Wind power-photovoltaic-photo-thermal new energy base bundling delivery optimization scheduling method

Abstract

The invention discloses a wind power-photovoltaic-photo-thermal new energy base bundling delivery optimization scheduling method. The method comprises the steps that historical environment data and prediction data of a wind power-photovoltaic-photo-thermal new energy base are acquired; taking the total power grid stability, the power generation cost and the energy scheduling penalty of the new energy base as objective functions, considering the operation characteristic constraint, the power balance constraint and the power transmission line capacity constraint of wind power, photovoltaic and photo-thermal power generation, and constructing a combined power generation optimal scheduling model; and solving by adopting improved mixed integer linear programming, performing real-time scheduling control on each new energy base according to a solving result, and determining an optimal delivery strategy. Through cooperative scheduling of wind power, photovoltaic, photo-thermal and other energy systems, power output in the power generation and delivery process is optimized, the power grid load matching capacity is improved, volatility and scheduling conflicts in the power system scheduling process are reduced, efficient utilization of new energy and maximized economic benefits are achieved, and the method is suitable for large-scale popularization and application. And the stability and economy of the power grid are ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of power dispatching, and in particular relates to a method for optimizing the dispatching of bundled transmission of wind power-photovoltaic-solar thermal new energy bases. Background Art

[0002] As the global energy transition accelerates, renewable energy sources, such as wind power, photovoltaics, and solar thermal, are increasingly contributing to the power system. To fully tap the potential of renewable energy, large-scale renewable energy base construction is in full swing. However, the intermittent and volatile nature of renewable energy poses significant challenges to the stable operation of the power grid. How to effectively integrate large-scale, widely distributed renewable energy bases into the power system and achieve safe, economical, and efficient power transmission has become a hot topic in current power system research.

[0003] Optimal dispatching of bundled transmission from renewable energy bases, a key technology for addressing the aforementioned issues in the power system sector, has garnered widespread attention. Its core objective is to rationally coordinate large-scale renewable energy generation with complex power grids, achieving efficient utilization of renewable energy and stable grid operation. However, this field still faces numerous challenges. Random variations in natural conditions such as wind speed and sunshine lead to a high degree of uncertainty in renewable energy output, complicating dispatching decisions. With the continuous expansion of renewable energy integration and the increasing complexity of the grid structure, traditional dispatching methods are struggling to adapt. During the dispatching optimization process, objectives such as economic efficiency, reliability, and environmental friendliness often conflict, making finding the optimal balance a complex problem. Power systems operate in real time, and dispatching decisions must rapidly respond to changes in grid conditions, placing high demands on the computational efficiency of algorithms. The large scale of renewable energy bases, coupled with the numerous variables and constraints involved, poses challenges to computing resources. These issues not only impact the integration of renewable energy but also the safe and stable operation of the entire power system. In-depth research on optimal dispatching strategies for bundled transmission from renewable energy bases is crucial for improving renewable energy utilization efficiency, reducing power system operating costs, and promoting energy transformation. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for optimizing the dispatching of bundled transmission of wind power-photovoltaic-solar thermal energy bases. By coordinating the dispatching of wind power, photovoltaic and solar thermal energy systems, the power output during power generation and transmission is optimized, the load matching capability of the power grid is improved, and the volatility and dispatching conflicts during the power system dispatching process are reduced, thereby achieving efficient utilization of new energy and maximizing economic benefits, and ensuring the stability and economy of the power grid.

[0005] In order to achieve the above object, the solution of the present invention is:

[0006] A method for optimizing the dispatch of bundled transmission of wind power, photovoltaic power, and solar thermal power generation bases includes the following steps:

[0007] Step 1: Obtain historical environmental data for the wind power, photovoltaic, and solar thermal energy bases, and obtain forecast data for wind speed, light intensity, and solar irradiance over the next period of time, as well as forecast data for wind power, photovoltaic power generation, and heat storage and release conditions of solar thermal power stations.

[0008] Step 2: Taking the overall grid stability, power generation cost, and energy dispatch penalty of the new energy base as the objective function, and considering the operating characteristics constraints of wind power, photovoltaic power generation, and solar thermal power generation, power balance constraints, and transmission line capacity constraints, a joint power generation optimization dispatch model is constructed;

[0009] Step 3: Use improved mixed integer linear programming to solve the joint power generation optimization scheduling model to obtain the optimal power generation plan of each renewable energy source in each scheduling period. Based on the solution results, each renewable energy base is dispatched and controlled in real time. The scheduling strategy is adjusted in real time according to the actual situation to determine the optimal transmission strategy.

[0010] In step 1 above, historical meteorological data is collected and real-time meteorological data is obtained. A prediction model is established using machine learning and long-short-term memory networks to predict wind speed, light intensity, solar irradiance, wind power, photovoltaic power generation, and heat storage and release of solar thermal power stations over a period of time in the future.

[0011] In the above step 2, the objective function of the joint power generation optimization scheduling model is:

[0012]

[0013] Where: T represents the scheduling time period, α1 is the weight coefficient of the load error term, α2 is the weight coefficient of the scheduling priority of the power generation cost, and α3 is the penalty coefficient; P load (t) is the power generation demanded by the grid load, P wind (t) is the wind power generation, P pv (t) is the photovoltaic power generation, P csp (t) is the amount of solar thermal power generation; C wind is the power generation cost of the wind farm, C pv is the power generation cost of the photovoltaic power station, C csp is the power generation cost of the CSP station, C penalty Penalty costs for curtailing wind and solar power.

[0014] In step 2 above, the constraints of the joint generation optimization scheduling model include:

[0015] Power balance constraints:

[0016] P wind (t)+Ppv (t)+P csp (t) = P load (t)+P loss (t)

[0017] Among them, P load (t) is the load demand power of the power system at time t, P loss (t) is the power loss of the transmission line at time t;

[0018] Grid transmission constraints:

[0019] P out (t)≤P max ,

[0020] Among them, P out (t) is the power output of the power grid, P max is the maximum transmission capacity of the power grid;

[0021] Thermal storage system constraints:

[0022] T min ≤T storage (t)≤T max

[0023] Among them, T min and T max are the minimum and maximum temperatures of the heat storage system, respectively;

[0024]

[0025]

[0026] Among them, P heat (t) is the power of the external heating heat storage system, P hrelease (t) is the power of the heat storage system supplying heat to the outside world;

[0027] Restrictions on wind power, photovoltaic power generation, and solar thermal power generation:

[0028] P wind (t)≤P wind,max (t)

[0029] P PV (t)≤P PV,max (t)

[0030] P CSP (t)≤P CSP,max (t)

[0031] Constraints on the thermodynamic characteristics of the photothermal system:

[0032] E store (t+1)=E store(t)+η collector Q collector (t)-E loss,store (t)-E release (t)

[0033] Among them, η collector is the efficiency of the solar thermal collector, Q collector (t) is the heat collected by the solar thermal collector at time t, E loss,store (t) is the heat loss of the heat storage system at time t.

[0034] In the above step 3, the improved mixed integer linear programming is used to solve the joint power generation optimization scheduling model, and the form of the mixed integer linear programming problem is set as:

[0035] Objective function: z = c T x+θ

[0036] Constraints: Ax≤b,A eq x=b eq ,x∈X

[0037] Where x is the decision variable vector, c is the objective function coefficient vector, A is the inequality constraint coefficient matrix, b is the inequality constraint right-hand side term vector, and A eq is the equality constraint coefficient matrix, b eq is the vector of terms on the right side of the equality constraint, and S is the domain of the variable;

[0038] The mixed integer linear programming problem is decomposed into a main problem and multiple sub-problems using Benders decomposition method;

[0039] Assume that the mixed integer linear programming problem is min{c T x+d T y:Ax+By≤b,x∈X,y∈Y}, where x is an integer variable and y is a continuous variable; the main problem is an integer programming problem, which only considers the integer variable y, and the objective function is z=c T x+θ, the constraints are Ax≤b and x∈X, where θ is a variable to be determined; the subproblem is a linear programming problem, for a given x, solve θ(x)=min{d T y:By≤b-Ax,y∈Y};

[0040] During the solution process, the main problem and sub-problems are solved alternately. The solution of the main problem provides constraints for the sub-problem, and the solution of the sub-problem is fed back to the main problem to update the objective function.

