Method for configuring capacity of energy system based on inaccurate prediction of renewable energy output

Abstract

The present invention discloses a method for configuring the capacity of an energy system based on the inaccurate prediction of renewable energy output. It solves the problems in the prior art that the calculation of wind and solar power generation is carried out with a fixed output curve and only the uncertainty of renewable energy output is considered, resulting in differences in operation scheduling and installed capacity configuration. The method includes establishing a renewable energy output model by combining the renewable energy benchmark output curve and the renewable energy output prediction curve; establishing an energy system configuration optimization model with the goal of the overall optimal operation of the energy system; and using a rolling window optimization algorithm to solve the energy system configuration optimization model to obtain the optimal capacity configuration plan. The present invention comprehensively considers the impacts of the uncertainty of renewable energy output and the inaccuracy of renewable energy prediction on the energy system planning, achieving the optimal system economics while improving the stability of the energy system operation.

Description

Technical Field

[0001] The present invention relates to the technical field of energy optimization allocation, and in particular to a method for configuring the capacity of an energy system based on inaccurate prediction of renewable energy output. Background Art

[0002] With the continuous increase in the proportion of renewable energy such as wind energy and solar energy in the energy system, the impact and challenges brought by the uncertainty of its output to the energy system are becoming greater and greater. This makes it necessary to reserve a part of flexible resources to cope with the negative impact brought by the uncertainty of renewable energy output when planning the energy system. Most of the current technologies aim at the overall optimal benefit of the energy system. First, the uncertainty curve of renewable energy output is characterized, and then the installed capacity of each unit and the real-time scheduling of the energy system are obtained through optimization calculation. In addition, the current technology does not analyze how high the prediction accuracy of renewable energy output should be to basically meet the demand.

[0003] For example, the invention patent with the patent number CN202211005903.X and the name of a distributionally robust configuration method for an integrated energy system considering the uncertainty of renewable energy output. The key points of its technical solution are that this configuration method comprehensively considers the uncertainty of renewable energy output, aims at the overall optimal operation of the energy system, adopts a distributionally robust optimization method for the uncertainty of error distribution, and continuously searches for the probability distribution of the most extreme weather conditions within the constraints, ensuring the optimal value of the result of the energy system under the probability distribution of the most extreme climate conditions, and the configuration result is more conservative to achieve a better configuration effect.

[0004] The existing technology still has disadvantages. When optimizing the configuration of the energy system, most of the calculations of wind and solar power generation are based on a fixed output curve, or only the uncertainty of renewable energy output is considered, lacking the consideration of the differences in operation scheduling and installed capacity configuration caused by the prediction accuracy of renewable energy output. Since the prediction of renewable energy output has a very important impact on the scheduling of the energy system, and further affects the overall configuration planning of the energy system, it is necessary to study an energy system capacity configuration optimization method that can consider both the uncertainty of renewable energy output and the uncertainty of the short-term output prediction accuracy of renewable energy. Summary of the Invention

[0005] The present invention mainly solves the problems that the existing technology calculates wind and solar power generation based on a fixed output curve and only considers the uncertainty of renewable energy output, resulting in differences in operation scheduling and installed capacity configuration, and provides a method for configuring the capacity of an energy system based on inaccurate prediction of renewable energy output.

[0006] The above technical problems of the present invention are mainly solved by the following technical solutions: An energy system capacity configuration method based on inaccurate renewable energy output prediction, comprising the following steps:

[0007] Obtain the historical output data of renewable energy at each moment to construct a renewable energy baseline output curve, calculate a renewable energy output prediction curve based on the renewable energy baseline output, and establish a renewable energy output model by combining the renewable energy baseline output curve and the renewable energy output prediction curve;

[0008] Establish an energy system configuration optimization model with the goal of overall optimal operation of the energy system;

[0009] Adopt a rolling window optimization algorithm to solve the energy system configuration optimization model and obtain an optimal capacity configuration plan.

[0010] The present invention comprehensively considers the impacts of the uncertainty of renewable energy output and the inaccuracy of renewable energy prediction on energy system planning, and achieves the best system economics while improving the stability of energy system operation. By introducing inaccurate prediction accuracy, the impact of short-term weather prediction on energy system configuration optimization is taken into account, which makes the model planning process closer to the actual situation and ultimately improves the adaptability of the energy system to extreme weather conditions or situations with increased errors. By analyzing, the specific requirements for improving the prediction accuracy of renewable energy output in practical applications are clarified, so as to avoid excessive resource investment caused by pursuing too high prediction accuracy.

[0011] As a preferred solution,

[0012] Renewable energy includes wind power and photovoltaic power generation. Obtain the historical wind power and photovoltaic output data of the research area in the past many years, and sample at each time node through simple random sampling to construct a renewable energy baseline output curve.

[0013] The renewable energy explored in the present invention mainly includes wind power and photovoltaic power generation. The renewable energy baseline output can be established based on the historical data of renewable energy in the research area. Specifically, obtain the historical wind power and photovoltaic output data of the research area in the past N years through satellite remote sensing data or ground monitoring stations. According to the historical output data of wind power and photovoltaic power in each year, sample at each time node of each year through simple random sampling to construct a renewable energy baseline output curve. Preferably, it can be in units of years. According to the historical output data of renewable energy in N years, sample at each time node of each historical year to construct multiple renewable energy baseline output curves.

