including a method for implementing the multi-type energy storage multi-timescale collaborative scheduling method described in Embodiment 1. Figure 4 As shown, the system also includes:

[0211] The multi-day rolling optimization module is configured to execute step S1, which involves acquiring historical operating data of the power system, multi-day load and new energy forecast data, and real-time rolling ultra-short-term forecast data, constructing a long-cycle optimization model, and calculating and outputting the end-of-day SOC boundary constraints of pumped storage.

[0212] Specifically, it is responsible for acquiring historical operating data of the power system and forecasting data of new energy sources and loads for the next few days; performing multi-day rolling optimization calculations that decouple the prediction time domain and the execution time domain, comprehensively considering the coal-fired cost, start-up and shutdown cost of thermal power units and the operation and maintenance cost of pumped storage, and outputting the end-of-day state-of-charge boundary value of pumped storage under the conditions of satisfying system power balance constraints, thermal power unit ramp-up and start-up and shutdown time constraints, and pumped storage cross-time period energy coupling constraints.

[0213] The dynamic partitioning scheduling module is currently configured to execute step S2, which calculates net load statistical characteristics and thresholds, and generates dynamic time zone division instructions and adaptive objective functions.

[0214] Specifically, based on the daily end-of-day charge state boundary value of pumped storage output by the multi-day rolling optimization module, and combined with the day-ahead renewable energy power prediction curve and net load fluctuation characteristics, the system net load sequence and its rate of change characteristics are calculated; with the goal of adapting to the source-load time sequence characteristics, the scheduling day is adaptively divided into renewable energy consumption time zone, auxiliary peak shaving time zone and regular time zone according to the preset net load amplitude and ramp rate threshold, and a multi-type energy storage collaborative scheduling objective function containing adaptive weight coefficients is constructed.

[0215] The two-stage robust optimization module is configured to execute step S3, construct and solve the two-stage robust optimization model, and output the robust feasible region of multi-type energy storage SOC covering uncertainties.

[0216] Specifically, the collaborative scheduling objective function is constructed based on the day-ahead dynamic partition scheduling module, and a set of uncertainties in new energy output is constructed by combining historical prediction error data; a two-stage robust optimization model of Min-Max-Min is established, and the column and constraint generation algorithm is used to solve it. Under the condition of satisfying the system operation safety constraints in the worst scenario, the robust feasible domain of energy storage state of charge and the benchmark output plan of thermal power units in the scheduling day are output for multiple types of energy storage.

[0217] The intraday real-time rolling correction module is configured to execute step S4, which receives SOC robust feasible region constraints and ultra-short-term forecast data, and generates real-time correction instructions based on the MPC strategy.

[0218] Specifically, based on the robust feasible region of the energy storage state of charge output by the two-stage robust optimization module, the system uncertainty state is updated using ultra-short-term renewable energy power prediction data; a model predictive control strategy is adopted to minimize the real-time power imbalance and adjustment cost, and under the condition that the real-time energy storage state does not exceed the robust feasible region, the real-time corrected output command of each energy storage power station is solved and output.

[0219] Example 3

[0220] This embodiment discloses a storage medium storing a program. When the program is executed by a processor, it implements the multi-type energy storage multi-timescale collaborative scheduling method described in Embodiment 1. The storage medium in this embodiment can be a disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), USB flash drive, portable hard drive, or other media.

[0221] Example 4

[0222] This embodiment discloses a computing device, including a processor and a memory for storing processor-executable programs. When the processor executes the program stored in the memory, it implements the multi-type energy storage multi-timescale collaborative scheduling method described in Embodiment 1. The computing device described in this embodiment can be a desktop computer, laptop computer, smartphone, PDA handheld terminal, tablet computer, programmable logic controller (PLC), or other terminal device with processor functionality.