[0041] In step 3 above, a greedy algorithm is first used to obtain an initial solution. Starting from the current solution, a better solution is searched within the defined neighborhood. If a better solution is found, the current solution is updated and the search continues. If no better solution is found after a certain number of iterations, the search is terminated.

[0042] When the convergence conditions are met, the optimal scheduling plan is output, that is, the optimal power generation plan of each renewable energy source in each scheduling period.

[0043] In step 3 above, based on the obtained optimal power generation plan, the wind power base, photovoltaic base and solar thermal power station are dispatched and controlled in real time through the power dispatch control system, and the power generation situation of the new energy base and the operating status of the power system are monitored in real time. When the actual situation deviates greatly from the predicted results or the system operating parameters exceed the constraint range, the dispatch plan is adjusted and optimized.

[0044] In step 3 above, distributed parallel computing is used to increase the solution speed. This means that the problem is broken down into multiple subproblems, which are solved in parallel on multiple processors. The parallel computing capabilities of the GPU are used to accelerate linear algebra operations. CPLEX or Gurobi tools are used for the solution.

[0045] The above method also includes step 4, obtaining the operating status of the new energy base in real time and evaluating the transmission strategy. The evaluation indicators include the new energy consumption rate, the stability of the transmitted power, and the power generation cost. The acquisition process of the transmission strategy is optimized and adjusted according to the evaluation results.

[0046] After adopting the above scheme, the present invention predicts power load based on machine learning and LSTM model, and establishes a multi-objective optimization scheduling model through multi-energy complementary scheduling algorithm. Through the improved mixed integer linear programming method and distributed parallel computing, real-time scheduling and dynamic adjustment are achieved to ensure the load balance and stable operation of the power grid. The application of real-time dynamic adjustment and feedback mechanism effectively improves the response speed and flexibility of the system, and can cope with sudden changes in power demand and fluctuations in natural conditions. This strategy can dynamically adjust the matching degree between new energy output and load according to the actual power grid operation status, maximize the new energy absorption rate, and improve the stability and economy of the transmitted power. It is particularly suitable for dealing with complex working conditions with large fluctuations in new energy output and high response requirements.

[0047] The present invention has the following advantages:

[0048] This paper proposes an optimization strategy based on the combined dispatch of wind power, photovoltaic power, and solar thermal power. This strategy aims to address the uncertainty of renewable energy generation, improve the level of renewable energy absorption, reduce power generation costs, ensure the safe and stable operation of the power system and the quality of transmitted power, and promote the efficient utilization and sustainable development of renewable energy. This strategy fully utilizes the complementary characteristics of wind power, photovoltaic power, and solar thermal power to reduce grid instability caused by fluctuations in a single energy source and improve the economic benefits of the system. Specifically, this strategy achieves optimization through the following aspects:

[0049] Through the coordinated scheduling of wind power, photovoltaics, and solar thermal energy, excess wind and photovoltaic power is converted into thermal energy storage in the solar thermal system. This effectively reduces the waste of fluctuating energy sources such as wind power and photovoltaics, improves the utilization rate of renewable energy generation, reduces wind, photovoltaic, and thermal curtailment, maximizes the utilization of renewable energy, and enhances the regulation capability of solar thermal systems within the power grid. A mixed-integer linear programming model, combined with real-time load forecasting and dynamic scheduling, enables precise scheduling and optimization, accurately matching grid load demand, ensuring stable grid operation, reducing system fluctuations, and ensuring load balance and efficient utilization of renewable energy. The application of real-time dynamic adjustment and feedback mechanisms effectively improves the system's responsiveness and flexibility, enabling it to cope with sudden changes in power demand and fluctuating natural conditions. Compensating for the fluctuations in wind and photovoltaic power generation through scheduling of the solar thermal system addresses the volatility of wind and photovoltaic power generation, enhancing system reliability and controllability. Optimizing scheduling algorithms reduces manual intervention and unnecessary adjustments in power system scheduling, reducing scheduling costs and improving system efficiency. This method is highly scalable and applicable to large-scale wind power, photovoltaic, and solar thermal power plants of varying sizes. System parameters can be adjusted based on actual needs, promoting the efficient utilization and sustainable development of renewable energy. By seamlessly integrating with existing power grid dispatch systems, it enables real-time optimized scheduling of renewable energy output, improving grid operational efficiency and stability. By coordinating the scheduling of different types of renewable energy generation equipment, it is possible to maximize the proportion of renewable energy utilization while ensuring power supply, reducing fossil energy consumption and minimizing environmental pollution.

[0050] The wind power-photovoltaic-solar thermal energy base bundled transmission optimization scheduling strategy provided by the present invention can not only solve the uncertainty problem of renewable energy power generation, but also improve the operational safety and economic benefits of the power grid, effectively improve the stability of renewable energy power generation and the load matching capability of the power grid, reduce the problems caused by volatility and uncertainty, and improve the scheduling efficiency and economic benefits of the power system. It has important practical significance for promoting the widespread application of new energy and promoting the transformation of energy structure, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0052] Figure 2 It is a real-time dynamic adjustment process timing diagram;

[0053] Figure 3 It is a time series diagram of the energy-based multi-energy multi-objective joint scheduling model;

[0054] Figure 4 It is a schematic diagram of electrical-thermal coupling;

[0055] Figure 5 It is a real-time monitoring system architecture and data flow diagram. DETAILED DESCRIPTION

[0056] The technical solutions and beneficial effects of the present invention will be described in detail below with reference to the accompanying drawings.

[0057] like Figure 1 As shown, the present invention provides a method for optimizing the dispatch of bundled transmission of wind power, photovoltaic power and solar thermal power new energy bases, comprising the following steps:

[0058] Step 1: Data Collection and Power Forecasting: For wind power, photovoltaic power, and solar thermal power, historical environmental data (including but not limited to temperature, humidity, air pressure, wind speed, and light intensity) from the new energy bases are collected. A prediction model is then established using machine learning and long-short-term memory (LSTM) networks combined with historical and real-time meteorological data to predict future wind speed, light intensity, and solar irradiance. This model also predicts the power generation of each energy source and the heat storage and release of the solar thermal power station. Data related to power transmission channels and receiving power grids is also obtained. In particular, the new energy bases are equipped with thermal storage electric boilers for converting electricity into heat and storing heat.

[0059] In step 1, historical wind speed data for wind power, photovoltaic power, and solar thermal power plants are collected. These data are used to develop power generation characteristic models for each of the three renewable energy sources: wind power, photovoltaic power, and solar thermal power plants. These data include, but are not limited to, the impact of environmental factors such as wind speed, sunshine, temperature, and humidity on power generation. A prediction model is then developed to forecast wind speed, sun intensity, and solar irradiance for the future. Combining historical and real-time meteorological data, machine learning and data-driven approaches are employed, using a long-short-term memory (LSTM) network, to generate short-term and medium- to long-term forecasts for wind power, photovoltaic power generation, solar thermal power plant power, and the heat storage and release characteristics of the solar thermal system. The forecast period can be flexibly set based on actual scheduling needs. Short-term forecasts can be for the next 1-6 hours to meet the rapid response requirements of real-time scheduling; medium- to long-term forecasts are for the next 1-7 days, enabling the development of longer-term power generation plans and resource allocation strategies, accurately estimating the power generation volatility of each renewable energy source. Furthermore, relevant data such as the transmission capacity limitations of power transmission channels and the load demands of receiving power grids are collected to inform subsequent scheduling decisions.