[0014] As a preferred solution,

[0015] Based on the reference output curve of renewable energy, set the prediction accuracy of renewable energy output, and generate the prediction curve of renewable energy output through a stochastic optimization method.

[0016] During the actual operation process, operators generally determine the action amount of the key operation parameters of the energy system through short-term weather prediction values. The additional error caused by the inaccuracy of renewable energy output prediction usually resulting from short-term weather prediction errors is represented as the prediction curve of renewable energy output, which follows a normal distribution.

[0017] As a preferred solution,

[0018] The renewable energy output model is the sum of the reference output curve of renewable energy and the prediction curve of renewable energy output.

[0019] The final renewable energy output curve is obtained by combining the uncertainty of renewable energy output with the inaccuracy of renewable energy output prediction.

[0020] As a preferred solution, an optimization model for energy system configuration is established with the goal of the overall optimal operation of the energy system, including:

[0021] Establish an objective function with the goal of minimizing the sum of the operation and maintenance investment of renewable energy equipment, the purchase investment of renewable energy equipment, the operation and maintenance investment of energy storage equipment, and the purchase investment of energy storage equipment;

[0022] Establish the constraints existing in the operation of the energy system, including the supply-demand balance constraint of the energy system, the stored electricity constraint of the lithium battery, and the constraints on the charge-discharge power and installed capacity of the energy storage.

[0023] The constructed energy system simulation model includes four themes: wind power generation, photovoltaic power generation, electrochemical energy storage, and users. Configuration optimization design is carried out with the goal of the overall optimal operation of the energy system. The energy system configuration optimization model includes an objective function and constraint conditions. The objective function is established with the goal of optimal configuration, that is, minimizing the sum of the operation and maintenance investment of renewable energy equipment, the purchase investment of renewable energy equipment, the operation and maintenance investment of energy storage equipment, and the purchase investment of energy storage equipment. In addition, there are supply-demand balance constraints of the energy system, stored electricity constraints of lithium batteries, and constraints on charge-discharge power and installed capacity during the operation of the energy system.

[0024] As a preferred solution, the supply-demand balance constraint of the energy system includes:

[0025] The sum of wind power output, photovoltaic power output, and energy storage discharge power at the same moment is equal to the sum of energy storage charging power and load energy consumption power at the same moment;

[0026] Among them, wind power output and photovoltaic power output are calculated according to the renewable energy output model.

[0027] Calculate the wind power output and photovoltaic power output according to the renewable energy output model. For the energy system supply-demand balance constraint conditions, it includes that the sum of the wind power output, photovoltaic power output, and energy storage discharge power at time t is equal to the sum of the energy storage power supply and the load energy consumption power at time t.

[0028] As a preferred solution, the storage capacity constraint of the lithium battery includes:

[0029] Under the condition of setting the energy storage charging efficiency and discharging efficiency,

[0030] The energy storage power at the next moment is the difference between the sum of the energy storage power at the previous moment and the product of the energy storage charging efficiency and the energy storage charging power, divided by the energy storage discharge power and the energy storage discharge efficiency.

[0031] Specifically, the energy storage capacity constraint condition of the lithium battery storage pool is that, considering the energy storage charging and discharging efficiency, the energy stored in the energy storage at time t + 1 is equal to the energy stored in the energy storage at time t, plus the product of the energy storage charging efficiency and the energy storage charging power at time t, minus the ratio of the energy storage discharge power at time t to the energy storage discharge efficiency.

[0032] As a preferred solution, the energy storage charging and discharging power and installed capacity limit constraints include:

[0033] The energy storage charging power is in the closed interval range of 0 and the product of the maximum energy storage charging and discharging power and the energy storage charging adjustment coefficient;

[0034] The energy storage discharge power is in the closed interval range of 0 and the product of the maximum energy storage charging and discharging power and the energy storage discharge adjustment coefficient.

[0035] The specific energy storage charging and discharging power and installed capacity limit constraint conditions at time t in this solution are that the energy storage charging power is greater than or equal to 0 and less than or equal to the product of the maximum energy storage charging and discharging power and the energy storage charging adjustment coefficient, and the energy storage discharge power is greater than or equal to 0 and less than or equal to the product of the maximum energy storage charging and discharging power and the energy storage discharge adjustment coefficient. The values of the energy storage charging adjustment coefficient and the energy storage discharge adjustment coefficient are both 0 or 1.

[0036] As a preferred solution, the rolling window optimization algorithm is used to solve the energy system configuration optimization model, including:

[0037] Based on the historical output data of renewable energy, construct N renewable energy reference output curves;

[0038] Set the renewable energy output prediction accuracy, and generate M renewable energy output prediction curves for each renewable energy reference output curve based on the stochastic optimization method;

[0039] Based on each renewable energy baseline output curve and the renewable energy output prediction curve, according to the constraint conditions, the energy system configuration optimization model is solved within each time step to obtain a candidate optimal capacity configuration plan;

[0040] Select the most complex plan from all candidate optimal capacity configuration plans as the final optimal capacity configuration plan.

[0041] The rolling window optimization algorithm of the present invention considers that the output of renewable energy has strong seasonality. To achieve annual cycle simulation, the annual time T is divided into K rolling windows, the length of each window, i.e., the time step, is Δt, and the rolling step size is Δt roll , and the annual time T = K * Δt. The end time t start k+1 of each rolling window is the start time t start k of the previous window plus the rolling step size, t start k+1 = t start k + Δt roll .

