Source load storage control method for sustainable and stable output of new energy output power of active power grid

Abstract

The invention relates to a source load storage control method for sustainable and stable output of new energy output power of an active power grid, and belongs to the technical field of new energy power systems. The method mainly comprises the following steps: based on ARIMA model prediction and load elastic response, guiding a user to optimize power consumption through real-time state perception, power prediction and time-of-use electricity price; a reference value is set and dynamically adjusted through multi-source data fusion, historical data and a scheduling instruction are integrated, and an output value of the source-storage combined system is corrected and output through an ARIMA model; a collaborative optimization mechanism is constructed, and load prediction, energy storage life and power grid interaction economy are optimized through multiple objective functions. Dynamic monitoring is realized through generalized power modeling and bidirectional flow discrimination, precision is improved by combining ARIMA prediction and rolling correction, and abandoned power is reduced. The load side actively participates in the electricity market, and the smooth curve relieves peak pressure; the intelligent charging and discharging strategy prolongs the energy storage life, optimizes economical efficiency and environmental protection, and provides a feasible technical path for a high-proportion new energy power grid.

Description

Technical Field

[0001] The present invention belongs to the technical field of new energy power systems, and in particular relates to a source-load-storage control method for achieving sustainable and stable output of new energy output power in an active power grid. Background Art

[0002] As the global energy structure transitions toward cleaner energy, the proportion of renewable energy sources, represented by wind power and photovoltaics, in the power system continues to rise. The power grid is gradually transforming from a traditional one-way power supply model to an active network with two-way interaction between sources and loads, significantly increasing the complexity and uncertainty of power flow. In this context, source-load-storage collaborative control technology has become key to supporting the high proportion of renewable energy consumption. Currently, countries are actively exploring energy storage configuration optimization, demand-side response mechanisms, and the construction of multi-energy complementary systems, aiming to flexibly adjust resources to match the fluctuating characteristics of renewable energy. However, the contradiction between the temporal and spatial imbalance of renewable energy generation and the lag in the grid's regulation capacity remains prominent, and more intelligent solutions are urgently needed.

[0003] In terms of renewable energy output regulation, existing technologies mainly implement power planning through time-sharing scheduling combined with prediction algorithms. For example, machine learning models based on historical meteorological data predict power generation and dynamically adjust output based on electricity price signals. Load-side management focuses on the classification and aggregation of adjustable loads, using price incentives to guide users to use electricity during off-peak hours. Energy storage systems often use fixed charging and discharging strategies or simple peak-valley arbitrage models to smooth out short-term power fluctuations. In addition, some solutions attempt to combine multiple energy sources such as wind, solar, and storage to improve system stability through multi-energy complementarity. These methods have alleviated the pressure on renewable energy grid connection to a certain extent, but they still have limitations in terms of real-time coordination and stability of the three ends of source, load, and storage.

[0004] Existing technologies often exhibit a fragmented state of coordinated control of power generation, load, and storage. New energy stations typically aim to maximize power generation, but the grid dispatch instructions lack a dynamic connection with energy storage charging and discharging strategies, leading to frequent power curtailment. The load-side response mechanism relies too much on manual experience to set the intensity of electricity price incentives, and user participation is limited by the deviation between the response model and actual behavior. Energy storage systems often use static charge and discharge thresholds, which fail to fully integrate the real-time needs of the grid with the health status of the battery. This not only affects economic efficiency but also accelerates equipment loss, making it difficult for the three terminals of power generation, load, and storage to operate stably for a long time. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to construct a smart grid dynamic optimization system that is adaptable to a high proportion of renewable energy access, and to solve the problem of unstable grid operation caused by the volatility of renewable energy through real-time perception, accurate prediction and multi-objective collaborative control.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A source-load-storage control method for enabling sustainable and stable output of new energy output power of an active power grid comprises the following steps:

[0008] S1. Autoregressive Integrated Moving Average Model (ARIMA) data forecasting and load elastic response based on real-time monitoring, including the following steps:

[0009] S1.1. By building a real-time state perception and dynamic power direction determination model and a multi-dimensional perception system, we can achieve power difference calculation, bidirectional power balance control, and dynamic grid status acquisition and processing for the combined source and storage system. We can also monitor source output, grid connection point power flow, and load demand in real time to establish a judgment mechanism.

[0010] S1.2. Construct an ARIMA forecasting model to predict the short-term power of the combined source and storage system, providing data support for the forecast correction term in the dynamic baseline value;

[0011] S1.3. By using time-of-day electricity prices and a load utility function with diminishing marginal utility, we can guide users to use more electricity during low-price periods and less during high-price periods to maximize utility. By using Nash equilibrium mediation to solve the optimal response strategy, we can enable users to voluntarily reduce their electricity consumption during high-price periods when electricity prices rise, thus smoothing the load curve and alleviating peak pressure on the power grid.

[0012] S2. Setting and dynamically adjusting the initial benchmark value of the new energy source and storage combined system based on multi-source data fusion includes the following steps:

[0013] S2.1. Build a combined source-storage system by calculating the difference in power generated at the source and establishing the state equation of the energy storage system.

[0014] S2.2. Integrate historical source and storage output with dispatch instructions, use the exponential moving average algorithm and smoothing coefficient to set the baseline value, establish a model library to select initial values ​​and trend items, and support rolling corrections;

[0015] S2.3. Based on the initial baseline value, the weather forecast correction is generated through the ARIMA model, the grid dispatch offset is calculated, and after weighted integration of historical trends, short-term forecasts, and dispatch instructions, the corrected power generation and storage system output is output;

[0016] S3. Build a source-load-storage collaborative optimization mechanism:

[0017] S3.1. Construct an objective function based on variables such as renewable energy output, energy storage charging and discharging, and electricity purchase and sales to achieve instantaneous power balance between power supply sources and power consumption destinations.

[0018] S3.2. Multi-objective optimization function based on source-load-storage collaborative optimization: A reference load prediction objective function is constructed by combining peak load and historical load data. The objective function for energy storage system life loss cost is constructed to minimize the loss cost by minimizing the absolute deviation between the actual load adjustment and the reference load within a specified time limit. A squared charge and discharge power term and a penalty term for state of charge (SOC) variation are introduced to suppress drastic SOC fluctuations and extend battery cycle life. An economic cost objective function for grid interaction is constructed to minimize the difference between electricity purchase costs and electricity sales revenue.

[0019] S3.3. System stability optimization: Minimize the deviation between the energy storage state and the target value, as well as the deviation of the response time, by constructing a total cost objective function for system stability.