[0060] These new energy base systems are equipped with thermal storage electric boilers, which convert electricity into heat, balancing direct heating and heat storage capabilities. During periods of grid downturn, the electric boilers can convert wind and photovoltaic power into thermal energy for storage in the solar thermal system.

[0061] The following main factors are considered when setting up a joint power generation dispatch model for three energy sources: wind power, photovoltaic power, and solar thermal power:

[0062] Wind power generation P wind (t), its value is affected by the change of wind speed at a certain time point t and is expressed as uncertainty.

[0063] Photovoltaic power generation P pv (t), whose value at a certain time point t is affected by factors such as solar radiation intensity and efficiency of photovoltaic panels.

[0064] Solar thermal power generation P csp (t), its power generation has certain adjustability under sunshine conditions and can be adjusted within a certain range through the thermal energy storage system.

[0065] The power generation models of wind power, photovoltaic power and solar thermal power are:

[0066] Wind power generation model: P wind (t) = f wind (v(t))

[0067] Among them, f wind It is the function between wind speed and wind turbine power generation, and v(t) is the wind speed data at time t.

[0068] Photovoltaic power generation model: P pv (t) = η pv ·G(t)·A

[0069] Among them, η pv is the efficiency of the photovoltaic panel, G(t) is the solar radiation intensity at time t, and A is the effective area of ​​the photovoltaic module.

[0070] Solar thermal power generation model: P csp (t) = η csp ·G(t)·A csp +ΔP csp

[0071] Among them, η csp is the efficiency of the solar thermal system, A csp is the effective area of ​​the CSP system, ΔP csp The power generation fluctuations caused by the regulation of the energy storage system, that is, the power generation caused by the additional thermal energy.

[0072] Monitor the grid's load demand in real time and forecast future load demand based on historical load data, seasonal variations, and external factors. Analyze the grid's load fluctuations and trends based on load demand forecasts, providing a basis for developing dispatch strategies.

[0073] Step 2: Construct a joint power generation optimization scheduling model: Taking the total grid stability, power generation cost and energy scheduling penalty of the new energy base as the objective function, a multi-energy joint scheduling model is established considering the operating characteristics constraints of wind power, photovoltaic power generation, and solar thermal power generation, power balance constraints, and transmission line capacity constraints. The model covers the output characteristics, power generation capacity, load demand and scheduling constraints of each energy source. The optimization objective function includes the load error term, the total power generation cost of the new energy base, and the energy scheduling penalty term. The cost calculation formulas and the wind and solar curtailment penalty coefficients are clarified. The utilization and heat storage control strategies of wind power and photovoltaic over-generation are given. The transmission power and strategy are determined based on the scheduling optimization results and grid demand. By adjusting the relevant parameters, the grid load demand is balanced, the energy curtailment waste is reduced, and the economy is ensured.

[0074] In step 2, a multi-energy joint scheduling model is established, taking the overall grid stability, power generation costs, and energy scheduling penalties of the new energy base as objective functions, and considering the operational characteristics constraints of wind power, photovoltaic power generation, and solar thermal power generation, power balance constraints, and transmission line capacity constraints. The model considers the output characteristics, power generation capacity, load demand, and scheduling constraints of wind power, photovoltaic power generation, and solar thermal power generation. The power generation costs include the equipment operation and maintenance costs of wind power, photovoltaic power generation, and solar thermal power generation, as well as the backup costs caused by power generation fluctuations. The operational characteristics constraints include the relationship between the upper and lower output limits of wind power and wind speed, the relationship between the upper and lower output limits of photovoltaic power generation and light intensity, and the heat storage and heat release characteristics of the heat storage device in the solar thermal power generation system.

[0075] By modeling the input-output relationships of various energy systems and the time-varying characteristics of resources, and comprehensively considering factors such as power generation volatility, dispatch frequency, response time, energy storage systems, and grid load, a multi-objective optimization problem is constructed. A data-driven real-time dispatch approach is employed to acquire grid load, meteorological data, and power generation forecast information in real time. Using efficient computational models, real-time dispatch is performed to adjust the output of wind, photovoltaic, and solar thermal power generation methods, ensuring energy supply stability and grid load balance.

[0076] To further optimize the dispatch strategy and avoid wasting wind, photovoltaic, and solar thermal energy, penalty terms for curtailed wind, photovoltaic, and solar thermal power are added to the optimization objective function. This will encourage the dispatch strategy to prioritize the use of available renewable energy while reducing reliance on traditional energy sources.

[0077] In the case of multi-energy complementarity, the joint scheduling problem of wind power, photovoltaic and solar thermal systems can be expressed as a multi-objective optimization problem. The specific scheduling objectives are:

[0078]

[0079] The objective function consists of three parts:

[0080] Load error term: It is used to measure the deviation between the grid load demand and the power generation output of each energy source, aiming to minimize the load matching error.

[0081] Total power generation cost of new energy base: including wind power, solar thermal and solar thermal power generation costs

[0082] Energy dispatch penalty: Considers dispatch failures due to generation limitations or volatility of different energy sources, including abandonment of wind and solar power.

[0083] in:

[0084] T represents the scheduling time period (which can be one day or one week).

[0085] α1 is the weight coefficient of the load error term, reflecting the priority of load balancing.

[0086] α2 is the weight coefficient of the dispatch priority of the generation cost, reflecting the trade-off between load balancing and resource optimization.

[0087] α3 is the penalty coefficient, which is specifically weighted for wind and power curtailment, aiming to minimize the waste of these energies. A larger penalty coefficient will more strongly encourage the system to avoid wasting available energy.

[0088] C wind is the power generation cost of the wind farm, including wind turbine operation and maintenance costs, which can be expressed as where c wind is the operation and maintenance cost coefficient of unit wind power.

[0089] C pv is the power generation cost of the photovoltaic power station, c pv is the operation and maintenance cost coefficient per unit photovoltaic power.

[0090] C csp is the power generation cost of the CSP station, where c csp is the unit CSP power cost coefficient, c store is the unit heat storage cost coefficient, E store (t) is the heat storage at time t, c release is the unit heat release benefit coefficient (considering the energy value recovery of the solar thermal system when it generates heat for power generation), E release(t) is the heat release at time t.

[0091] C penalty Penalty costs for curtailing wind and solar power, where c wtdump is the penalty cost coefficient for unit wind power abandonment, P wtdump is the wind power curtailment, c pwdump is the penalty cost coefficient for unit abandoned optical power, P pvdump The discarded optical power.

[0092] Wind power curtailment occurs when wind power generation exceeds grid demand or grid transmission capacity. Curtailment is defined as:

[0093] P wtdump (t)=max(0,P wind (t)-P wind,consume (t))

[0094] Photovoltaic curtailment occurs when photovoltaic power generation exceeds grid demand or grid transmission capacity. The calculation method for curtailment is similar to that for wind power:

[0095] P pvdump (t)=max(0,P pv (t)-P pv,consume (t))

[0096] α3 is the penalty coefficient for wind and solar curtailment, controlling the severity of the penalty. Its selection is based on the system's economic objectives and environmental benefits. If the penalty for wind and solar curtailment is too low, it may result in significant curtailment; conversely, if the penalty coefficient is too high, it may affect the system's economic viability. Therefore, it needs to be adjusted based on specific circumstances.

[0097] When wind and photovoltaic power generation exceeds grid demand or external power, it can be directly used to heat the thermal storage medium of the solar thermal system, such as thermal oil or salt solution. The specific heating control strategy is as follows:

[0098] Assume that the temperature of the heat storage medium of the solar thermal system is T storage (t), the wind power and photovoltaic power generation used to heat the thermal storage medium are P h wind (t) and P h PV (t), the heating power of the heat storage medium is P heat (t), the heat storage heating process can be expressed as:

[0099] P heat (t) = η heat ·(P h wind (t)+P h PV(t)),

[0100] Among them, η heat is the conversion efficiency of the heating process.