[0042] First, based on the historical output data of renewable energy, the output range of renewable energy at each time node is obtained, and random sampling is performed according to the output range to construct N renewable energy baseline output curves. On this basis, the prediction accuracy of renewable energy output is set, and the N generated renewable energy baseline output curves are corrected based on the stochastic optimization method, and M renewable energy output prediction curves are generated for each renewable energy baseline output curve through Monte Carlo simulation. Based on this, the optimal configuration calculation is carried out. By initializing the installed capacities of wind power, photovoltaic, and energy storage, under the premise of considering the constraints of the energy system operation, the parallel computing technology is used to operate the objective function of the energy system configuration optimization model, and the optimal scheduling within each time step and the installed capacity data of various required devices are obtained through Yalmip and Gurobi optimization simulations, and the stored electricity of the energy storage facility after the end of this simulation period is obtained, and this is used as the initial value for the calculation of the next time step. During this process, the installed capacities of various devices in the energy system are continuously adjusted until all time periods are simulated. Finally, the candidate optimal capacity configuration of the energy system is obtained, and at the same time, the operating conditions of the energy system under various renewable energy output scenarios can be obtained and compared to obtain the final optimal capacity configuration. Since the most complex plan represents the most extreme weather conditions, it is necessary to ensure that the energy system can operate normally under the most extreme weather conditions. Therefore, the installed capacity configuration plan of the most complex plan is used as the final optimal capacity configuration plan.

[0043] As a preferred solution, it further includes the steps of calculating the prediction accuracy of the optimal capacity configuration plan, including:

[0044] Set a threshold for capacity configuration change,

[0045] Determine whether the optimal capacity configuration change amount is not less than the capacity configuration change threshold,

[0046] If not, output the prediction accuracy;

[0047] If so, increment the prediction accuracy, return to the step of generating M renewable energy output prediction curves, continue to execute the subsequent steps until the final optimal capacity configuration is obtained, and repeat the prediction accuracy calculation until the prediction accuracy is output.

[0048] This solution can determine the best demand for the prediction accuracy of renewable energy output in the future energy system. Based on the above rolling window optimization algorithm to solve the energy system configuration optimization model to obtain the final optimal capacity configuration, improve the prediction accuracy of renewable energy output, and perform cyclic parallel operations to obtain the operation of the energy system after improving the prediction accuracy of renewable energy. Until the change amount of the optimal operation of the system before and after improving the prediction accuracy is less than the capacity configuration threshold Q, it means that the effect of improving the prediction accuracy of renewable energy will be very limited, thus proving that the prediction accuracy of renewable energy output in the p-1 case can meet the daily production planning requirements.

[0049] Therefore, the advantages of the present invention are as follows: comprehensively considering the impacts of the uncertainty of renewable energy output and the inaccuracy of renewable energy prediction on the energy system planning, while improving the stability of the energy system operation, achieving the best system economics. By introducing the inaccurate prediction accuracy to consider the impact of short-term weather prediction on the energy system configuration optimization, this makes the model planning process closer to the actual situation, and finally improves the adaptability of the energy system to extreme weather conditions or situations with increased errors. By analyzing and clarifying the specific requirements for improving the prediction accuracy of renewable energy output in practical applications, thus avoiding excessive resource investment caused by pursuing too high prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a schematic flow chart of the rolling window optimization algorithm in the present invention for solving the energy system configuration optimization model.

[0051] Figure 2 is a schematic flow chart of the prediction accuracy calculation of the optimal capacity configuration scheme in the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0052] The technical solution of the present invention will be further specifically described below through embodiments and in conjunction with the drawings.

[0053] Embodiment 1:

[0054] A method for capacity configuration of an energy system based on inaccurate prediction of renewable energy output in this embodiment includes the following steps:

[0055] S1. Obtain the historical output data of renewable energy at each moment to construct the renewable energy baseline output curve, calculate the renewable energy output prediction curve based on the renewable energy baseline output, and establish a renewable energy output model by combining the renewable energy baseline output curve and the renewable energy output prediction curve.

[0056] Renewable energy includes wind power and photovoltaic power generation. Obtain the historical wind power and photovoltaic output data in the research area over the past years, and sample at each time node through simple random sampling to construct the renewable energy baseline output curve.

[0057] The renewable energy explored in the present invention mainly includes wind power and photovoltaic power generation. Since both of them are affected by weather and have strong uncertainty although their output forms are different, the same method is adopted for output uncertainty. Assume that the output P of renewable energy at a certain time period t a t is expressed as:

[0058] P a t =P fore t +ε fore t

[0059] where P fore t is the renewable energy baseline output, and ε fore t is the predicted value of renewable energy output and follows a normal distribution: ε fore t ~N(0,σ 2 fore ).

[0060] The renewable energy baseline output can be established based on the historical data of renewable energy in the research area. Specifically, obtain the historical wind power and photovoltaic output data in the research area over the past N years through satellite remote sensing data or ground monitoring stations. The historical renewable energy output value at the t-th period in the n-th year is P a t,n ~(δ t1 ,δ t2 ), where δ t1 and δ t2 respectively represent the upper and lower limits of the historical renewable energy output data.