[0020] Furthermore, the real-time state perception and power direction dynamic determination model is constructed in S1.1, specifically including the following steps:

[0021] Build a high-frequency data acquisition system to monitor distributed renewable energy output power and other data in real time through sensors and smart meters;

[0022] Obtain data from a high-frequency data acquisition system, embed new energy parameters into the grid node model, and derive a generalized node power vector.

[0023] A bidirectional power flow formula is constructed to describe the power flow direction of the power grid to determine whether to supply or withdraw power to the grid.

[0024] Furthermore, the ARIMA forecasting model is constructed in S1.2, specifically including the following steps:

[0025] Preprocess and differentiate the renewable energy power generation, load, and electric vehicle charging curves to ensure the stationarity of the time series;

[0026] Use autocorrelograms and partial autocorrelograms to set the optimal values ​​of ARIMA model parameters;

[0027] An autoregressive term is used to describe the relationship between the current value and its past value, and an autoregressive model is established;

[0028] The weighted average of past errors is used to describe the forecast error of the current value and a sliding average model is established;

[0029] Fit the ARIMA model to generate the short-term power forecast value of the source-storage combined system for the next 24 hours.

[0030] Furthermore, the utility function of the load in S1.3 is expressed as the following formula:

[0031] U i =βln(1+αi )-λ(t)α i P shift,i

[0032] In the formula, βln(1+α i ) is the electricity satisfaction, reflecting the diminishing marginal utility of users when electricity consumption increases; λ(t) is the time-dependent penalty coefficient, which is used to indicate the penalty intensity for subsequent items; α i P is the load transfer ratio, which indicates the ratio of the user's electricity consumption in a specific period to his original electricity consumption; shift,i The load transfer amount made by the load end based on the electricity price signal in the power consumption management;

[0033] The solution to the Nash equilibrium mediation in S1.3 is expressed as the following formula:

[0034]

[0035] Where, It is the optimal response that maximizes the load utility under the current given parameters; It is the negative impact that may be caused after the response occurs. The larger the value, the greater the negative impact that may be caused, which will lead to a lower optimal response value.

[0036] Furthermore, the difference in the power generated by the source end in S2.1 is expressed as the following formula:

[0037] ΔP(t)=P(t)-P ref

[0038] Where, P(t) is the actual output power of the new energy source at the current moment; P ref is the initial reference value pre-set by the grid dispatching system or joint control strategy within the time period; ΔP(t) is the power difference, reflecting the deviation between the actual output of the source end and the plan; if ΔP(t)>0, the energy storage is charged, indicating that the actual power generation at the source end exceeds the dispatching expectation; if ΔP(t)<0, the energy storage is discharged, indicating that the power generation at the source end is lower than expected.

[0039] Furthermore, the state equation of the energy storage system in S2.1 is expressed as the following formula:

[0040]

[0041] Where SOC(t) is the energy storage state of charge at time t, expressed as a percentage; η ch is the charging efficiency, which indicates the ratio of energy lost during the charging process; P ch (t) is the charging power at time t; Δt is the time step, which represents the time interval between each state update; P dis(t) is the discharge power at time t; η dis is the discharge efficiency, which indicates the energy loss during the discharge process.

[0042] Furthermore, the step S2.2 specifically includes the following steps:

[0043] Based on historical data, the power generation output of sources and storage is integrated with grid dispatch instructions, and the corresponding benchmark value is set through the exponential moving average algorithm and smoothing coefficient;

[0044] Set the time-based benchmark values ​​for the total output of the combined power generation and storage system and the initial benchmark values ​​for power grid dispatch;

[0045] Establish a typical benchmark value pattern library based on time period characteristics and usage scenarios, and select initial benchmark values ​​and trend items for each time period to support subsequent rolling correction operations;

[0046] Based on real-time grid dispatch instructions, calculate grid dispatch offset and quantify grid dynamic adjustment requirements;

[0047] The initial benchmark value, meteorological forecast value and power grid demand value are weighted and integrated, and the corrected power generation and storage system output value is output through historical trend smoothing, short-term forecast response and dispatch instruction compensation.

[0048] Furthermore, the step S2.3 specifically includes the following steps:

[0049] Based on the initial benchmark value, the real-time dispatch instructions of the power grid, load forecast results and weather forecast data are integrated as correction inputs;

[0050] The meteorological data is processed through the ARIMA forecast model to generate short-term power meteorological forecast values, based on which the baseline value correction at the forecast level is calculated.

[0051] Furthermore, the objective function in S3.2 selects different weight coefficients according to the specific time period, which can be expressed as the following formula:

[0052]

[0053] in, is the weight of the i-th optimization objective, the larger the value, the higher the priority; β is the sensitivity coefficient of weight allocation, the larger the value, the larger the weight difference; Score the urgency of the target in time period t, the higher the score, the lower the weight; R min The minimum score among all objectives, used for numerical stability guarantee;

[0054] The urgency score is mainly composed of three indicators, which can be expressed as the following formula:

[0055]

[0056] in, It is a safety indicator, indicating the risk of SOC exceeding the limit; It is an economic indicator that indicates the deviation of electricity prices from the benchmark; To adjust the demand index, it represents the new energy forecast error.

[0057] Furthermore, the generalized node power vector It is expressed as the following formula:

[0058]

[0059] Where, P gen The electric power generated by all new energy power generation units in the power grid; P load The power consumed by various users in the power grid, that is, the load demand of the power grid; P storage The electricity stored or released by the energy storage system in the power grid; EV The electrical power involved in charging and discharging electric vehicles.

[0060] The bidirectional power flow formula of the power grid is expressed as follows:

[0061] P net (t) = P gen (t)+P dis (t)-P ch (t)-P load (t)

[0062] Where, P net is the net power difference in the grid, a positive value indicates power is supplied to the grid, and a negative value indicates power is taken from the grid; P dis P is the power released by the energy storage system to the grid or local load; ch P is the power absorbed by the energy storage system from the grid or renewable energy generation; load It is the total power consumption of local loads, including rigid loads and transferable loads.

[0063] The beneficial effects of the present invention are:

[0064] Based on generalized power modeling and bidirectional flow discrimination technology, dynamic monitoring of the power grid status is achieved; combined with an improved ARIMA forecasting model and rolling correction mechanism, the forecast accuracy of renewable energy output and load demand is improved, and the corresponding initial benchmark values ​​are generated; and a dynamic benchmark value rolling correction mechanism is established to effectively guide the three elements of source, load and storage to work together and jointly participate in grid regulation, thereby ensuring the real-time power balance of the power grid.

[0065] The introduction of a time-based benchmark value system allows for real-time optimization and adjustment of power output by dynamically balancing renewable energy generation with grid demand, ensuring maximum utilization of renewable energy. This optimization mechanism allows renewable energy generation to be more closely integrated with traditional energy and grid demand, reducing power curtailment due to energy mismatch.