[0223] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A multi-type energy storage multi-time scale collaborative scheduling method, characterized in that: Includes the following steps: Step S1: Obtain historical operating data of the power system and forecast data of new energy sources and loads for the next few days, construct a multi-day rolling optimization model for long-term optimization calculation, and obtain the end-of-day state of charge boundary value of pumped storage as the boundary condition for day-ahead optimization. Step S2: Based on the end-of-day state-of-charge boundary value of pumped storage obtained in Step S1, and combined with the day-ahead renewable energy power prediction curve and net load fluctuation characteristics, the scheduling day is divided into renewable energy consumption time zone, auxiliary peak shaving time zone and regular time zone using dynamic programming method. An objective function for the coordinated scheduling of pumped storage and electrochemical energy storage is constructed to adapt to the characteristics of different time zones. Step S3: Based on the objective function for coordinated scheduling of multiple types of energy storage adapted to the characteristics of different time zones constructed in Step S2, and considering the uncertainty of new energy prediction error, a two-stage robust optimization model is established to solve for the robust feasible domain of energy storage state of charge of multiple types of energy storage within the scheduling day. Step S4: Based on the robust feasible region of the state of charge of the multi-type energy storage obtained in step S3, during the intraday real-time operation phase, the uncertainty set is updated using ultra-short-term forecast data, and the output command is solved in real time to realize the coordinated control of the multi-type energy storage within the robust feasible region.

2. The method for coordinated scheduling of multiple types of energy storage across multiple time scales according to claim 1, characterized in that: The step S1 is to plan the start-stop of the thermal power unit and the power management of the pumped storage in a longer time dimension by introducing a multi-day rolling optimization mechanism, the multi-day rolling mechanism adopts an optimization framework of decoupling of a prediction time domain and an execution time domain; including: setting the prediction time domain of the multi-day rolling optimization as , taking 3-7 days, i.e. 72-168 hours, and the execution time domain as , 24 hours, and taking the minimum total system operation cost in the prediction time domain as the target to build a target function: In the formula, The total operating cost of the system, For time indexing, For thermal power units, It is a pumped storage power station. and It is a collection of thermal power units and pumped storage power stations. For thermal power units The cost of coal, For thermal power units Startup costs, For thermal power units The cost of no-load operation Pumped storage power station The operation and maintenance costs For time period The amount of wind and solar power that has been curtailed. For time period The shear load, and These are their corresponding penalty coefficients. As the terminal deviation penalty factor, For the first The decision-making process at the end of the day accumulates energy. This is the expected terminal water level pre-set based on long-term statistical patterns; The constraints that the multi-day rolling optimization needs to meet include: System power balance constraints: In the formula, and Pumped storage power stations The discharge power and charging power, The day-ahead forecast values ​​that contribute to the system's wind and solar power. This is the day-ahead forecast of the total system load; Thermal power unit operating constraints: Upper and lower limits of thermal power unit output constraints: In the formula, and thermal power units The lower and upper limits of technical output, For unit start-up and shutdown status variables, when When taking 1, thermal power units exist It is always in a shutdown state, when When the value is 0, the thermal power unit exist It is always powered on; Thermal power unit ramping constraints: In the formula, and thermal power units The limits for uphill and downhill climbing speeds; Minimum start-up and shutdown time constraints for thermal power units: In the formula, and thermal power units During the period The duration of continuous power-on time and the duration of continuous power-off time. and thermal power units Minimum continuous power-on time and minimum continuous downtime; Energy coupling constraints of pumped storage across time periods: To characterize the dynamic change of water storage capacity in the upper reservoir of a pumped storage power station over continuous time, and to prevent power depletion due to optimization window truncation, the following energy balance equation and reservoir capacity constraints are established: In the formula, Pumped storage power station exist The energy storage of the upper reservoir during the period, for Energy storage during a certain period; They are respectively Pumping power and power generation during the same period; These are pumping efficiency and power generation efficiency, respectively. , These are the minimum and maximum allowable storage capacity of the upper reservoir, respectively. The duration is the length of the time period. For the first The stored energy at the end of each operating day. The preset target energy threshold at the end of the day; this constraint mandates that the system retain sufficient adjustment capacity at the end of each day to support continuous operation for several more days to come; Mutual exclusion constraints for pumped storage operation status: In the formula: , The units exist The pumping and power generation status for a given time period is represented by 1 for operation and 0 for shutdown. and These are the minimum and maximum power limits for the unit under the corresponding operating conditions.