[0061] According to the historical output data of wind power and photovoltaic power generation in each year, sample at each time node in each year through simple random sampling to construct the renewable energy baseline output curve,

[0062] P foret ∈(δ t1 , δ t2 ), t ∈ (1, T)

[0063] where T is the annual time length.

[0064] During the actual operation, operators generally determine the action amount of the key operation parameters of the energy system through short-term weather prediction values. The additional error caused by the inaccuracy of renewable energy output prediction usually generated by short-term weather prediction errors is denoted as ε fore t , which follows a normal distribution: ε fore t ~N(0, σ 2 fore ).

[0065] where σ 2 fore is the prediction accuracy factor, which is used to adjust the prediction error range. The renewable energy output basic value is adjusted through the renewable energy output prediction, and then the energy system scheduling action is adjusted until the adjustment.

[0066] The specific process of constructing the renewable energy output prediction curve includes initializing the prediction accuracy of the renewable energy output and obtaining the fluctuation range of the renewable energy output prediction. Based on the stochastic optimization method, the renewable energy benchmark output curve is corrected, and multiple renewable energy output prediction curves are generated through Monte Carlo simulation.

[0067] The final renewable energy output curve is obtained by combining the uncertainty of the renewable energy output and the inaccuracy of the renewable energy output prediction. Specifically, the renewable energy output model is the sum of the renewable energy benchmark output curve and the renewable energy output prediction curve.

[0068] S2. Establish an energy system configuration optimization model with the overall optimal operation of the energy system as the goal.

[0069] Establish an objective function with the minimum sum of the operation and maintenance investment of renewable energy equipment, the purchase investment of renewable energy equipment, the operation and maintenance investment of energy storage equipment, and the purchase investment of energy storage equipment as the goal;

[0070] Establish the constraints existing in the operation of the energy system, including the energy system supply-demand balance constraint, the stored electricity constraint of lithium batteries, and the energy storage charge-discharge power and installed capacity limit constraints.

[0071] Construct an energy system simulation model, which includes four main bodies: wind power generation, photovoltaic power generation, electrochemical energy storage, and users. Conduct configuration optimization design with the overall optimal operation of the energy system as the goal. The established objective function is:

[0072] min(F EP ) = C om EP + C co EP + C om ES + C co ES

[0073] Among them, C om EP is the operation and maintenance investment of renewable energy equipment, and C co EP is the equipment purchase investment of renewable energy, and C om ES is the operation and maintenance investment of energy storage equipment, and C co ES is the equipment purchase investment of energy storage. Renewable energy refers to wind power and photovoltaic power.

[0074] The equipment purchase investment of renewable energy is the installed capacity investment C co EP and the operation and maintenance investment C om EP The specific calculation is as follows:

[0075] C co EP = C w * c w inv + C p * c p inv

[0076] C om EP =(C w * c w om + C p * c p om ) * ∑ Y t=1 1 / (1 + r)^t

[0077] P w t = P w_fore t + ε w_fore t

[0078] P s t = P s_fore t + ε s_fore t

[0079] Among them, c w inv is the investment per unit of wind power installed capacity, c p inv is the investment per unit of photovoltaic installed capacity, C w is the wind power installed capacity, C p is the photovoltaic installed capacity, c w om is the operation and maintenance investment per unit of installed wind power, c p om is the operation and maintenance investment per unit of installed photovoltaic, P w t is the output of wind power at time t, P w_fore t is the reference output of wind power, ε w_fore t is the predicted value of wind power output, P s t is the output of photovoltaic at time t, P s_fore t is the reference output of photovoltaic output, ε s_fore t is the predicted value of photovoltaic output. r is the discount rate, and Y is the service life of the equipment.

[0080] The equipment purchase investment of energy storage, that is, the installed capacity investment C co ES and the equipment operation and maintenance investment C om ES The specific calculation is as follows:

[0081] C co ES = C es *c es inv + C x *c est inv

[0082] C om ES = C es *c es om ∑ Y t=1 1 / (1 + r)^t

[0083] Among them, C es is the energy storage installed capacity, c es inv is the investment per unit of energy storage installed capacity, c es om is the operation and maintenance investment per unit of installed energy storage, C x is the charge and discharge power of energy storage, cest inv Input the charge and discharge power of the energy storage unit.

[0084] There are constraints in the operation of the energy system. Among them, the energy supply - demand balance constraint of the energy system means that at the same moment, the sum of the wind power output, photovoltaic power output, and energy storage discharge power is equal to the sum of the energy storage charging power and the load energy consumption power. The formula is as follows:

[0085] P w t + P s t + P d t = P ch t +P l t

[0086] Among them, P w t is the output of wind power at time t, P s t is the output of photovoltaic power at time t, P d t is the discharge power of the energy storage at time t, P ch t is the charging power of the energy storage at time t, P l t is the power of the load energy consumption.

[0087] The stored - electricity constraint of the lithium - ion battery is specifically:

[0088] Under the condition of setting the energy storage charging efficiency and discharge efficiency,

[0089] The energy storage electricity at the next moment is the difference between the sum of the energy storage electricity at the previous moment and the product of the energy storage charging efficiency and the energy storage charging power, and the quotient of the energy storage discharge power and the energy storage discharge efficiency. The formula is as follows:

[0090] S t+1 = S t + η t P ch t - P d t / η f

[0091] Among them, S t+1 is the energy stored in the energy storage at time t + 1, S t is the energy stored in the energy storage at time t, η t is the energy storage charging efficiency, P ch t is the energy storage charging power at time t,d t is the energy storage discharge power at time t, and η f is the energy storage discharge efficiency.