[0101] The temperature change of the heat storage medium is determined by its heat capacity C storage and heating power P heat (t) determine:

[0102]

[0103] Heat storage medium temperature dT storage The change of (t) is related to the time integral of the heating power. In order to avoid overheating and waste, the temperature of the thermal storage system should be kept within a suitable range to ensure its effective energy storage.

[0104] Penalties are designed to encourage dispatch strategies to reduce wind, solar, and thermal power curtailment. If a particular energy source's generation capacity is underutilized (i.e., below maximum output), a penalty cost is incurred. By incorporating penalty factors, dispatch strategies are biased toward minimizing the waste of renewable energy, particularly in light of fluctuations in uncontrollable energy sources like wind and solar.

[0105] Finally, the amount of electricity to be transmitted is determined based on the dispatch optimization results and grid demand. The transmission strategy is dynamically adjusted based on the grid's dispatch requirements and electricity market prices. By dynamically optimizing the amount of electricity to be transmitted, the transmission strategy can be adjusted based on market demand, minimizing economic losses during the transmission process.

[0106] By adjusting parameters such as α1, α2, and α3, the dispatch strategy can find a balance between meeting grid load demand, reducing energy waste, and ensuring economic efficiency. For example, if grid load demand is too high, it may prioritize relying on adjustable solar thermal systems; if load demand is low and renewable energy is abundant, the system will prioritize using wind power and photovoltaics.

[0107] In actual operation, when using the improved mixed integer linear programming (MILP) to solve this optimization problem, the solver will automatically adjust the power generation and scheduling strategies according to the given penalty coefficient to avoid excessive energy abandonment and improve the overall economy and environmental performance of the system.

[0108] The constraints are as follows:

[0109] Power balance constraints:

[0110] P wind (t)+P pv (t)+P csp (t) = P load (t)+P loss (t)

[0111] That is, the load demand of the power grid should be met by the total amount of power generation from different energy sources.

[0112] Among them, P load (t) is the load demand power of the power system at time t, P loss (t) is the power loss of the transmission line at time t, which can be calculated based on the line parameters and transmission power.

[0113] Grid transmission constraints:

[0114] P out (t)≤P max ,

[0115] Among them, P out (t) is the power output of the power grid, P max is the maximum transmission capacity of the power grid.

[0116] Thermal storage system constraints:

[0117] T min ≤T storage (t)≤T max

[0118] Among them, T min and T max are the minimum and maximum temperatures of the heat storage system, respectively.

[0119]

[0120]

[0121] Among them, P heat (t) is the power of the external heating heat storage system, P hrelease (t) is the power of the heat storage system supplying heat to the outside world.

[0122] Restrictions on wind power, photovoltaic power generation, and solar thermal power generation:

[0123] P wind (t)≤P wind,max (t)

[0124] P PV (t)≤P PV,max (t)

[0125] P CSP (t)≤P CSP,max (t)

[0126] Constraints on the thermodynamic characteristics of the photothermal system:

[0127] The heat storage and release process of the solar thermal system follows the law of conservation of energy. The heat storage process can be expressed as:

[0128] Estore (t+1)=E store (t)+η collector Q collector (t)-E loss,store (t)-E release (t)

[0129] where η collector is the efficiency of the solar thermal collector, Q collector (t) is the heat collected by the solar thermal collector at time t, E loss,store (t) is the heat loss of the thermal storage system at time t, which can be calculated based on factors such as the thermal conductivity characteristics of the thermal storage system and the ambient temperature. There is a certain conversion relationship between the power generated by solar thermal power generation and the amount of heat released.

[0130] Based on real-time data and forecast results, a dynamic scheduling algorithm is used to adjust the output power of the three renewable energy sources to address grid load fluctuations. During the actual scheduling process, optimal power resource allocation is achieved by controlling scheduling priorities and response times based on the real-time power generation and load demand of wind power, photovoltaic power, and solar thermal power. Combined with the energy storage system's scheduling strategy, the adjustability of the solar thermal system is fully utilized to supplement the insufficient supply of more volatile energy sources such as wind power and photovoltaic power, thereby improving the reliability of renewable energy and the stability of power supply.

[0131] Step 3: Model solution and scheduling implementation: The constructed optimization scheduling model is solved using an improved mixed integer linear programming (MILP) to obtain the optimal power generation plan for each energy source in each scheduling period. The solution process fully considers the complementary characteristics of different energy sources, and performs real-time scheduling control on each base based on the solution results. The actual parameters are compared with the predicted values ​​in real time, and large deviations are adjusted in time. For large-scale MILP problems, distributed parallel computing is used to improve the solution speed. In combination with existing numerical solution tools, the Benders decomposition method, heuristic algorithm, and local search are used to improve the MILP solution. The optimal scheduling plan is output when the convergence conditions are met. The power generation status of the new energy base and the operating status of the power system are monitored in real time. If there is a deviation or it exceeds the constraint range, the scheduling program is triggered to be re-optimized in time. The scheduling strategy is adjusted in real time based on the scheduling execution status and system load feedback, and the optimal transmission strategy is determined based on meeting the local power grid needs.

[0132] In step 3, a modified mixed-integer linear programming (MILP) algorithm is used to solve the constructed optimization scheduling model, resulting in the optimal generation plans for wind power, photovoltaic power, and CSP power for each scheduling period. The scheduling optimization results are used to determine the optimal wind-photovoltaic-CSP power base bundling and transmission strategy, taking into account the generation characteristics of each renewable energy source and meeting local grid needs. During the solution process, the complementary nature of different energy sources is fully considered. For example, when wind power generation is high and photovoltaic output is low, CSP power generation is prioritized for energy storage to balance power fluctuations. When sunlight is abundant and wind speeds are low, photovoltaic and CSP power generation output is increased, and wind power reserve capacity is adjusted as needed. Based on the resulting optimal generation plan, real-time scheduling and control are implemented for wind power bases, photovoltaic bases, and CSP power stations. During the scheduling process, actual parameters such as wind speed, sunlight intensity, and generated power are monitored in real time and compared with predicted values. If significant deviations occur, the scheduling plan is promptly adjusted to ensure the safe and stable operation of the power system and the quality of the transmitted power.

[0133] For large-scale MILP problems, distributed parallel computing is used to improve the solution speed. That is, the problem is decomposed into multiple sub-problems and solved in parallel on multiple processors. At the same time, the parallel computing capabilities of GPU are used to accelerate linear algebra operations.

[0134] The improved mixed-integer linear programming (MILP) is solved using existing numerical solvers, preferably CPLEX or Gurobi. At each time period t, the improved MILP solves the optimal wind power, photovoltaic, and solar thermal power generation output and thermal storage scheduling strategy based on real-time meteorological data, load demand, and grid status to achieve grid load balancing and maximize energy utilization.

[0135] The mixed integer linear programming (MILP) problem has the form:

[0136] Objective function: z = c T x+θ

[0137] Constraints: Ax≤b,A eq x=b eq ,x∈X

[0138] Where x is the decision variable vector, c is the objective function coefficient vector, A is the inequality constraint coefficient matrix, b is the inequality constraint right-hand side term vector, and A eq is the equality constraint coefficient matrix, b eq is the vector of terms on the right side of the equality constraint, S is the domain of the variables, and some variables are required to be integers.

[0139] The improvements are as follows:

[0140] The Benders decomposition method is used to decompose the MILP problem into a main problem and multiple subproblems.

[0141] Assume that the MILP problem is min{c T x+d T y:Ax+By≤b,x∈X,y∈Y}, where x is an integer variable and y is a continuous variable.

[0142] The main problem is an integer programming problem, which only considers integer variables y and the objective function is z = c T x+θ, with the constraints Ax≤b and x∈X, where θ is a variable to be determined.

[0143] The sub-problem is a linear programming problem. For a given x, solve θ(x) = min{d T y:By≤b-Ax,y∈Y}.

[0144] During the solution process, the main problem and sub-problems are solved alternately. The solution of the main problem provides constraints for the sub-problem, and the solution of the sub-problem is fed back to the main problem to update the objective function.