3. The multi-type energy storage multi-timescale collaborative scheduling method according to claim 2, characterized in that: Step S2 involves constructing a day-ahead collaborative scheduling model based on dynamic time zone division. This model is achieved by dynamically dividing scheduling time zones and constructing an adaptive objective function. The specific process is as follows: Given the limitations of traditional static peak-valley time zone division in adapting to the source-load timing characteristics under high-proportion renewable energy access, a dynamic time zone division mechanism based on net load morphology characteristics is introduced. Based on the current renewable energy and load forecast data, the system equivalent net load sequence is constructed. Then, considering the amplitude distribution and rate of change characteristics of the net load, a multi-dimensional time zone determination criterion is established, and the 24 time periods in the scheduling cycle are adaptively deconstructed into three types of characteristic time zones: renewable energy consumption, auxiliary peak shaving, and conventional time zones. The mathematical criterion for the dynamic time zone division is expressed as follows: In the formula: for Forecasted system net load for the time period The rate of change of net load over adjacent time periods. and These are the net load thresholds for determining the renewable energy consumption time zone and the auxiliary peak-shaving time zone, respectively. To determine the ramp rate threshold for drastic load fluctuations, , , These represent the time zones for renewable energy consumption, auxiliary peak shaving, and regular time zones, respectively. To address the characteristics of different time zones, electrochemical energy storage is incorporated into the optimization framework. Leveraging its fast response speed in conjunction with pumped hydro storage, a day-ahead coordinated scheduling objective function with adaptive weighting coefficients is constructed. These adaptive weighting coefficients are dynamically adjusted based on the characteristic time zone type of the current period. In renewable energy consumption time zones, the proportion of the penalty weight for wind and solar curtailment is increased; in auxiliary peak-shaving time zones, the proportion of the penalty weight for safety margin is increased. The expression is as follows: In the formula, The total cost of the coordinated scheduling system as of now. , , The adaptive weighting coefficients for the renewable energy consumption time zone, auxiliary peak-shaving time zone, and conventional balance time zone during time period t are determined by the characteristic time zone to which this period belongs. , The system's cost of curtailing wind and solar power. This is a system safety margin penalty item used to penalize the risk of insufficient spinning reserve. This represents the total cost of system collaborative operation up to date. For thermal power units The cost of coal, Pumped storage power station Start-up and shutdown costs For electrochemical energy storage The unit charge / discharge loss coefficient calculated over the entire life cycle. and Electrochemical energy storage The charging and discharging power, It is an electrochemical energy storage system; Regarding the value of the weight, when t is in the new energy consumption time zone, set... This significantly increases the penalty weight for wind and solar curtailment, forcing the system to prioritize the use of energy storage for charging and consumption. When t is in the auxiliary peak-shaving time zone, the following settings are made: Increasing the safety margin penalty weight forces energy storage to reserve discharge capacity at its peak. When t is in the normal equilibrium time zone, the following settings are applied. As the dominant weight, , The base value is taken, and the system then reverts to the economy-first mode. This mapping mechanism eliminates the blindness of traditional manual weight assignment and enables the scheduling strategy to switch automatically according to the net load characteristics. Regarding constraints, in addition to conventional system balancing and unit operation constraints, differentiated operation constraints for various types of energy storage were constructed, and the inter-day energy boundary passed in step S1 was introduced; the mathematical expressions for the relevant core constraints are as follows: Pumped storage constraints: In the formula, The target electricity consumption at the end of the day, derived from the multi-day rolling optimization in step S1, For the allowed boundary relaxation range; Electrochemical energy storage operation constraints: In the formula, For electrochemical energy storage Percentage of state of charge at time t For its rated capacity, and These are their charging and discharging efficiencies, respectively. and These are the safe upper and lower limits for its state of charge. and This refers to the battery's charge / discharge state. Take 1 as discharge. Set 0 to indicate charging; Thermal power output constraints based on stationary unit combinations: The real-time output of thermal power units is only allowed to be adjusted within a fixed physical range. In the formula, Given the known parameters passed in step S1, this constraint ensures that the day-ahead fine-grained scheduling strictly follows the multi-day rolling optimization mechanism; System power balance constraints: 。 4. The method for coordinated scheduling of multiple types of energy storage across multiple time scales according to claim 3, characterized in that: In step S3, considering the uncertainty of new energy prediction errors, a two-stage robust optimization model is established. This model constructs an uncertainty set of new energy output and seeks the optimal response strategy under the worst-case scenario using a "pre-scheduling-rescheduling" framework. The specific process is as follows: To quantify the uncertainty of new energy output, a box-shaped uncertainty set is constructed based on historical forecast error data to describe the potential deviation of actual output from the range of the current day forecast value. The mathematical expression for the uncertainty set of new energy output is as follows: In the formula, and Wind power and solar power during different time periods The real-time power output value and the day-ahead forecast value, For time period The maximum permissible range of prediction error fluctuation, i.e., the radius of uncertainty. and These are auxiliary variables used to characterize positive and negative prediction biases; This is the uncertainty budget parameter, used to control the extreme degree of uncertainty scenarios considered. The larger the value, the more conservative the model. Based on the aforementioned set of uncertainties, a two-stage robust optimization objective function, Min-Max-Min, is constructed; this function aims to minimize the total system cost under the worst-case uncertainty scenario. The mathematical expression for the two-stage robust optimization objective function is as follows: In the formula, The first phase of day-ahead pre-dispatch decision variables includes the planned output of thermal power units, the energy storage infrastructure charging and discharging plan, and the boundary of its feasible region. Its feasible region is... , The basic operating cost for the first stage is calculated based on the model in step S2; Belongs to the set of uncertainty Any new energy power output scenario, The second-stage real-time rescheduling decision variables include unit output adjustment, energy storage power correction, and load shedding, etc., and its feasible region... Subject to the first phase of decision With uncertain scenarios Common constraints This refers to the penalty costs incurred during the rescheduling phase due to deviation adjustments, such as load shedding costs and emergency energy curtailment costs. Unlike fixed trajectories, this method models the state of charge of energy storage as a dynamic band-shaped interval, ensuring that the real-time SOC can be maintained within the interval under any permissible uncertainties, thereby reserving sufficient adjustment capability; Constraints of the two-stage robust optimization model: Robust operation and boundary connection constraints of pumped storage: In the formula, to ensure the convergence of optimization in extreme scenarios, a minimal relaxation variable is introduced into the boundary constraints. This is a permissible minimal robust relaxation; Electrochemical energy storage robust SOC constraint: Thermal power output constraints based on stationary unit combinations: The real-time output of thermal power units is only allowed to be adjusted within a fixed physical range. By solving the above model using a column and constraint generation algorithm, robust feasible regions of various energy storage SOCs capable of withstanding the most severe renewable energy fluctuation scenarios can be obtained, along with the corresponding baseline output plan for thermal power generating units; as long as The energy storage state at any given time lies within this interval, for the set of uncertainties. Any perturbation in All systems have feasible control strategies to enable... The state at time 1 remains within the interval of the next time 1, and satisfies the end-of-day boundary constraints.