[0092] Constraints on the energy storage charge-discharge power and installed capacity are as follows:

[0093] The energy storage charging power is within the closed interval range of 0 and the product of the maximum energy storage charge-discharge power and the energy storage charging adjustment coefficient;

[0094] The energy storage discharge power is within the closed interval range of 0 and the product of the maximum energy storage charge-discharge power and the energy storage discharge adjustment coefficient.

[0095] Among them, the values of the energy storage charging adjustment coefficient and the energy storage discharge adjustment coefficient are both 0 or 1. The formula is expressed as follows:

[0096] 0 < P ch t < X c t C x

[0097] 0 ≤ P d t ≤ X d t C x

[0098] X c t , X d t ∈ {0, 1}

[0099] Among them, C x is the maximum energy storage charge-discharge power, and X c t is the energy storage charging adjustment coefficient, and X d t is the energy storage discharge adjustment coefficient.

[0100] S3. Use the rolling window optimization algorithm to solve the energy system configuration optimization model and obtain the optimal capacity configuration plan.

[0101] In the rolling window optimization algorithm of the present invention, considering that the output of renewable energy has strong seasonality, to achieve annual cycle simulation, the annual time T is divided into K rolling windows, the length of each window, i.e., the time step, is Δt, and the rolling step size is Δt roll , and the annual time T = K * Δt. The end time t start k+1 of each rolling window is the start time t start k of the previous window start k+1 plus the rolling step size, t start k+1 = tstart k +Δt roll 。

[0102] As Figure 1 shown, it specifically includes the following steps:

[0103] S31. Based on the historical output data of renewable energy, construct N benchmark output curves of renewable energy.

[0104] Specifically, based on the historical output data of renewable energy, obtain the output range of renewable energy at each time node, and perform random sampling according to the output range to construct N benchmark output curves of renewable energy.

[0105] S32. Set the prediction accuracy of renewable energy output, and generate M renewable energy output prediction curves for each benchmark output curve of renewable energy based on the stochastic optimization method;

[0106] Specifically, on the basis of step S31, set the prediction accuracy of renewable energy output, and correct the N generated benchmark output curves of renewable energy based on the stochastic optimization method, and generate M renewable energy output prediction curves for each benchmark output curve of renewable energy through Monte Carlo simulation.

[0107] S33. Based on each benchmark output curve of renewable energy and the renewable energy output prediction curve, according to the constraint conditions, solve the energy system configuration optimization model at each time step to obtain the candidate optimal capacity configuration plan. The present invention calculates the candidate optimal capacity configuration plan for each benchmark output curve of renewable energy, combines each corresponding generated renewable energy output prediction curve, and performs the calculation at each time step until all the candidate optimal capacity configuration plans obtained by all combinations are obtained.

[0108] Specifically, on the basis of steps S31 and S32, perform the optimal configuration calculation. By initializing the installed capacities of wind power, photovoltaic, and energy storage, under the premise of considering the constraints of the energy system operation, run the parallel computing technology for the objective function of the energy system configuration optimization model, and obtain the optimal scheduling situation at each time step through Yalmip and Gurobi optimization simulations, and perform the calculation of the next time step based on this. During this process, continuously adjust the installed capacities of each device in the energy system until the simulation is completed for all time lengths and all scenarios.

[0109] S34. Select the most complex plan from all candidate optimal capacity configuration plans as the final optimal capacity configuration plan.

[0110] Since the most complex scenario represents the most extreme weather conditions, it is necessary to ensure that the energy system can operate normally under the most extreme weather conditions. Therefore, the installed capacity configuration plan of the most complex scenario is taken as the final optimal capacity configuration plan.

[0111] The present invention comprehensively considers the impacts of the uncertainty of renewable energy output and the inaccuracy of renewable energy prediction on the energy system planning, and achieves the optimal system economics while improving the operation stability of the energy system. By introducing the inaccurate prediction accuracy, the impact of short-term weather prediction on the optimization of energy system configuration is taken into account, which makes the model planning process closer to the actual situation, and finally improves the adaptability of the energy system to extreme weather conditions or scenarios with increased errors. By analyzing, the specific requirements for improving the prediction accuracy of renewable energy output in practical applications are clarified, so as to avoid excessive resource investment caused by pursuing too high prediction accuracy.

[0112] Embodiment 2:

[0113] A method for configuring the capacity of an energy system based on the inaccurate prediction of renewable energy output in this embodiment includes the following steps:

[0114] S1. Obtain the historical output data of renewable energy at each moment to construct a renewable energy benchmark output curve, calculate a renewable energy output prediction curve based on the renewable energy benchmark output, and establish a renewable energy output model by combining the renewable energy benchmark output curve and the renewable energy output prediction curve.

[0115] Renewable energy includes wind power and photovoltaic power generation. Obtain the historical wind power and photovoltaic output data in the research area over the past years, and sample at each time node through simple random sampling to construct a renewable energy benchmark output curve.

[0116] The renewable energy explored in the present invention mainly includes wind power and photovoltaic power generation. Since both of them are affected by weather and have strong uncertainty although their output forms are different, the same method is adopted for output uncertainty. Assume that the output P of renewable energy at a certain time period t a t is expressed as:

[0117] P a t =P fore t +ε fore t

[0118] where P fore t is the renewable energy benchmark output, and ε fore t is the renewable energy output prediction value, which follows a normal distribution: ε foret ~N(0, σ 2 fore ).