[0145] Heuristic algorithms can quickly find high-quality approximate solutions and are commonly used to solve large-scale MILP problems. This paper improves the MILP problem using local search. Starting from an initial feasible solution, the quality of the solution is gradually improved by making local adjustments to the solution.

[0146] Preferably, the initial solution is generated using a heuristic method, which utilizes the specific structure of the problem and designs heuristic rules to generate the initial solution. Before solving the MILP, a greedy algorithm can be used to obtain an initial solution.

[0147] After obtaining a MILP solution, a local search algorithm is applied to improve it. Starting from the current solution, a better solution is searched within a defined neighborhood. If a better solution is found, the current solution is updated and the search continues. If no better solution is found after a certain number of iterations, the search is terminated.

[0148] Starting from different initial solutions, multiple local searches are performed to increase the probability of finding the global optimal solution. The neighborhood size is dynamically adjusted according to the search process, using a larger neighborhood in the early stages of the search to speed up the search and a smaller neighborhood in the later stages of the search to refine the search.

[0149] Different solvers have different parameter settings, and adjusting these parameters can improve solution efficiency. Preferably, a pivoting branch-and-bound strategy is used, which selects different branching variables and branching methods. Optionally, a cutting plane generation strategy is selected: which types of cutting planes are generated.

[0150] When the convergence conditions of the algorithm are met, the optimal scheduling plan is output, which determines the power generation plan of each power generation equipment in different time periods and the charging and discharging strategy of the energy storage equipment. According to the optimal power generation plan obtained by the solution, the wind power base, photovoltaic base and solar thermal power station are dispatched and controlled in real time through the power dispatch control system. During the scheduling process, the power generation situation of the new energy base (including the actual power generation power of wind power, photovoltaic and solar thermal) and the operating status of the power system (such as load demand, line flow, etc.) are monitored in real time through the power system monitoring system. When the actual situation deviates significantly from the predicted results or the system operating parameters exceed the constraint range, the re-optimization scheduling program is triggered in time to adjust and optimize the scheduling plan to ensure the safe and stable operation of the power system and the efficient absorption of new energy. The adjustment method adopts the idea of ​​rolling optimization. Taking the current moment as the starting point, short-term power forecasts and optimized scheduling calculations are re-performed, and the power generation plan is updated to ensure the safe and stable operation of the power system.

[0151] During the solution process, the complementary characteristics of different energy sources are fully considered. For example, when the wind power forecast value is high and the photovoltaic power forecast value is low in a certain period of time, the solar thermal power generation plan is appropriately adjusted to give priority to storing excess thermal energy to balance power fluctuations. Conversely, when there is sufficient sunlight and low wind speed, the output of photovoltaic and solar thermal power generation is increased, and the backup capacity of wind power is adjusted as needed.

[0152] The dispatch strategy is adjusted in real time based on dispatch execution and system load feedback. If the system experiences load imbalance or sudden abnormalities (such as extreme weather causing large fluctuations in power generation), adjustments can be made by adjusting the dispatch strategy, increasing energy storage release, or activating backup energy sources to ensure stable grid operation.

[0153] Based on meeting local grid demand, dispatch optimization results are used to determine the optimal transmission strategy, including the amount of power to be transmitted, the transmission time window, and the target transmission area. Based on changes in transmission demand, the generation plan of the new energy base is adjusted to maximize the advantages of the three energy sources. Dynamic optimization is also performed based on electricity market prices to reduce economic losses during power transmission.

[0154] Step 4: Effect evaluation and feedback: During the implementation process, the system continuously adjusts and optimizes according to the actual operating status, dynamically adjusts the scheduling strategy based on real-time data feedback, and regularly evaluates the implementation effect of the new energy base bundled transmission scheduling strategy. The evaluation indicators include new energy absorption rate, stability of transmitted power, power generation cost, etc. Based on the evaluation results, the data collection and prediction model, optimization scheduling model, and scheduling control strategy are optimized and adjusted.

[0155] During step 4, the system continuously adjusts and optimizes based on actual operating conditions, dynamically adjusting the dispatch strategy based on real-time data feedback to ensure continuous optimized operation under complex climate and grid operating conditions. The effectiveness of the bundled transmission dispatch strategy for new energy bases is regularly evaluated, using metrics such as the new energy absorption rate, the stability of transmitted power, and the cost of power generation. Based on the evaluation results, adjustments are made to the data acquisition and prediction model, the optimized dispatch model, and the dispatch control strategy to continuously improve the performance and adaptability of the dispatch strategy.

[0156] The formula for calculating the new energy consumption rate is:

[0157]

[0158] The power fluctuation rate calculation formula is:

[0159]

[0160] in, is the average value of the total generated power.

[0161] Based on the evaluation results, optimization and adjustments are made to the data acquisition and prediction model, the optimized dispatch model, and the dispatch control strategy. For example, if the wind power forecast error is found to be large, the wind power forecast model may need to be retrained, with more meteorological variables or adjustments to the model structure. If the power generation cost is too high, the cost structure should be analyzed and the operation and maintenance cost coefficient or the backup cost coefficient should be adjusted. If the stability of the transmitted power is insufficient, the constraints in the dispatch model should be optimized or the adjustment thresholds and adjustment methods in the dispatch control strategy should be adjusted to continuously improve the performance and adaptability of the dispatch strategy.

[0162] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0163] Figure 1The entire process of the present invention is shown. First, the data collection and power prediction stage begins. In this stage, historical environmental data is collected from the corresponding bases for the three new energy sources of wind power, photovoltaic power, and solar thermal power. For example, historical wind speed data for wind power bases, historical light intensity data for photovoltaic bases, and historical solar irradiance data for solar thermal power stations are collected. This data is used to construct the respective power generation characteristic models, fully considering the impact of environmental factors such as wind speed, sunshine, temperature, and humidity on power generation. Then, with the help of machine learning and long short-term memory networks (LSTM), combined with historical meteorological data and real-time meteorological data, wind speed, light intensity and solar irradiance in the future are predicted. Then, short-term (next 1-6 hours, to meet the needs of real-time scheduling and rapid response) and medium- and long-term (next 1-7 days, for formulating long-term power generation plans and resource allocation strategies) predictions are made for the power generation of wind power and photovoltaic power, as well as the power of solar thermal power stations and the heat storage and release of solar thermal systems. At the same time, key data such as the transmission capacity limit of the power transmission channel and the load demand of the receiving power grid are obtained. In addition, the heat storage electric boilers equipped in the new energy base play a role, which can convert the green electricity of wind power and photovoltaic power into thermal energy of the solar thermal system when the power grid is at a low point, laying the foundation for subsequent scheduling.

[0164] After completing data collection and prediction, the optimization scheduling model construction process begins. With the objective functions of ensuring the overall grid stability of the new energy base, controlling power generation costs, and rationally setting energy scheduling penalties, the system comprehensively considers the operational characteristics and constraints of wind, photovoltaic, and solar thermal power generation. These include the relationship between wind power output limits and wind speed, the relationship between photovoltaic output limits and light intensity, and the heat storage and release characteristics of thermal storage devices in solar thermal power generation systems. Furthermore, factors such as power balance constraints and transmission line capacity constraints are also considered. A multi-energy joint scheduling model is constructed, which comprehensively incorporates the output characteristics, power generation capacity, load demand, and scheduling constraints of wind, photovoltaic, and solar thermal power. The optimization objective function is subdivided into three key components: a load error term, used to measure the deviation between grid load demand and the power generation output of each energy source, aiming to minimize load matching errors; the total power generation cost of the new energy base, encompassing the costs of wind power, solar thermal power, and CSP power generation; and an energy dispatch penalty term, which accounts for scheduling failures due to generation limitations or volatility of different energy sources, including wind and solar curtailment. Detailed calculation formulas for each cost component are also defined. For example, the costs of wind power, photovoltaic power, and CSP power generation each include equipment operation and maintenance costs, as well as backup costs due to power generation fluctuations. Curtailment penalty costs are calculated based on the penalty cost coefficients per unit wind and solar curtailment power and the actual curtailment power. A strategy for utilizing excess wind and photovoltaic power and controlling thermal storage is also proposed. Specifically, when wind and photovoltaic power generation exceeds grid demand or export capacity, it can be directly used to heat the thermal storage medium of the CSP system. Reasonable control is achieved based on formulas such as thermal storage medium temperature change and heating power. Finally, the export capacity and strategy are determined based on the dispatch optimization results and grid demand. By flexibly adjusting relevant parameters, a precise balance is achieved between meeting grid load demand, reducing energy curtailment waste, and ensuring economic efficiency.