5. The multi-type energy storage multi-timescale collaborative scheduling method according to claim 4, characterized in that: In step S4, based on the robust feasible region of the energy storage state of charge obtained in step S3, during the intraday real-time operation phase, the uncertainty set is updated using ultra-short-term forecast data, and the specific process for solving the real-time corrected output command is as follows: During the intraday real-time operation phase, a Model Predictive Control (MPC) strategy is employed, with a rolling execution time window set as follows: The time resolution is ; Based on the latest ultra-short-term renewable energy power forecast data Update the uncertain input state of the system; The core objective of optimization at this point is to track the real-time power imbalance of the system with minimal adjustment cost while satisfying the safety boundaries determined in the previous day. The objective function for intraday real-time rolling correction is constructed, and its mathematical expression is as follows: In the formula, For the current moment, For the predicted time within the scrolling window, This represents the output adjustment of thermal power units relative to the day-ahead baseline plan. To adjust the cost coefficient, Representing pumped hydro storage and electrochemical energy storage, This refers to the power deviation of various types of energy storage relative to the day-ahead pre-dispatch command. This is the corresponding deviation penalty coefficient. For real-time power imbalance residuals, This is the real-time power imbalance penalty factor; Regarding constraints, a robust feasible region constraint for the energy storage charge state is introduced; in order to ensure the robustness of the day-ahead scheduling strategy, the energy storage energy state in real-time operation is forcibly constrained within the envelope range calculated in step S3. Real-time collaborative operation constraints: In the formula, and These represent the actual energy states of pumped hydro storage and electrochemical energy storage during real-time operation, respectively. and These are the lower and upper boundaries of the robust feasible region of the energy storage charged state obtained in step S3, respectively. The real-time ramp rate limitation for electrochemical energy storage The real-time ramp rate limit for pumped storage; Terminal state guidance constraints: In the formula, The rolling prediction time domain length for model predictive control; This is the end time of the prediction window; Penalty cost for deviation of terminal state; By solving the above model, the system generates the current time. The final power command is issued and distributed to each energy storage power station for execution. This strategy utilizes the high ramp rate of electrochemical energy storage to prioritize smoothing out high-frequency and small-amplitude power fluctuations, and utilizes the large capacity characteristics of pumped hydro storage to bear continuous energy imbalances within the range allowed by the robust feasible domain, thereby achieving closed-loop collaborative control of multiple types of energy storage on the entire time scale.