[0119] The baseline output of renewable energy can be established based on the historical data of renewable energy in the study area. Specifically, the historical wind power and photovoltaic output data of the study area in the past N years are obtained through satellite remote sensing data or ground monitoring stations. The historical renewable energy output value at the t-th period in the n-th year is P a t,n ~(δ t1 , δ t2 ), where δ t1 and δ t2 represent the upper and lower limits of the historical output data of renewable energy respectively.

[0120] According to the historical output data of wind power and photovoltaic power generation in each year, sampling is carried out at each time node in each year through the simple random sampling method to construct the baseline output curve of renewable energy.

[0121] P fore t ∈(δ t1 , δ t2 ), t ∈ (1, T)

[0122] where T is the annual time length.

[0123] Since in the actual operation process, operators generally determine the action amount of the key operation parameters of the energy system through short-term weather prediction values, and the additional error caused by the inaccuracy of the renewable energy output prediction usually generated by the short-term weather prediction error is represented as ε fore t , which follows a normal distribution: ε fore t ~N(0, σ 2 fore ).

[0124] where σ 2 fore is the prediction accuracy factor, used to adjust the prediction error range, adjust the basic value of renewable energy output through the renewable energy output prediction, and then until the adjustment of the energy system dispatching action.

[0125] The specific process of constructing the renewable energy output prediction curve includes initializing the prediction accuracy of the renewable energy output and obtaining the prediction fluctuation range of the renewable energy output. Based on the stochastic optimization method, the baseline output curve of renewable energy is corrected, and multiple renewable energy output prediction curves are generated from the baseline output curve of renewable energy through Monte Carlo simulation.

[0126] The final renewable energy output curve is obtained by combining the uncertainty of renewable energy output and the inaccuracy of renewable energy output prediction. Specifically, the renewable energy output model is the sum of the renewable energy baseline output curve and the renewable energy output prediction curve.

[0127] S2. Establish an energy system configuration optimization model with the goal of overall optimal operation of the energy system.

[0128] Establish an objective function with the goal of minimizing the sum of the operation and maintenance investment of renewable energy equipment, the purchase investment of renewable energy equipment, the operation and maintenance investment of energy storage equipment, and the purchase investment of energy storage equipment;

[0129] Establish the constraints existing in the operation of the energy system, including the energy system supply-demand balance constraint, the stored electricity constraint of lithium batteries, and the energy storage charge-discharge power and installed capacity limit constraint.

[0130] Construct an energy system simulation model, which includes four entities: wind power generation, photovoltaic power generation, electrochemical energy storage, and users. Conduct configuration optimization design with the goal of overall optimal operation of the energy system. The established objective function is:

[0131] min(F EP )=C om EP + C co EP + C om ES + C co ES

[0132] Where, C om EP is the operation and maintenance investment of renewable energy equipment, C co EP is the purchase investment of renewable energy equipment, C om ES is the operation and maintenance investment of energy storage, C co ES is the purchase investment of energy storage. Renewable energy refers to wind power and photovoltaic power.

[0133] The purchase investment of renewable energy equipment is the installed capacity investment C co EP and the operation and maintenance investment C om EP The specific calculation is as follows:

[0134] C co EP =C w *c w inv +C p *c pinv

[0135] C om EP =( C w *c w om +C p *c p om )*∑ Y t=1 1 / (1+r)^t

[0136] P w t =P w_fore t +ε w_fore t

[0137] P s t =P s_fore t +ε s_fore t

[0138] Among them, c w inv is the investment per unit of wind power installed capacity, c p inv is the investment per unit of photovoltaic installed capacity, C w is the wind power installed capacity, C p is the photovoltaic installed capacity, c w om is the operation and maintenance investment per unit of installed wind power, c p om is the operation and maintenance investment per unit of installed photovoltaic, P w t is the output of wind power at time t, P w_fore t is the reference output of wind power, ε w_fore t is the predicted value of wind power output, P s t is the output of photovoltaic at time t, P s_fore t is the reference output of photovoltaic, ε s_fore t is the predicted value of photovoltaic output, r is the discount rate, and Y is the service life of the equipment.

[0139] The equipment purchase investment of energy storage, that is, the installed capacity investment C co ES and the equipment operation and maintenance investment C om ES The specific calculation is as follows:

[0140] C co ES = C es *c es inv + C x *c est inv

[0141] C om ES = C es *c es om ∑ Y t=1 1 / (1 + r)^t

[0142] Wherein, C es is the installed capacity of energy storage, c es inv is the investment per unit of installed energy storage, c es om is the operation and maintenance investment per unit of installed energy storage, C x is the charge and discharge power of energy storage, c est inv is the investment per unit of charge and discharge power of energy storage.

[0143] There are constraints in the operation of the energy system. Among them, the energy supply - demand balance constraint of the energy system is that the sum of the wind power output, photovoltaic power output, and energy storage discharge power at the same moment is equal to the sum of the energy storage charging power and the load energy consumption power at the same moment. The formula is expressed as follows:

[0144] P w t + P s t + P d t = P ch t +P l t

[0145] Wherein, P w t is the output of wind power at time t, P s t is the output of photovoltaic power at time t, P d t is the discharge power of energy storage at time t, P ch t is the charging power of energy storage at time t, P l t is the power of load energy consumption.