[0066] Transform the load-side response mode from passively accepting power dispatch to actively participating in the power market. In this process, the optimized load response mechanism can adjust user electricity consumption behavior in real time to adapt to changes in grid demand. This not only improves the smoothness of the load curve, but also enhances the grid's overall ability to respond to load fluctuations. Users make flexible adjustments based on electricity price information and their own electricity needs, forming a more proactive participation attitude, thereby improving the operating efficiency of the power system and enhancing users' electricity experience;

[0067] Intelligent charging and discharging strategies have been designed to not only provide power buffering but also extend service life. These strategies use dynamic programming models to comprehensively consider various loss factors during the charging and discharging process, thereby optimizing the efficiency of energy storage. This innovative design enables energy storage equipment to manage energy flow more efficiently, reducing the risk of attenuation caused by frequent charging and discharging. Overall, the operation of this system not only takes into account the reliability of power supply, but also optimizes economic efficiency and environmental friendliness, providing a feasible and effective technical path for a high-proportion new energy grid;

[0068] A new energy source-storage integrated system based on time-based benchmark values ​​was constructed. By integrating multiple sources of information, including renewable energy generation data, energy storage equipment operating status, and grid load demand, a dynamic benchmark value correction model was established. This system analyzes renewable energy output volatility and energy storage regulation capabilities in real time, dynamically correcting power allocation benchmark values ​​to optimize the source-storage coordinated operation strategy. Combining historical data mining with real-time forecasting technology, the system dynamically adjusts benchmark values ​​to external factors such as weather conditions and load fluctuations, significantly improving the system's adaptability to renewable energy uncertainties.

[0069] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below with reference to the accompanying drawings, in which:

[0071] Figure 1A schematic diagram of a flow chart of an embodiment of the present invention;

[0072] Figure 2 This is a schematic diagram of the source-storage combined system architecture according to an embodiment of the present invention;

[0073] Figure 3 This is a module-based flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0074] The following describes the embodiments of the present invention by means of specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and the features in the following embodiments and embodiments can be combined with each other without conflict.

[0075] Among them, the accompanying drawings are only for illustrative purposes and represent only schematic diagrams rather than actual pictures, and should not be understood as limiting the present invention. In order to better illustrate the embodiments of the present invention, some parts of the accompanying drawings may be omitted, enlarged or reduced, and do not represent the dimensions of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions may be omitted in the accompanying drawings.

[0076] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", "front", "back", etc. indicating directions or positional relationships, they are based on the directions or positional relationships shown in the drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0077] This paper addresses the core issues of large power fluctuations, low prediction accuracy, and difficulty in multi-objective coordination brought about by the high proportion of renewable energy integrated into smart grids. It proposes a closed-loop "perception-prediction-optimization" solution. This approach utilizes generalized power vector modeling and a bidirectional flow discriminant model to achieve real-time perception of the grid's dynamic state. Furthermore, based on time-series forecasting using the Autoregressive Integrated Moving Average (ARIMA) model, it improves the 24-hour forecast accuracy of renewable energy output and load demand, generating initial time-based benchmark values.

[0078] Based on multi-source data fusion and dynamic benchmark value rolling correction, a source-load-storage multi-objective collaborative optimization system is established. The State of Charge (SOC) equation and the squared charge and discharge power terms are introduced on the energy storage side to extend battery life and stabilize system operation. On the grid side, power interaction strategies are optimized through time-of-use electricity pricing. Dynamic weight allocation is used to adjust the priorities of safety, economy, and regulation objectives in real time, achieving rolling corrections to dynamic benchmark values. Ultimately, this system achieves the coordinated optimization of new energy consumption rate, operating costs, and grid frequency compliance, breaking through the technical bottlenecks of traditional power grids, which suffer from "inaccurate measurement, inability to adjust, and inaccurate control."

[0079] The present invention mainly includes the following three technical points:

[0080] 1. Build a smart grid power optimization system based on real-time monitoring and dynamic prediction:

[0081] First, a high-frequency data acquisition system is used to perceive the state of the power grid in real time. New energy parameters are embedded in traditional node models to form a generalized power vector. A bidirectional flow discrimination model is established that considers net power difference, new energy output, energy storage charging and discharging, and local load to dynamically determine the direction of power flow. Secondly, ARIMA time series forecasting technology is used to generate accurate forecasts of new energy output and load demand for the next 24 hours through differential processing, autoregressive, and sliding average modeling.

[0082] 2. Based on multi-source data fusion and dynamic time-based benchmark value modeling technology, a new energy source and storage joint system control system featuring "historical data driven + real-time prediction and correction + multi-objective optimization" is constructed:

[0083] By generating time-based benchmark values, accurately modeling the energy storage SOC equation, and integrating three weighted rolling corrections—historical trends, weather forecasts, and grid dispatch instructions—multi-objective collaborative optimization is achieved, ultimately forming a stable power output solution that takes into account grid dispatch requirements, new energy consumption, and economic efficiency.

[0084] 3. Build a source-load-storage collaborative optimization mechanism based on bidirectional power flow and multi-objective dynamic optimization:

[0085] By establishing a power balance equation, a flexible interaction between renewable energy, energy storage, and the power grid is achieved. On the load side, an exponential moving average is used to generate a dynamic reference load, minimizing tracking deviation to improve demand response accuracy. The squared charge and discharge power term and the SOC variation penalty term are introduced on the energy storage side to extend battery life. On the grid side, a "buy low, sell high" strategy is optimized through a time-of-use electricity pricing model. A dynamic weight allocation algorithm is also proposed, adjusting target priorities in real time based on safety, economy, and regulatory requirements. Ultimately, the SOC operating range and response speed are constrained by a system stability function.

[0086] See also Figure 1 , is a flow chart of a specific embodiment of the present invention, comprising the following steps:

[0087] 1. ARIMA data forecasting and load elastic response based on real-time monitoring

[0088] 1.1 Real-time state perception and power direction dynamic judgment model

[0089] To ensure rapid response to renewable energy generation and grid status, a high-frequency data acquisition system is required. This system uses sensors and smart meters to implement real-time monitoring and obtain data such as the output power of distributed renewable energy.

[0090] Based on the above data collection, the new energy parameters are embedded in the traditional power grid node model to obtain the generalized node power vector, as shown in formula (1):

[0091]

[0092] Where, P gen The electric power generated by all new energy power generation units in the power grid; P load The power consumed by various users in the power grid, that is, the load demand of the power grid; P storage The energy storage system in the power grid can store or release excess electricity, and release it when demand peaks, thereby regulating the load on the power grid. EV The electric power involved in the charging and discharging of electric vehicles (EVs). This power represents the electricity consumed by EVs as user loads, or the electricity provided to the grid when EVs are discharged as virtual energy storage devices.