[0165] Based on the constructed optimization scheduling model, an improved mixed-integer linear programming (MILP) method is used for model solution and scheduling implementation. The solution fully exploits the complementary characteristics of different energy sources. For example, when wind power generation is high and photovoltaic output is low, CSP is prioritized for energy storage to balance power fluctuations. When sunlight is sufficient and wind speed is low, the output of photovoltaic and CSP is increased, and the reserve capacity of wind power is adjusted as needed. Through a series of complex calculations, the optimal power generation plan for wind power, photovoltaic power, and CSP in each scheduling period is obtained. Then, based on the solved plan, real-time scheduling and control of wind power bases, photovoltaic bases, and CSP power stations are carried out. During this process, actual parameters such as wind speed, sunlight intensity, and power generation are monitored in real time and rigorously compared and analyzed with predicted values. In the event of significant deviations, the scheduling plan is quickly adjusted to ensure the safe and stable operation of the power system and the quality of the transmitted power. Taking into account the large-scale MILP problems that may be encountered in practical applications, distributed parallel computing is used to significantly improve the solution speed, that is, to cleverly decompose complex problems into multiple sub-problems, and solve them in parallel on multiple processors. At the same time, the parallel computing capabilities of GPUs are fully utilized to accelerate linear algebra operations, and when solving, existing numerical solution tools are combined, preferably CPLEX tools, and Gurobi tools are optional. The Benders decomposition method is used to decompose the MILP problem into a main problem and multiple sub-problems. The main problem is an integer programming problem that only considers integer variables, and the sub-problem is a linear programming problem. The two are solved alternately, and the solutions of the sub-problems are fed back to the main problem to update the objective function. At the same time, a heuristic algorithm is used to quickly find high-quality approximate solutions. Starting from an initial feasible solution, the quality of the solution is gradually improved by making local adjustments to the solution. A local search algorithm can also be applied to start from the current solution and search for a better solution within the defined neighborhood. If a better solution is found, the current solution is updated and the search continues. If no better solution is found after a certain number of iterations, the search is terminated. Starting from multiple different initial solutions, multiple local searches are performed to increase the probability of finding the global optimal solution. The neighborhood size is dynamically adjusted according to the search process. A larger neighborhood is used in the early stage of the search to speed up the search, and a smaller neighborhood is used in the later stage of the search for a fine search. Different solvers set different parameters according to their characteristics, such as using a rotating branch and bound strategy to select different branch variables and branching methods, and an optional cutting plane generation strategy to select which types of cutting planes to generate. When the convergence conditions of the algorithm are met, the optimal scheduling plan is output, that is, the power generation plan of each power generation equipment in different time periods and the charging and discharging strategy of the energy storage equipment are determined.At the same time, the power system monitoring system monitors the power generation situation of the new energy base (including the actual power generation power of wind power, photovoltaic power, and solar thermal power) and the operating status of the power system (such as load demand, line flow, etc.) in real time. When the actual situation deviates greatly from the predicted results or the system operating parameters exceed the constraint range, the re-optimization scheduling program is triggered in time to adjust and optimize the scheduling plan to ensure the safe and stable operation of the power system. The scheduling strategy is adjusted in real time according to the scheduling execution situation and system load feedback. On the basis of meeting the needs of the local power grid, the scheduling optimization results are used to determine the best transmission strategy, including the transmission power, transmission time window and transmission target area. According to the changes in the demand for transmission power, the power generation plan of the new energy base is adjusted to ensure that the advantages of the three energy sources are maximized, and dynamic optimization is carried out according to the electricity market price to reduce the economic losses in the process of power transmission.

[0166] Throughout the entire dispatch strategy implementation process, the effectiveness evaluation and feedback loop plays a continuous role. The system continuously adjusts and optimizes based on actual operating conditions, dynamically adjusting the dispatch strategy based on real-time data feedback. Regular comprehensive evaluations are conducted on the effectiveness of the new energy base bundled transmission dispatch strategy. Evaluation metrics cover key factors such as the new energy absorption rate, the stability of the transmitted power, and the generation cost. The new energy absorption rate is calculated using a specific formula, and the power fluctuation rate also has a corresponding precise formula. Based on the evaluation results, targeted optimization and adjustments are made to the data collection and prediction model, the optimized dispatch model, and the dispatch control strategy. For example, if large wind power forecast errors are found, the wind power forecast model may need to be retrained, with additional meteorological variables or adjustments to the model structure. If the generation cost is too high, a detailed analysis of the cost structure is performed, and the operation and maintenance cost coefficient or the backup cost coefficient is adjusted. If the transmission power stability is insufficient, the constraints in the dispatch model are optimized or the adjustment thresholds and methods in the dispatch control strategy are adjusted. This cycle continuously improves the performance and adaptability of the dispatch strategy, ensuring the efficiency and stability of the new energy base bundled transmission dispatch.

[0167] Figure 2The dispatch process sequence diagram shows the time series relationship between output forecasting, aggregated dispatching, and bundled transmission for wind, photovoltaic, and solar thermal systems. First, each renewable energy subsystem independently performs short-term power forecasting. The wind power forecasting system estimates future power curves based on historical wind speed and meteorological data, the photovoltaic forecasting system predicts output based on parameters such as sunshine intensity and temperature, and the solar thermal system combines thermal storage tank status with pre-set operating strategies to generate a combined heat and power output forecast. Subsequently, a unified dispatch center aggregates the forecast data for the three energy sources to form a total available power curve. This data is then combined with the grid's transmission plan and electricity price curve to calculate the optimal output combination for each time period. Based on this, the dispatch strategy module outputs a bundled transmission plan, reserving spare capacity and taking into account the responsiveness of each subsystem. The execution layer, comprised of the EMS and energy management PLC, receives dispatch commands, controls the actual output power of the wind power converter, photovoltaic inverter, and solar thermal turbine system, and monitors any deviations from the schedule. If an output deviation or unplanned failure occurs in a certain time slice, the feedback mechanism will trigger the emergency regulation strategy and dynamically adjust the bundling plan to ensure that the output power of the entire system is stable, economical and meets the synchronization and coordination requirements.

[0168] Cooperate Figure 3 As shown in the figure, in the complex operation scenario of bundled delivery from new energy bases, the time sequence diagram of the multi-energy joint scheduling model depicts the dynamic process of orderly development and interaction of various key links as time goes by.

[0169] During the initial phase, the system initiates the data collection process. Simultaneously, wind speed sensors at wind farms, light intensity sensors at photovoltaic plants, and solar irradiance sensors at solar thermal power plants operate synchronously, collecting real-time environmental data that accurately reflects current natural conditions. Almost simultaneously, load monitoring equipment within the power system closely tracks the load demand of the receiving grid, continuously transmitting real-time load data to the dispatch center. Furthermore, transmission line monitoring devices continuously monitor line status, capturing critical information such as transmission capacity limitations and feeding it back to the system.