[0146] The storage - capacity constraint of lithium - ion batteries is specifically:

[0147] When setting the energy storage charging efficiency and discharging efficiency,

[0148] The energy storage power at the latter moment is the difference between the sum of the energy storage power at the previous moment and the product of the energy storage charging efficiency and the energy storage charging power, and the quotient of the energy storage discharging power and the energy storage discharging efficiency. The formula is as follows:

[0149] S t+1 = S t + η t P ch t - P d t / η f

[0150] Among them, S t+1 is the energy stored in the energy storage at time t + 1, S t is the energy stored in the energy storage at time t, η t is the energy storage charging efficiency, P ch t is the energy storage charging power at time t, P d t is the energy storage discharging power at time t, η f is the energy storage discharging efficiency.

[0151] Constraints on the energy storage charging and discharging power and the installed capacity are as follows:

[0152] The energy storage charging power is in the closed interval range of 0 and the product of the maximum energy storage charging and discharging power and the energy storage charging adjustment coefficient;

[0153] The energy storage discharging power is in the closed interval range of 0 and the product of the maximum energy storage charging and discharging power and the energy storage discharging adjustment coefficient.

[0154] Among them, the values of the energy storage charging adjustment coefficient and the energy storage discharging adjustment coefficient are both 0 or 1. The formula is as follows:

[0155] 0 < P ch t < X c t C x

[0156] 0 ≤ P d t ≤ X d t C x

[0157] X c t , X d t ∈ {0, 1}

[0158] Among them, C x is the maximum charge and discharge power of energy storage, and X c t is the charge regulation coefficient of energy storage, and X d t is the discharge regulation coefficient of energy storage.

[0159] S3. Use the rolling window optimization algorithm to solve the energy system configuration optimization model, obtain the optimal capacity configuration plan, and calculate the prediction accuracy of the optimal capacity configuration plan.

[0160] For the rolling window optimization algorithm of the present invention, considering that the output of renewable energy has strong seasonality, in order to achieve annual cycle simulation, the annual time T is divided into K rolling windows, the length of each window, i.e., the time step, is Δt, and the rolling step size is Δt roll , and the annual time T = K * Δt. The end time t start k+1 of each rolling window is the start time t start k of the previous window plus the rolling step size, and t start k+1 = t start k + Δt roll .

[0161] As Figure 2 shown, it specifically includes the following steps:

[0162] S31. Based on the historical output data of renewable energy, construct N renewable energy reference output curves.

[0163] Specifically, based on the historical output data of renewable energy, obtain the output range of renewable energy at each time node, and perform random sampling according to the output range to construct N renewable energy reference output curves.

[0164] S32. Set the prediction accuracy p of the renewable energy output, and generate M renewable energy output prediction curves for each renewable energy reference output curve based on the stochastic optimization method;

[0165] Specifically, on the basis of step S31, set the prediction accuracy of the renewable energy output, and correct the N generated renewable energy reference output curves based on the stochastic optimization method, and generate M renewable energy output prediction curves for each renewable energy reference output curve through Monte Carlo simulation.

[0166] S33. Based on each renewable energy reference output curve and renewable energy output prediction curve, according to the constraint conditions, solve the energy system configuration optimization model at each time step to obtain the candidate optimal capacity configuration plan.

[0167] Specifically, based on steps S31 and S32, optimal configuration calculation is carried out. By initializing the installed capacities of wind power, photovoltaic, and energy storage, and considering the constraints of the operation of the energy system, the parallel computing technology of the objective function of the energy system configuration optimization model is run. Through Yalmip and Gurobi optimization simulations, the optimal scheduling situation in each time step is obtained, and based on this, the calculation of the next time step is carried out. During this process, the installed capacities of various devices in the energy system are continuously adjusted until the simulation is completed for all time lengths and all scenarios.

[0168] S34. Select the most complex scheme from all candidate optimal capacity configuration schemes as the final optimal capacity configuration scheme.

[0169] Since the most complex scheme represents the most extreme weather conditions, it is necessary to ensure that the energy system can operate normally under the most extreme weather conditions. Therefore, the installed capacity configuration scheme of the most complex scheme is used as the final optimal capacity configuration scheme.

[0170] S35. Set the capacity configuration change threshold Q.

[0171] Judge whether the optimal capacity configuration change amount is not less than the capacity configuration change threshold, specifically, judge whether 1 - F(p) / F(p - 1)≥Q is satisfied.

[0172] If not, output the prediction accuracy p - 1.

[0173] If so, increment the prediction accuracy p = p + 1%, and return to the step of generating M renewable energy output prediction curves, that is, step S32, and repeat steps S32 - S35 until the obtained optimal capacity configuration satisfies 1 - F(p) / F(p - 1)≥Q to obtain the final optimal capacity configuration, and output the prediction accuracy p - 1.

[0174] The present invention can determine the best demand for the prediction accuracy of renewable energy output in the future energy system. Based on the above rolling window optimization algorithm to solve the energy system configuration optimization model to obtain the final optimal capacity configuration, the prediction accuracy of renewable energy output is improved, and cyclic parallel operations are carried out to obtain the operation situation of the energy system after improving the prediction accuracy of renewable energy. Until the change amount of the optimal operation situation of the system before and after improving the prediction accuracy is less than the capacity configuration threshold Q, it indicates that the effect of improving the prediction accuracy of renewable energy will be very limited, thus proving that the prediction accuracy of renewable energy output in the case of p - 1 can meet the daily production planning requirements.