[0093] On this basis, combined with the net power and the impact of electric vehicles on the power grid, a power bidirectional flow formula is constructed to describe the power flow direction of the power grid, which helps to determine whether to inject power into the grid or absorb power from the grid, as shown in formula (2):

[0094] P net (t) = P gen (t)+P dis (t)-P ch (t)-P load (t) (2)

[0095] Where, P net P is the net power difference in the grid, usually the difference between all generated power and load power. This quantity reflects whether there is excess generated power in the grid or insufficient power generation. A positive value indicates that power is supplied to the grid, while a negative value indicates that power needs to be taken from the grid. disP is the power released by the energy storage system to the grid or local load; ch P is the power absorbed by the energy storage system from the grid or renewable energy generation; load It is the total power consumption of local loads, including rigid loads and transferable loads.

[0096] When the power grid has excess electricity, the excess electricity can be exported to other places; when the power grid is overloaded, external power can be dispatched in time or energy storage equipment can be enabled to prevent power grid overload, optimize the allocation of power resources, and ensure the stable operation of the power grid.

[0097] By building a multi-dimensional real-time perception system, we can accurately calculate power differences in the combined source and storage system, dynamically collect and pre-process grid operating status for bidirectional power balance control. By real-time monitoring of source output, power flow at the grid connection point, and load demand, we form a triple judgment mechanism, providing highly timely data support for subsequent dynamic baseline value correction and coordinated control.

[0098] 1.2 Short-term power forecasting based on ARIMA

[0099] The ARIMA (Autoregressive Integrated Moving Average) model is a commonly used time series forecasting method for renewable energy power generation and load demand. It is suitable for modeling and forecasting stationary time series data, especially those with trends or seasonal variations, and can effectively capture time series correlations.

[0100] An ARIMA model consists of three main components:

[0101] AR (Autoregressive): The autoregressive component represents the relationship between one observation and multiple lagged observations (i.e., the values ​​at the previous time point). This component describes the inertia or persistence of the data.

[0102] I (Integrated): The differencing part is to make the time series stationary by differencing it. A stationary time series means that its statistical properties do not change over time. Differencing helps remove trend and seasonality.

[0103] MA (Moving Average): The moving average part describes the relationship between the error of an observation and a lagged observation. It captures the model error by smoothing the noise in the data.

[0104] There are three main parameters of the ARIMA model:

[0105] p: The lag order of the autoregressive (AR) part, that is, how many historical data points are used to predict the current value.

[0106] d: The number of differencing operations used to convert a non-stationary time series into a stationary series. Usually 1 or 2 differencing operations.

[0107] q: The order of the moving average (MA) part, indicating how many lagged errors are used for modeling.

[0108] Steps to build an ARIMA model:

[0109] (1) Data collation: Before using the ARIMA model, it is necessary to preprocess and differentiate the renewable energy power generation, load, and electric vehicle charging curves to ensure the stationarity of the time series, as shown in formula (3):

[0110] Y t ′=Y t -Y t-1 (3)

[0111] Among them, Y t ′ is the time series after difference, Y t is the original time series.

[0112] If the data after differentiation is still unstable, multiple differentiation operations can be performed to ensure the stability of the time series.

[0113] (2) Setting the parameters of the ARIMA model: For each parameter (p, d, q), the optimal value can be determined by the autocorrelation function (ACF) and partial autocorrelation function (PACF).

[0114] (3) Establish an AR model. The AR model describes the relationship between the current value and its past value through the autoregressive term, as shown in formula (4):

[0115]

[0116] Among them, φ n is the coefficient of the AR part, Y t is the current value of the time series, ∈ t is the error term.

[0117] (4) Establishing MA model: MA model describes the prediction error of the current value by weighted average of past errors, as shown in formula (5):

[0118] Y t =θ1 t-1 +θ2 t-2 +…+θ q t-q +μ+ t (5)

[0119] Among them, θ n is the coefficient of the MA part, and μ is the constant term of the sequence.

[0120] (5) Establish the combined formula of the ARIMA model and fit the ARIMA model.

[0121] (6) Generate forecast values ​​for the next 24 hours based on the trained ARIMA forecast model.

[0122] By building the above-mentioned ARIMA model, accurate prediction of the short-term power of the source-storage combined system can be achieved, thereby providing data support for the prediction correction item in the dynamic benchmark value.

[0123] 1.3 Load-side demand elasticity response game

[0124] When determining the bidirectional flow of power in an active grid, load demand response is crucial. Based on their importance and priority, loads are categorized into two types: rigid loads (such as lighting) and shiftable loads (such as electric vehicle charging). These two types of loads differ significantly in their flexibility in responding to price fluctuations. Shiftable loads, in particular, allow users to choose to use their loads during periods of lower prices, thereby better aligning with grid demand.

[0125] (1) Load response mechanism

[0126] Based on the time-of-day electricity price, price elasticity modeling is performed to make an economic judgment on the load demand response. The utility function of the load is shown in formula (6):

[0127] U i =βln(1+α i )-λ(t)α i P shift,i (6)

[0128] In the formula, βln(1+α i ) is the electricity satisfaction, reflecting the diminishing marginal utility of users as electricity consumption increases. Simply put, as electricity consumption increases, the user's satisfaction will gradually decrease, indicating that users will be more cautious in their electricity choices; λ(t) is the time-dependent penalty coefficient, which is used to indicate the penalty intensity for subsequent items; α i P is the load transfer ratio, which indicates the ratio of the user's electricity consumption in a specific period to his original electricity consumption; shift,i It is the load transfer amount made by the load side based on the electricity price signal in electricity consumption management.

[0129] Load-end users hope to increase electricity consumption during periods of low electricity prices, thereby reducing electricity consumption during periods of high electricity prices, and then adjust their electricity consumption behavior according to real-time electricity prices to maximize their own utility.

[0130] (2) Nash equilibrium mediation

[0131] By differentiating the utility function, we can find the optimal response of the user. In the case of a mature market and rational user behavior, Nash equilibrium can guide load-side users to reasonably allocate loads, thereby achieving the effect of "peak shaving and valley filling". The solution is shown in Equation (7):

[0132]

[0133] Where, It is the optimal response that maximizes the load utility under the current given parameters; It is the negative impact that may be caused after the response occurs. The larger the value, the greater the negative impact that may be caused, which will lead to a lower optimal response value.