6. A multi-type energy storage multi-timescale collaborative scheduling system, characterized in that: The method includes a multi-type energy storage multi-timescale collaborative scheduling method for implementing any one of claims 1 to 5, and further includes: The multi-day rolling optimization module is responsible for acquiring historical operating data of the power system and forecasting data of new energy sources and loads for the next few days; it performs multi-day rolling optimization calculations that decouple the prediction time domain from the execution time domain, comprehensively considering the coal cost of thermal power units, start-up and shutdown costs, and operation and maintenance costs of pumped storage, and outputs the end-of-day state-of-charge boundary value of pumped storage under the conditions of satisfying system power balance constraints, thermal power unit ramp-up and start-up and shutdown time constraints, and pumped storage cross-time period energy coupling constraints; The day-ahead dynamic zone scheduling module, based on the day-ahead renewable energy power prediction curve and net load fluctuation characteristics output by the multi-day rolling optimization module, calculates the system net load sequence and its rate of change characteristics. With the goal of adapting to the source-load time sequence characteristics, the scheduling day is adaptively divided into renewable energy consumption time zone, auxiliary peak shaving time zone and regular time zone according to the preset net load amplitude and ramp rate threshold. A multi-type energy storage collaborative scheduling objective function containing adaptive weight coefficients is constructed. The two-stage robust optimization module is based on the collaborative scheduling objective function constructed by the day-ahead dynamic partition scheduling module, and constructs a set of uncertainties in new energy output by combining historical prediction error data; a two-stage robust optimization model of Min-Max-Min is established and solved by column and constraint generation algorithm. Under the condition of satisfying the system operation safety constraints in the worst scenario, it outputs the robust feasible domain of energy storage state of charge and the benchmark output plan of thermal power units in the scheduling day for multiple types of energy storage. The intraday real-time rolling correction module, based on the robust feasible region of the energy storage state of charge output by the two-stage robust optimization module, updates the system uncertainty state using ultra-short-term renewable energy power prediction data; adopts a model predictive control strategy, with the goal of minimizing real-time power imbalance and adjustment costs, and solves and outputs the real-time corrected output command of each energy storage power station under the condition that the real-time energy storage state does not exceed the robust feasible region.

7. A storage medium storing a program, characterized in that, When the program is executed by the processor, it implements a multi-type energy storage multi-timescale collaborative scheduling method as described in any one of claims 1 to 5.

8. A computing device, comprising a processor and a memory for storing a processor-executable program, characterized in that, When the processor executes the program stored in the memory, it implements the multi-type energy storage multi-timescale collaborative scheduling method as described in any one of claims 1 to 5.

Images