[0170] Once data collection is complete, the prediction phase begins. Based on the collected historical and real-time environmental data, machine learning and LSTM models quickly initiate predictions for future wind speed, light intensity, and solar irradiance. During this timeframe, the model, drawing on patterns mined from massive amounts of historical data and the dynamic changes in real-time data, accurately infers key indicators related to energy generation for the coming period. This includes forecasts for wind power, photovoltaic power generation, and heat storage and release at solar thermal power stations. The prediction cycle is flexibly adjusted based on scheduling needs, ranging from as short as 1-6 hours to meet real-time scheduling needs, or as long as 1-7 days to facilitate long-term power generation planning.

[0171] The moment the prediction result is generated, it serves as a key input to trigger the construction of the optimization scheduling model. At this time, the system quickly integrates various constraints with the core demands of ensuring grid stability, controlling power generation costs, and reducing energy scheduling penalties. On the one hand, the operating characteristics and constraints of wind power, photovoltaic power generation, and solar thermal power generation are considered in detail, such as the correspondence between wind speed and output for wind power, the impact of photovoltaic power generation combined with light intensity, and the boundaries set by solar thermal power generation based on heat storage and heat release characteristics; on the other hand, the power balance constraint is taken into account to ensure that the total power generation matches the load demand, and the transmission line capacity constraint prevents transmission overload. On this basis, a rigorous multi-energy joint scheduling model is constructed, which uses complex algorithms to accurately calculate the load error term, total power generation cost, energy scheduling penalty term and other objective function components.

[0172] Once the model is built, the solution phase begins. Using an improved mixed-integer linear programming (MILP) method, combined with distributed parallel computing resources, the solver operates at high speed, fully exploiting the complementary characteristics of various energy sources. For example, when wind power is high and photovoltaic power is insufficient, solar thermal energy storage is prioritized for power balance. When sunlight is abundant and wind speeds are low, photovoltaic and solar thermal power generation are increased while wind power reserve is adjusted. During the solution process, Benders decomposition, heuristic algorithms, and local search are also used to accelerate optimization. Multiple searches are performed starting from multiple initial solutions, and the neighborhood size is dynamically adjusted until convergence conditions are met, resulting in the output of the optimal power generation plan.

[0173] Once the optimal power generation plan is determined, it is seamlessly integrated into the real-time dispatch implementation process. Based on the plan, the dispatch center issues precise power generation instructions to wind power bases, photovoltaic bases, and solar thermal power stations. Each base responds quickly and adjusts the operating status of power generation equipment in real time. Simultaneously, the system initiates a rigorous monitoring program, continuously comparing actual wind speed, light intensity, power generation, and other parameters with predicted values. Once a significant deviation is detected, the system immediately triggers a re-optimization of the dispatch program, rapidly adjusting the power generation plan to ensure power system stability. Furthermore, throughout the entire process, effect evaluation and feedback mechanisms operate in parallel. The effectiveness of dispatch strategy implementation is regularly evaluated based on indicators such as the new energy absorption rate, the stability of transmitted power, and the cost of power generation. Based on the evaluation results, data collection, prediction models, dispatch models, and dispatch strategies are reversely optimized, forming a closed-loop process of continuous improvement to ensure that the bundled transmission dispatch of new energy bases remains efficient over time.

[0174] Figure 4The electrical-thermal coupling diagram shown here illustrates the energy conversion relationship between excess wind and photovoltaic power and the CSP thermal storage system. When wind or photovoltaic power generation exceeds grid demand or export capacity, the excess electricity is converted to heat by an electric heating device (such as an electrode boiler) and stored in the CSP system's molten salt thermal storage tank. This process involves three key steps: First, the curtailed wind / photovoltaic energy is converted to voltage / current parameters suitable for the heating device by a power electronic converter; second, the electrical energy is heated by a thermal storage medium such as molten salt through resistance heating or electromagnetic induction, with the thermal power conversion efficiency determined by the characteristics of the heating device; finally, the temperature of the thermal storage medium changes according to the laws of thermodynamics, with its temperature change rate over time being related to the input thermal power, heat loss coefficient, and ambient temperature. During peak grid load periods, the thermal storage system releases this heat energy through a steam generator to generate electricity, forming an "electricity-heat-electricity" energy cycle.

[0175] Cooperate Figure 5 In the wind power-photovoltaic-solar thermal energy base bundled transmission optimization scheduling strategy of the present invention, a distributed coordination optimization method is used to solve the scheduling problem. This process is based on the main problem-subproblem architecture, and divides the optimization scheduling problem of the entire new energy base into multiple subproblems, corresponding to the wind power subsystem, photovoltaic subsystem, solar thermal energy storage subsystem and power transmission channel resource scheduling subsystem. Each subproblem independently constructs an optimization model and solves the sub-optimal solution based on its own local information and constraints. Subsequently, the main coordinator uniformly adjusts the system-level coupling variables (such as total power constraints, channel capacity limitations, etc.) according to the local solutions reported by each subsystem, and feeds the adjustment results back to the subproblems for iterative updates. This process adopts a distributed coordination mechanism that combines the augmented Lagrangian method with the alternating direction multiplier method (ADMM), and converges to the global optimal or approximate optimal scheduling solution of the overall system through parallel iterative optimization. This method effectively improves the computational efficiency and robustness of multi-energy collaborative scheduling, and is particularly suitable for engineering scenarios with a high proportion of new energy access and large computing scale.

[0176] In summary, the present invention mainly includes the following points:

[0177] (1) In the data collection and power prediction steps, historical environmental data of wind power, photovoltaic, and solar thermal bases are collected to establish a power generation characteristic model. Machine learning and LSTM are used to predict wind speed, light intensity, and solar irradiance based on historical and real-time meteorological data. The wind power and photovoltaic power generation power and the heat storage and heat release of solar thermal power stations are predicted. A joint power generation scheduling model is set up, considering the factors affecting the power generation of each energy source and the model formula, monitoring the load demand of the power grid and making predictions. The new energy base is equipped with a heat storage electric boiler to realize the conversion of green electricity into heat storage.

[0178] (2) In the step of constructing the optimization dispatching model, the grid stability, power generation cost, and energy dispatching penalty are taken as the objective functions, and the multi-energy joint dispatching model is constructed by considering the constraints such as the operating characteristics of each energy, power balance, and transmission line capacity. The model covers the output characteristics of each energy, power generation capacity and other conditions. The optimization objective function includes load error, total power generation cost, and energy dispatching penalty items. The cost calculation formulas and the effects of wind and solar power abandonment penalty coefficients are clarified. The utilization and heat storage control strategies of wind power and photovoltaic power generation excess are given. The transmission power and strategy are determined according to the dispatching results and grid demand, and the parameters are adjusted to balance the grid load, abandoned energy and economy.

[0179] (3) In the model solving and scheduling implementation steps, the improved mixed integer linear programming (MILP) is used to solve the optimization scheduling model to obtain the optimal power generation plan. The solution takes into account the complementary characteristics of energy sources, and each base is scheduled in real time according to the plan. The parameters are monitored and compared with the predicted values. If there is a large deviation, it is adjusted in time. For large-scale MILP problems, distributed parallel computing is used. Combined with CPLEX or Gurobi tools, Benders decomposition method, heuristic algorithm, and local search are used to improve MILP. The optimal scheduling plan is output according to the convergence conditions. The power generation and operation status are monitored in real time. If the deviation is large or exceeds the constraint range, re-optimization is triggered. The strategy is adjusted in real time according to the scheduling execution and load feedback. The transmission strategy is determined and the power generation plan is adjusted according to demand.

[0180] (4) In the effect evaluation and feedback step, the system dynamically adjusts the dispatching strategy according to the actual operation and regularly evaluates the implementation effect. The indicators include the new energy consumption rate, the stability of the external power transmission, the power generation cost, etc. Based on the results, the data collection and prediction model, the optimized dispatching model, and the dispatching control strategy are optimized.

[0181] The present invention has the following key improvements:

[0182] (1) Multi-energy complementary coordinated dispatch: Based on the characteristics of three new energy sources, wind power, photovoltaic power, and solar thermal power, coordinated optimized dispatch is achieved to give full play to their complementary advantages and improve the overall stability and economy of the system.