[0134] When electricity prices rise, users will voluntarily reduce their electricity consumption during high-price periods. This behavior helps smooth the load curve and reduces the stress on the power system during peak periods.

[0135] Through a demand response game model, we reveal users' positive attitudes towards electricity price signals and their impact on power system load regulation. By designing a reasonable utility function and solving Nash equilibrium problems, we can better understand and predict user electricity consumption behaviors, thereby achieving more efficient and economical power system operations.

[0136] 2. Initial benchmark value setting and dynamic adjustment of new energy source and storage combined system based on multi-source data fusion

[0137] 2.1 New Energy Source and Storage Combined System

[0138] See also Figure 2 , which is a schematic diagram of the source-storage combined system architecture of an embodiment of the present invention; the new energy source-storage combined system refers to the combination of new energy energy production and energy storage equipment, organically combining new energy power generation and energy storage to improve energy utilization efficiency, scheduling flexibility and system stability.

[0139] When constructing a combined source and storage system, the difference in power generation at the source end is an extremely important factor, and its calculation method is shown in formula (8):

[0140] ΔP(t)=P(t)-P ref (8)

[0141] Where P(t) is the actual output power of the new energy source at the current moment, which is affected by multiple real-time factors such as light intensity, temperature, and equipment efficiency; P ref It is the expected source output target value set by the power grid dispatching system or joint control strategy within the time period based on historical data, weather forecasts, power grid demand, etc., that is, the pre-set initial reference value; ΔP(t) is the power difference, which changes continuously with time and reflects the deviation between the actual output of the source and the plan.

[0142] When the actual power generation at the source exceeds the scheduling expectations, that is, ΔP(t)>0, the energy storage is charged and the excess energy is stored in the battery, thereby avoiding overvoltage or frequency increase in the grid. When the power generation at the source is lower than expected, that is, ΔP(t)<0, the energy storage is discharged, releasing the stored energy, thereby ensuring power supply reliability and preventing frequency drop.

[0143] Energy storage systems can charge when electricity prices are low and discharge when they are high, maximizing economic benefits. Furthermore, because the output of renewable energy sources (such as wind and solar) is unstable, energy storage systems can act as power regulators, smoothing out fluctuations in renewable energy and maintaining a stable net power output for the system.

[0144] In the construction of a new energy source-storage system, the safety of the energy storage system is an extremely important part. The safety of the energy storage device can be considered based on its state equation. The state equation of the energy storage system is an important indicator for evaluating safety. It is used to describe the changes in the state of charge of the energy storage and accurately describes how the state of charge of the energy storage device at different time points evolves as the charging and discharging actions change. Its calculation method is shown in formula (9):

[0145]

[0146] Where SOC(t) is the energy storage state of charge at time t, usually expressed as a percentage; η ch is the charging efficiency, which indicates the ratio of energy lost during the charging process; P ch (t) is the charging power at time t; Δt is the time step, which represents the time interval between each state update; P dis (t) is the discharge power at time t; η dis is the discharge efficiency, which also indicates the energy loss during the discharge process.

[0147] Based on this, a source-storage joint system is built to integrate and quantify the output of the two, forming a whole, and then conducting subsequent benchmark value calculations and rolling corrections.

[0148] 2.2 Generation of initial benchmark values ​​by time period

[0149] When calculating the benchmark value, first, based on historical data, the corresponding benchmark values ​​are set for the source and storage power generation and the grid dispatch instructions using the exponential moving average (EMA) and smoothing coefficient, as shown in formula (10):

[0150]

[0151] in, is the initial benchmark value obtained based on historical data combined with EMA; α t It is a smoothing coefficient, corresponding to different time periods, reflecting the characteristics of different time periods; EMA is the exponential moving average, which emphasizes the overall data trend, adapts to volatility, and improves the stability of prediction; The collected historical data is used as the pre-data for calculating the initial benchmark value, including but not limited to the source-storage combined system and power grid dispatch instructions.

[0152] This formula can be used to generate and set the time-based benchmark values ​​of the total output of the source-storage combined system and the initial benchmark values ​​for power grid dispatch.

[0153] Then, according to the different benchmark values ​​corresponding to different time periods, they are classified into four modes to achieve different scenarios, namely constant, ramp, step and sine, and then a typical benchmark value mode library is established as shown in Table 1:

[0154] Table 1

[0155]

[0156] During a constant period, the generated power remains stable and does not change significantly. For example, during the night valley period, the corresponding power generation baseline value should also be a stable value, and the power generation baseline value will remain stable during this period.

[0157] During the ramp period, the power generation gradually increases or decreases, reflecting the transition process during sunrise or sunset, and its power generation baseline value shows a stable change during this period.

[0158] Different constant power generation powers are set by referring to historical data within the step-type period, which is suitable for preset power generation plans in different time periods throughout the day and consists of multiple constant periods.

[0159] The sinusoidal period reflects the intraday periodic fluctuations that change with the external environment. It is used to simulate intraday periodic changes and smooth the transition to prevent jumps.

[0160] Based on the generation of the above initial benchmark value and the establishment of four different benchmark value modes, we can select an initial benchmark value and general trend item for each time period by analyzing the different characteristics of the time period, and then implement the next step of operations such as rolling correction.

[0161] 2.3 Rolling Correction of Dynamic Baseline Values ​​of Source-Reservoir Joint System Based on Multi-Source Data Fusion

[0162] When making actual benchmark value corrections, this paper uses the initial benchmark value as the basis and performs rolling corrections in combination with real-time grid dispatch instructions, load forecast results, and meteorological forecast data.

[0163] First, combine the ARIMA forecast model mentioned above to generate the short-term power meteorological forecast value P based on meteorological forecast ARIMA , and then generate forecast corrections.

[0164] Then, a certain amount of adjustment is made according to the real-time grid dispatch instruction. The grid dispatch offset calculation method is shown in formula (11):

[0165]

[0166] Where, P gridreq is the desired grid dispatch offset, which is used to reflect the adjustment requirements based on the initial reference value according to the real-time grid dispatch information; P grid,t Real-time dispatch instructions for the power grid.

[0167] The above calculation can effectively reflect the adjustment needs of the power grid, and then adjust the rolling correction of the final dynamic reference value.

[0168] Combining the above initial reference value, weather forecast value and grid demand value, the final reference value of rolling correction can be obtained, as shown in formula (12):

[0169]

[0170] Where, is the output value of the source-storage combined system obtained after correction; is a trend term, which is derived from historical data and is used to reflect long-term trends and reduce short-term random fluctuations; λ2·P ARIMA is the prediction correction term; λ3·P gridreq is the grid demand item, taking into account the changes in grid load and the real-time dispatch instructions of the grid; λ is the weight coefficient corresponding to each part, the sum of the three is 1, and the specific values ​​of the three are dynamically adjusted according to the feedback prediction error.