[0175] The specific embodiments described herein are merely illustrative of the spirit of the present invention. Those skilled in the art to which the present invention pertains may make various modifications or supplements to the described specific embodiments or use similar means for substitution, but will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

[0176] Although terms such as renewable energy baseline output curve, renewable energy output prediction curve, renewable energy output model, energy system configuration optimization model, and optimal capacity configuration are used more frequently herein, the possibility of using other terms is not excluded. The use of these terms is only for more conveniently describing and explaining the essence of the present invention; interpreting them as any additional limitation is contrary to the spirit of the present invention.

Claims

1. A method for configuring the capacity of an energy system based on the inaccurate prediction of renewable energy output, characterized in that It includes the following steps: Obtain the historical output data of renewable energy at each moment to construct a renewable energy baseline output curve, calculate a renewable energy output prediction curve based on the renewable energy baseline output, and establish a renewable energy output model by combining the renewable energy baseline output curve and the renewable energy output prediction curve; Establish an energy system configuration optimization model with the goal of overall optimal operation of the energy system; Use a rolling window optimization algorithm to solve the energy system configuration optimization model and obtain an optimal capacity configuration plan, including: Based on the historical output data of renewable energy, construct N renewable energy baseline output curves; Set the prediction accuracy of renewable energy output, and generate M renewable energy output prediction curves for each renewable energy baseline output curve based on a stochastic optimization method; Based on each renewable energy baseline output curve and renewable energy output prediction curve, and according to the constraint conditions, solve the energy system configuration optimization model at each time step to obtain a candidate optimal capacity configuration plan; Select the most complex plan from all candidate optimal capacity configuration plans as the final optimal capacity configuration plan.

2. The method for configuring the capacity of an energy system based on inaccurate prediction of renewable energy output according to claim 1, characterized in that: Renewable energy includes wind power and photovoltaic power generation. Obtain the historical wind power and photovoltaic output data of the research area in the past years, and sample at each time node through simple random sampling to construct a renewable energy baseline output curve.

3. The method for configuring the capacity of an energy system based on inaccurate prediction of renewable energy output according to claim 2, characterized in that: Based on the renewable energy baseline output curve, set the prediction accuracy of renewable energy output, and generate a renewable energy output prediction curve through a stochastic optimization method.

4. The method for configuring the capacity of an energy system based on inaccurate prediction of renewable energy output according to claim 1 or 2 or 3, characterized in that: The renewable energy output model is the sum of the renewable energy baseline output curve and the renewable energy output prediction curve.

5. The method for configuring the capacity of an energy system based on the inaccurate prediction of renewable energy output according to claim 1, characterized in that, Establish an energy system configuration optimization model with the goal of overall optimal operation of the energy system, including: Establish an objective function with the goal of minimizing the sum of the operation and maintenance investment of renewable energy equipment, the purchase investment of renewable energy equipment, the operation and maintenance investment of energy storage equipment, and the purchase investment of energy storage equipment; Establish the constraints existing in the operation of the energy system, including the energy system supply-demand balance constraint, the stored electricity constraint of lithium batteries, and the energy storage charge-discharge power and installed capacity limit constraint.

6. The method for configuring the capacity of an energy system based on inaccurate prediction of renewable energy output according to claim 5, characterized in that, The energy system supply-demand balance constraint includes: The sum of the wind power output, photovoltaic power generation output, and energy storage discharge power at the same moment is equal to the sum of the energy storage charging power and the load energy consumption power at the same moment; Among them, the wind power output and photovoltaic power generation output are calculated according to the renewable energy output model.

7. The method for configuring the capacity of an energy system based on the inaccurate prediction of renewable energy output according to claim 5, wherein The stored electricity constraint of lithium batteries includes: Under the set energy storage charging efficiency and discharge efficiency, The energy storage electricity at the next moment is the difference between the sum of the energy storage electricity at the previous moment and the product of the energy storage charging efficiency and the energy storage charging power, and the quotient of the energy storage discharge power and the energy storage discharge efficiency.

8. The method for configuring the capacity of an energy system based on the inaccurate prediction of renewable energy output according to claim 5, characterized in that, The energy storage charge-discharge power and installed capacity limit constraint includes: The energy storage charging power is within the closed interval range of 0 and the product of the maximum charge-discharge power of the energy storage and the energy storage charging regulation coefficient; The energy storage discharging power is within the closed interval range of 0 and the product of the maximum charge-discharge power of the energy storage and the energy storage discharging regulation coefficient.

9. The method for configuring the capacity of an energy system based on inaccurate prediction of renewable energy output according to claim 1, characterized in that It also includes the steps for calculating the prediction accuracy of the optimal capacity configuration scheme, including: Setting a capacity configuration change threshold, Judging whether the change amount of the optimal capacity configuration is not less than the capacity configuration change threshold, If not, outputting the prediction accuracy; If so, increasing the prediction accuracy, returning to the step of generating M renewable energy output prediction curves, continuing to execute the subsequent steps until the final optimal capacity configuration is obtained, and repeating the prediction accuracy calculation until the prediction accuracy is output.

Images