[0183] (2) Power forecasting based on machine learning and LSTM: Using machine learning and long short-term memory network (LSTM) models, high-precision short-term and medium-term forecasts of wind power, photovoltaic, and solar thermal power generation and power load are made, providing accurate data support for optimized scheduling.

[0184] (3) Multi-objective optimization scheduling model: Construct a multi-objective optimization model with grid stability, power generation cost and energy scheduling penalty (for wind, solar and thermal power abandonment) as the objective function, comprehensively consider various operating constraints, and achieve economical, reliable and efficient scheduling.

[0185] (4) Improved Mixed Integer Linear Programming (MILP) Solving Method and Distributed Parallel Computing: Improved MILP methods (such as those combining Benders decomposition, heuristic algorithms, and local search) combined with distributed parallel computing can improve the efficiency and real-time performance of solving large-scale optimization problems and meet actual scheduling requirements. Distributed parallel computing architecture (such as GPU acceleration) supports high-frequency real-time scheduling decisions.

[0186] (5) Optimal utilization of energy storage in solar thermal systems: Emphasize the key role of energy storage devices in solar thermal systems in smoothing out fluctuations in wind power and photovoltaic power generation, improving system adjustability and reliability, and incorporate them into the overall optimization scheduling strategy.

[0187] (6) Real-time dynamic adjustment and feedback mechanism: Implement real-time system operation status monitoring and scheduling effect evaluation, dynamically adjust scheduling strategies and model parameters based on feedback information, and improve the system's adaptability and robustness.

[0188] (7) Synergy between thermal storage system and new energy: A temperature control model for the thermal storage medium of a solar thermal system is used to heat the thermal storage medium through excess wind / photovoltaic power, thereby improving system flexibility. Dynamic conversion constraints between thermal storage and power generation ensure that the thermodynamic characteristics match the grid demand.

[0189] It will be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented in various computer languages, for example, the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0190] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0191] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0192] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0193] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0194] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A wind power-photovoltaic-solar thermal energy base bundled delivery optimization scheduling method, characterized by The steps include: Step 1: Obtain historical environmental data for the wind power, photovoltaic, and solar thermal energy bases, and obtain forecast data for wind speed, light intensity, and solar irradiance over the next period of time, as well as forecast data for wind power, photovoltaic power generation, and heat storage and release conditions of solar thermal power stations. Step 2: Taking the overall grid stability, power generation cost, and energy dispatch penalty of the new energy base as the objective function, and considering the operating characteristics constraints of wind power, photovoltaic power generation, and solar thermal power generation, power balance constraints, and transmission line capacity constraints, a joint power generation optimization dispatch model is constructed; Step 3: Use improved mixed integer linear programming to solve the joint power generation optimization scheduling model to obtain the optimal power generation plan of each renewable energy source in each scheduling period. Based on the solution results, each renewable energy base is dispatched and controlled in real time. The scheduling strategy is adjusted in real time according to the actual situation to determine the optimal transmission strategy.

2. The method according to claim 1, wherein: In step 1, historical meteorological data is collected and real-time meteorological data is obtained, and a prediction model is established using machine learning and long-short-term memory networks to predict wind speed, light intensity, solar irradiance, wind power, photovoltaic power generation, and heat storage and release of solar thermal power stations in the future.

3. The method according to claim 1, wherein: In step 2, the objective function of the joint power generation optimization scheduling model is: Where: T represents the scheduling time period, α1 is the weight coefficient of the load error term, α2 is the weight coefficient of the scheduling priority of the power generation cost, and α3 is the penalty coefficient; P load (t) is the power generation demanded by the grid load, P wind (t) is the wind power generation, P pv (t) is the photovoltaic power generation, P csp (t) is the amount of solar thermal power generation; C wind is the power generation cost of the wind farm, C pv is the power generation cost of the photovoltaic power station, C csp is the power generation cost of the CSP station, C penalty Penalty costs for curtailing wind and solar power.

4. The method according to claim 1, wherein: In step 2, the constraints of the joint power generation optimization scheduling model include: Power balance constraints: P wind (t)+P pv (t)+P csp (t)=P load (t)+P loss (t) Among them, P load (t) is the load demand power of the power system at time t, P loss (t) is the power loss of the transmission line at time t; Grid transmission constraints: P out (t)≤P max , Among them, P out (t) is the power output of the power grid, P max is the maximum transmission capacity of the power grid; Thermal storage system constraints: T min ≤T storage (t)≤T max Among them, T min and T max are the minimum and maximum temperatures of the heat storage system, respectively; Among them, P heat (t) is the power of the external heating heat storage system, P hrelease (t) is the power of the heat storage system supplying heat to the outside world; Restrictions on wind power, photovoltaic power generation, and solar thermal power generation: P wind (t)≤P wind,max (t) P PV (t)≤P PV,max (t) P CSP (t)≤P CSP,max (t) Constraints on the thermodynamic characteristics of photothermal systems: E store (t+1)=E store (t)+η collector Q collector (t)-E loss,store (t)-E release (t) Among them, η collector is the efficiency of the solar thermal collector, Q collector (t) is the heat collected by the solar thermal collector at time t, E loss,store (t) is the heat loss of the heat storage system at time t.

5. The method according to claim 1, wherein: In step 3, the improved mixed integer linear programming is used to solve the joint power generation optimization scheduling model, and the form of the mixed integer linear programming problem is set as: Objective function: z = c T x+θ Constraints: Ax≤b,A eq x=b eq ,x∈X Where x is the decision variable vector, c is the objective function coefficient vector, A is the inequality constraint coefficient matrix, b is the inequality constraint right-hand side term vector, and A eq is the equality constraint coefficient matrix, b eq is the vector of terms on the right side of the equality constraint, and S is the domain of the variable; The mixed integer linear programming problem is decomposed into a main problem and multiple sub-problems using Benders decomposition method; Assume that the mixed integer linear programming problem is min{c T x+d T y:Ax+By≤b,x∈X,y∈Y}, where x is an integer variable and y is a continuous variable; the main problem is an integer programming problem, only considering the integer variable y, and the objective function is z=c T x+θ, the constraints are Ax≤b and x∈X, where θ is a variable to be determined; the subproblem is a linear programming problem, for a given x, solve θ(x)=min{d T y:By≤b-Ax,y∈Y}; During the solution process, the main problem and sub-problems are solved alternately. The solution of the main problem provides constraints for the sub-problem, and the solution of the sub-problem is fed back to the main problem to update the objective function.

6. The method according to claim 5, wherein: In step 3, a greedy algorithm is first used to obtain an initial solution; starting from the current solution, a better solution is searched within the defined neighborhood; if a better solution is found, the current solution is updated and the search continues; if no better solution is found after a certain number of iterations, the search is terminated; When the convergence conditions are met, the optimal scheduling plan is output, that is, the optimal power generation plan of each renewable energy source in each scheduling period.

7. The method according to claim 1, wherein: In step 3, based on the obtained optimal power generation plan, the wind power base, photovoltaic base and solar thermal power station are dispatched and controlled in real time through the power dispatching and control system, and the power generation situation of the new energy base and the operating status of the power system are monitored in real time. When the actual situation deviates greatly from the predicted result or the system operating parameters exceed the constraint range, the dispatching plan is adjusted and optimized.

8. The method according to claim 1, wherein: In step 3, distributed parallel computing is used to improve the solution speed, that is, the problem is decomposed into multiple sub-problems, which are solved in parallel on multiple processors. At the same time, the parallel computing capability of the GPU is used to accelerate linear algebra operations, and the CPLEX tool or the Gurobi tool is used for solving.

9. The method according to claim 1, wherein: It also includes step 4, obtaining the operating status of the new energy base in real time, evaluating the transmission strategy, and the evaluation indicators include the new energy consumption rate, the stability of the transmitted power, and the power generation cost. The acquisition process of the transmission strategy is optimized and adjusted according to the evaluation results.

Images