[0171] Through the weighted fusion of these three items, the smoothing of historical trends, the response to short-term forecasts and the compensation of scheduling instructions are achieved, forming a robust and reasonable benchmark value reference.

[0172] 3. Establish a coordinated optimization mechanism for sources, loads and storage

[0173] 3.1 Power balancing based on bidirectional flow in active grid

[0174] The active grid can not only receive power from renewable energy and energy storage, but also transmit power in the reverse direction. For example, the energy storage system can both charge and discharge, and the grid may also supply power to the grid or absorb excess power in some cases. Then, the corresponding objective function is constructed to achieve power balance, as shown in formula (13):

[0175]

[0176] in, is the actual output of new energy during period t; is the energy storage discharge power during period t; is the power purchased from the grid during period t; is the load demand after elastic adjustment during period t; is the energy storage charging power during period t; is the power sold to the grid during period t.

[0177] Based on the bidirectional flow characteristics of active power grids, the above function describes the instantaneous balance relationship of power in active power grids. The left side is the power supply source, and the right side is the power consumption destination, reflecting the bidirectional energy flow characteristics of active power grids.

[0178] 3.2 Multi-objective optimization function based on source-load-storage collaborative optimization

[0179] (1) Load end optimization

[0180] When optimizing the load side, it is necessary to first calculate a load forecast value and then generate a relatively stable reference load forecast by combining the peak load and historical load data, as shown in formula (14):

[0181] D ref,t =a·D peak +b·EMA(D hist,t ) (14)

[0182] Among them, D ref,t Represents the reference load value at time t, which serves as the target for load control and prediction. It is the standard for tracking actual load in the model; D peak The maximum load value during the period is a fixed value used to reflect the upper limit of the system capacity. It is usually calculated based on historical data; D hist,tA set of historical load data represents load records from a certain time period. It serves as an important basis for generating the reference load. EMA is the exponential moving average of historical load data. EMA smoothes historical data, reducing the impact of sudden short-term fluctuations and placing greater weight on recent values, making the reference load more flexible and responsive. The coefficients a and b represent the importance of maximum load and historical load in generating the reference load, respectively. a is the weight assigned to the maximum load value, emphasizing its practicality and importance, ensuring that the reference load does not exceed the system's capacity. b is the weight assigned to EMA, indicating the importance of historical data and allowing the reference load to be somewhat responsive to past demand fluctuations.

[0183] On this basis, the load tracking accuracy target is set to improve the load tracking accuracy by minimizing the absolute deviation between the actual adjusted load and the reference load within the specified time limit, as shown in formula (15):

[0184]

[0185] Where f is the objective function, which represents the total error measure of load tracking accuracy. The goal is to optimize the load adjustment to make it as close as possible to the reference value; T is the time range or total number of time steps, which determines the number of time nodes considered in the calculation; λ is a preset weight coefficient that reflects its importance in the objective function. If the value is high, it means that the load tracking accuracy requirements are more stringent and the model will pay more attention to it; D adj,t The actual load after adjustment. It reflects the actual load value required by the power system at that point in time, which may have been adjusted based on real-time data and demand response measures; D ref,t The reference load represents the ideal load value expected at time t. This value is typically calculated based on historical data and a forecasting model to provide a stable load tracking target. |·| represents the absolute value of the deviation, ensuring that the subtraction result is non-negative, reflecting the non-negative difference between the two load values.

[0186] The two sets of formulas above clearly illustrate how to set targets to optimize load tracking accuracy and establish a reasonable mechanism for generating reference loads. These formulas are designed to combine the system's maximum capacity with historical demand trends to achieve more accurate load forecasting and scheduling.

[0187] (2) Energy storage optimization

[0188] When optimizing the energy storage end, long-term use and excessively frequent charging and discharging may cause the performance of the energy storage device to deteriorate. Therefore, it is necessary to optimize the life loss cost of the energy storage device, as shown in formula (16):

[0189]

[0190] Where f is the objective function, which represents the total life loss cost of the energy storage system, and the goal is to minimize it; a is the coefficient of the square term of the charge and discharge power, and its unit is yuan / kW 2 This coefficient is used to measure the impact of charging and discharging power on energy storage equipment loss, reflecting the nonlinear relationship; P E,t represents the energy storage charge and discharge power at time t (discharge is positive). High power will lead to higher battery loss. b is the linear coefficient of SOC change, in yuan / %. It is used to penalize the change of the state of charge (SOC) of the energy storage battery and suppress drastic fluctuations to extend the battery life. t SOC is the state of the energy storage battery at time t, and its value is usually between 0% and 100%; t -SOC t-1 | is the absolute value of the difference between the SOC at time t and the previous moment. The introduction of this term is to prevent drastic changes in SOC and reduce the speed of battery aging.

[0191] The above equation reflects the nonlinear relationship between battery loss and charge / discharge power through the squared charge / discharge power term. This reduces the occurrence of high currents and avoids accelerating battery aging, which in turn increases costs. Furthermore, an SOC variation penalty term is added to suppress drastic SOC fluctuations, ensuring a smooth charge / discharge process, which helps extend the battery's cycle life.

[0192] (3) Grid-side optimization

[0193] Based on the principle of bidirectional energy flow in the active grid, the time-of-use electricity price strategy is used to maximize profits by using the "buy low, sell high" strategy. That is, electricity is purchased when the price is low and sold when the price is high, thereby obtaining economic benefits, as shown in formula (17):

[0194]

[0195] Where f is the economic cost of grid interaction. The goal is to minimize the difference between the cost of purchasing electricity and the revenue from selling electricity; C buy,t P is the electricity purchase price at time t, which changes with the time period to reflect the actual load of the power grid and the electricity price strategy. grid,buy is the actual amount of electricity purchased from the grid at time t; C sell,t P is the electricity price sold by the grid at that time; grid,sell is the amount of grid electricity actually sold at time t.

[0196] Clearly model the costs and benefits of purchasing and selling electricity separately to enable a clearer analysis of the economics and feasibility of grid interaction.

[0197] (4) Dynamic weight allocation optimization

[0198] In the construction of the above objective function, it is often necessary to select different weight coefficients according to the specific time period and then judge the importance of the factors in the objective function to achieve the optimization of the overall goal. The calculation is shown in formula (18):

[0199]

[0200] in, is the weight of the i-th optimization objective, the larger the value, the higher the priority; β is the sensitivity coefficient of weight allocation, the larger the value, the larger the weight difference; Score the urgency of the target in time period t, the higher the score, the lower the weight; R min The minimum score among all objectives, used for numerical stability.

[0201] The urgency score is mainly composed of three indicators, as shown in formula (19):

[0202]

[0203] in, It is a safety indicator, mainly determined by the SOC limit-exceeding risk; It is an economic indicator, which mainly depends on the deviation of electricity prices from the benchmark; To adjust the demand indicators, it mainly depends on the new energy forecast error.

[0204] Moreover, the optimal value of the sensitivity coefficient can be obtained by fitting historical data, as shown in formula (20):

[0205]

[0206] Where, are the standard deviation and mean of the weights, respectively, and the goal is to keep the weight dispersion around 30%.

[0207] 3.3 System stability optimization

[0208] On the basis of the above, it is also necessary to consider the optimization of the overall stability of the source-load-storage system in order to achieve system stability, as shown in formula (21):

[0209]

[0210] Where f is the total cost of system stability, and the goal is to minimize the deviation between the energy storage state and the target value, as well as the deviation of the response time; SOC t The current state of charge of the battery; SOC ref To consider the optimal operating midpoint of the charge and discharge capacity to ensure that the energy storage battery operates in a good working condition; |SOCt -SOC ref | is the absolute deviation between the SOC value at time t and the reference SOC value, which aims to keep the SOC as close to the ideal value as possible to maintain system stability; t r,t The deviation between the actual response time and the target time. This term is used to evaluate the system's response speed to new energy fluctuations.

[0211] By optimizing the SOC value and maintaining it in a sufficiently neutral range, the energy storage device can be prevented from being in an extremely high or low state for a long time. This helps reduce battery aging and failure, while ensuring that the energy storage system can respond quickly to load changes and fluctuations in renewable energy, thereby maintaining the stability of the power grid.

[0212] By combining these multiple objective functions, the design and optimization of a coordinated source-load-storage system not only improves economic efficiency but also prioritizes equipment longevity and grid stability. In practical applications, these objectives complement each other, collectively enhancing system resilience and adaptability. Through this optimization strategy, the grid and energy storage equipment can collaborate more effectively, improving overall power generation and consumption flexibility to adapt to changing market and environmental demands.

[0213] See also Figure 3 This is a modular flow chart of an embodiment of the present invention. Compared with the fragmented mode of independent optimization of source, load, and storage in traditional technology, the present invention realizes dynamic coupling of multiple links and global resource coordination by constructing a collaborative optimization framework. On the new energy side, relying on power prediction and dynamic constraint mechanisms, a source-storage joint system based on new energy is constructed to form an initial benchmark value and then flexibly respond to grid dispatching needs and weather changes, generating a final correction value with dynamic rolling correction, so that the source-storage joint system reduces sudden changes when connected to the grid and achieves stable control; the load side breaks through the limitations of passive response and activates the spatiotemporal transfer potential of adjustable resources based on the price elasticity model, making flexible loads such as air conditioners and electric vehicles active participants in grid balance; the energy storage system significantly extends its service life while smoothing power fluctuations through the collaborative optimization of charging and discharging strategies and battery health status, forming an operating mode of "no attenuation of adjustment capability and sustainable economy". This collaborative mechanism not only improves the adjustment accuracy of a single link, but also amplifies the overall value of system-level flexibility resources through global interaction.

[0214] At the same time, this solution, while reducing the cost of absorbing new energy and improving power supply reliability, uses the coordinated operation of sources, loads, and storage to make the power flow distribution more consistent with the physical characteristics of the power grid, reducing the risk of voltage over-limit caused by power imbalance, and essentially improving the stability of the system under the high proportion of new energy access. The technical framework adopts a modular design, which can not only independently adapt to the characteristics of the regional power grid, but also achieve interconnection and expansion through standardized interfaces. For example, in the future, it can seamlessly integrate new technologies such as hydrogen energy storage and vehicle-grid interaction, or be compatible with multiple mechanisms such as carbon quota trading and green certificate markets, providing an open and compatible technical foundation for the evolution of the energy Internet.

[0215] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should all be included in the scope of the claims of the present invention.

Claims

1. A source-load-storage control method for a new energy source in an active power grid that can output power in a sustainable and stable manner, characterized in that: The following steps are involved: S1. Autoregressive Integrated Moving Average Model (ARIMA) data forecasting and load elastic response based on real-time monitoring, including the following steps: S1.

1. By building a real-time state perception and dynamic power direction determination model and a multi-dimensional perception system, we can achieve power difference calculation, bidirectional power balance control, and dynamic grid status acquisition and processing for the combined source and storage system. We can also monitor source output, grid connection point power flow, and load demand in real time to establish a judgment mechanism. S1.

2. Construct an ARIMA forecasting model to predict the short-term power of the combined source and storage system, providing data support for the forecast correction term in the dynamic baseline value; S1.

3. By using time-of-day electricity prices and a load utility function with diminishing marginal utility, we can guide users to use more electricity during low-price periods and less during high-price periods to maximize utility. By using Nash equilibrium mediation to solve the optimal response strategy, we can enable users to voluntarily reduce their electricity consumption during high-price periods when electricity prices rise, thus smoothing the load curve and alleviating peak pressure on the power grid. S2. Setting and dynamically adjusting the initial benchmark value of the new energy source and storage combined system based on multi-source data fusion includes the following steps: S2.

1. Build a combined source-storage system by calculating the difference in power generated at the source and establishing the state equation of the energy storage system. S2.

2. Integrate historical source and storage output with dispatch instructions, use the exponential moving average algorithm and smoothing coefficient to set the baseline value, establish a model library to select initial values ​​and trend items, and support rolling corrections; S2.

3. Based on the initial baseline value, the weather forecast correction is generated through the ARIMA model, the grid dispatch offset is calculated, and after weighted integration of historical trends, short-term forecasts, and dispatch instructions, the corrected power generation and storage system output is output; S3. Build a source-load-storage collaborative optimization mechanism: S3.

1. Construct an objective function based on variables such as renewable energy output, energy storage charging and discharging, and electricity purchase and sales to achieve instantaneous power balance between power supply sources and power consumption destinations. S3.

2. Multi-objective optimization function based on source-load-storage collaborative optimization: A reference load prediction objective function is constructed by combining peak load and historical load data. The objective function is constructed to minimize the absolute deviation between the actual load adjustment and the reference load within a specified time limit. A lifetime loss cost objective function is constructed for the energy storage system to minimize the loss cost. A squared charge and discharge power term and a penalty term for state of charge (SOC) variation are introduced to suppress drastic SOC fluctuations and extend battery cycle life. A grid interaction economic cost objective function is constructed to minimize the difference between electricity purchase costs and electricity sales revenue. S3.

3. System stability optimization: Minimize the deviation between the energy storage state and the target value, as well as the deviation of the response time, by constructing a total cost objective function for system stability.

2. The source-load-storage control method for enabling sustainable and stable output of new energy power in an active power grid according to claim 1, characterized in that: The construction of a real-time state perception and power direction dynamic determination model in S1.1 specifically includes the following steps: Build a high-frequency data acquisition system to monitor distributed renewable energy output power and other data in real time through sensors and smart meters; Obtain data from a high-frequency data acquisition system, embed new energy parameters into the grid node model, and derive a generalized node power vector. A bidirectional power flow formula is constructed to describe the power flow direction of the power grid to determine whether to supply or withdraw power to the grid.

3. The source-load-storage control method for enabling sustainable and stable output of new energy power in an active power grid according to claim 1, characterized in that: The ARIMA forecasting model is constructed in S1.2, specifically including the following steps: Preprocess and differentiate the renewable energy power generation, load, and electric vehicle charging curves to ensure the stationarity of the time series; Use autocorrelograms and partial autocorrelograms to set the optimal values ​​of ARIMA model parameters; An autoregressive term is used to describe the relationship between the current value and its past value, and an autoregressive model is established; The weighted average of past errors is used to describe the forecast error of the current value and a sliding average model is established; Fit the ARIMA model to generate the short-term power forecast value of the source-storage combined system for the next 24 hours.

4. The source-load-storage control method for enabling sustainable and stable output of new energy power in an active power grid according to claim 1, characterized in that: The utility function of the load in S1.3 is expressed as follows: U i =βln(1+α i )-λ(t)α i P shift,i In the formula, βln(1+α i ) is the electricity satisfaction, reflecting the diminishing marginal utility of users when electricity consumption increases; λ(t) is the time-dependent penalty coefficient, which is used to indicate the penalty intensity for subsequent items; α i P is the load transfer ratio, which indicates the ratio of the user's electricity consumption in a specific period to his original electricity consumption; shift,i The load transfer amount made by the load end based on the electricity price signal in the power consumption management; The solution to the Nash equilibrium mediation in S1.3 is expressed as the following formula: Where, It is the optimal response that maximizes the load utility under the current given parameters; It is the negative impact that may be caused after the response occurs. The larger the value, the greater the negative impact that may be caused, which will lead to a lower optimal response value.

5. The source-load-storage control method for enabling sustainable and stable output of new energy power in an active power grid according to claim 1, characterized in that: The difference in power generated by the source end in S2.1 is expressed as follows: ΔP(t)=P(t)-P ref Where, P(t) is the actual output power of the new energy source at the current moment; P ref It is the initial reference value pre-set by the power grid dispatching system or joint control strategy within the time period; ΔP(t) is the power difference, reflecting the deviation between the actual power output of the source end and the plan. If ΔP(t)>0, the energy storage is charged, indicating that the actual power generation of the source end exceeds the scheduling expectations. If ΔP(t)<0, the energy storage is discharged, indicating that the power generation of the source end is lower than expected.

6. The source-load-storage control method for enabling sustainable and stable output of new energy power in an active power grid according to claim 1, characterized in that: The state equation of the energy storage system in S2.1 is expressed as the following formula: Where SOC(t) is the energy storage state of charge at time t, expressed as a percentage; η ch is the charging efficiency, which indicates the ratio of energy lost during the charging process; P ch (t) is the charging power at time t; Δt is the time step, which represents the time interval between each state update; P dis (t) is the discharge power at time t; η dis is the discharge efficiency, which indicates the energy loss during the discharge process.

7. The source-load-storage control method for enabling sustainable and stable output of new energy power in an active power grid according to claim 1, characterized in that: The S2.2 specifically includes the following steps: Based on historical data, the power generation output of sources and storage is integrated with grid dispatch instructions, and the corresponding benchmark value is set through the exponential moving average algorithm and smoothing coefficient; Set the time-based benchmark values ​​for the total output of the combined power generation and storage system and the initial benchmark values ​​for power grid dispatch; Establish a typical benchmark value pattern library based on time period characteristics and usage scenarios, and select initial benchmark values ​​and trend items for each time period to support subsequent rolling correction operations; Based on real-time grid dispatch instructions, calculate grid dispatch offset and quantify grid dynamic adjustment requirements; The initial benchmark value, meteorological forecast value and power grid demand value are weighted and integrated, and the corrected power generation and storage system output value is output through historical trend smoothing, short-term forecast response and dispatch instruction compensation.

8. The source-load-storage control method for enabling sustainable and stable output of new energy power in an active power grid according to claim 1, characterized in that: The S2.3 specifically includes the following steps: Based on the initial benchmark value, the real-time dispatch instructions of the power grid, load forecast results and weather forecast data are integrated as correction inputs; The meteorological data is processed through the ARIMA forecast model to generate short-term power meteorological forecast values, based on which the baseline value correction at the forecast level is calculated.

9. The source-load-storage control method for enabling sustainable and stable output of new energy power in an active power grid according to claim 1, characterized in that: The objective function in S3.2 selects different weight coefficients according to the specific time period, which can be expressed as the following formula: in, is the weight of the i-th optimization objective, the larger the value, the higher the priority; β is the sensitivity coefficient of weight allocation, the larger the value, the larger the weight difference; Score the urgency of the target in time period t, the higher the score, the lower the weight; R min The minimum score among all objectives, used for numerical stability guarantee; The urgency score is mainly composed of three indicators, which can be expressed as the following formula: in, It is a safety indicator, indicating the risk of SOC exceeding the limit; It is an economic indicator that indicates the deviation of electricity prices from the benchmark; To adjust the demand index, it represents the new energy forecast error.

10. The source-load-storage control method for enabling sustainable and stable output of new energy power in an active power grid according to claim 2, characterized in that: The generalized node power vector It is expressed as the following formula: Where, P gen The electric power generated by all new energy power generation units in the power grid; P load The power consumed by various users in the power grid, that is, the load demand of the power grid; P storage The electricity stored or released by the energy storage system in the power grid; EV The electrical power involved in charging and discharging electric vehicles; The power bidirectional flow formula of the power flow direction of the power grid is expressed as the following formula: P net (t)=P gen (t)+P dis (t)-P ch (t)-P load (t) Where, P net is the net power difference in the grid, a positive value indicates power is supplied to the grid, and a negative value indicates power is taken from the grid; P dis P is the power released by the energy storage system to the grid or local load; ch The power absorbed by the energy storage system from the grid or renewable energy generation; P load It is the total power consumption of local loads, including rigid loads and transferable loads.

Images