Virtual power plant real-time online scheduling method considering photovoltaic output nonlinear randomness

Abstract

The invention discloses a virtual power plant real-time online scheduling method considering photovoltaic output nonlinear randomness. The method specifically comprises the following steps: constructing a photovoltaic output prediction model; constructing an air conditioner cooling load model; constructing a virtual power plant optimization scheduling model; and solving the virtual power plant optimization scheduling model in real time by adopting a rolling optimization algorithm. The method adopts a nonlinear stochastic differential equation theory to predict solar radiation passing through the atmosphere, can consider the stochastic nonlinear characteristic of photovoltaic night output, and can further improve the accuracy of a scheduling result.

Description

Technical Field

[0001] The present invention belongs to the technical field of power plant dispatching, and in particular relates to a real-time online dispatching method for a virtual power plant taking into account the nonlinear randomness of photovoltaic output. Background Art

[0002] Distributed photovoltaics have the characteristics of randomness and small capacity. Their large-scale grid connection will bring huge challenges to the dispatching and control of the distribution network. The distribution network's ability to absorb photovoltaic power generation has become the main factor restricting the utilization rate of photovoltaic power generation systems in Henan Province. On the other hand, the scale of energy demand on the load side is growing rapidly. Virtual power plants, which include distributed power sources, energy storage systems and flexible loads, are an effective technical means to achieve a balance between supply and demand in the distribution network. By optimizing the dispatching of virtual power plants and coordinating the dispatchable generators on the power supply side and the air conditioning and refrigeration load demand on the load side, it is possible to improve energy utilization efficiency and the grid's ability to absorb photovoltaic power generation.

[0003] The existing virtual power plant optimization scheduling methods that take into account the randomness of photovoltaic power generation and the flexibility of air conditioning loads mainly have the following problems and shortcomings:

[0004] 1. Some methods have designed day-ahead optimization scheduling strategies for virtual power plants. These methods rely on rough day-ahead forecast information of PV output and load-side demand, which can easily lead to the phenomenon that the optimization results are inconsistent with the actual operation of the system.

[0005] 2. Some methods have also designed intraday optimization scheduling strategies for virtual power plants. These methods use linear regression methods to make real-time predictions of PV output and load peak and valley changes. When solar radiation is highly random, the prediction accuracy is poor and cannot reflect the random nonlinear characteristics of PV output, such as PV output at night.

[0006] 3. Some virtual power plant optimization scheduling methods consider the effect of air conditioning load participating in demand response. Such methods do not consider the impact of changes in building exterior wall temperature on indoor temperature, which reduces the accuracy of scheduling results. Summary of the invention

[0007] In order to address the deficiencies of the prior art, the present invention aims to provide a real-time online scheduling method for a virtual power plant taking into account the nonlinear randomness of photovoltaic output. The method adopts the theory of nonlinear stochastic differential equations to predict the solar radiation passing through the atmosphere, which can take into account the random nonlinear characteristics of photovoltaic nighttime output and further improve the accuracy of the scheduling results.

[0008] In order to achieve the above object, the technical solution adopted by the present invention is:

[0009] A real-time online scheduling method for a virtual power plant taking into account the nonlinear randomness of photovoltaic output is implemented based on the following steps:

[0010] S1. Construct a photovoltaic output prediction model: construct a nonlinear stochastic differential equation model of solar radiation passing through the atmosphere, use weather forecast information as the input of the prediction model; then use the extended Kalman filter to estimate the solar radiation; determine the parameters of the prediction model through likelihood estimation;

[0011] S2. Construct air conditioning cooling load model: Use the second-order equivalent thermal parameter model to model the temperature inside the building when the air conditioner is cooling;

[0012] S3. Considering the gas turbine model and the battery energy storage system model, taking the minimization of the virtual power plant operation cost as the optimization goal, a virtual power plant optimization scheduling model is constructed;

[0013] S4. Use the rolling optimization algorithm to solve the virtual power plant optimization scheduling model in real time.

[0014] Step S1 specifically includes the following steps:

[0015] S1-1. Constructing a model of solar radiation passing through the atmosphere

[0016] The photovoltaic output forecast can be expressed as:

[0017]

[0018] A(k+1)=A(k)+θ(μ 2 -A(k))dt+σ 2 dB 2 (2)

[0019] P solar (k) = γMX(k) (3)

[0020] Where: k represents discrete time; X(k) represents the ratio of the actual value of solar radiation intensity to its maximum value M; γ represents radiation efficiency; N(k) represents the input of solar radiation forecast data; A(k) represents the speed at which X(k) converges to N(k); P solar (k) represents the observed value of solar radiation; δ represents a small constant; B 1 and B 2 represents mutually independent standard Wiener processes; dt represents the micro-increment of time; γ, β, θ, μ 1 , σ 1 , μ 2 , σ 2 ,ω 1 and ω 2 Indicates parameters to be determined;

[0021] S1-2. Constructing a solar radiation estimation model

[0022] Estimated solar radiation intensity It can be expressed as

[0023]

[0024] Where: represents the predicted value of solar radiation at time k; express variance estimate; represents the predicted value of solar radiation intensity at time k; express variance estimate; represents the predicted value of solar radiation intensity at time k+1; express Variance estimation value; M represents the maximum solar radiation intensity; K represents the Kalman gain; L represents the differential operator g=σ 1 X(k)(1-X(k)); (·) T represents transpose;

[0025] S1-3. Estimation of the parameters to be determined in the model of equations (1)-(4) θ = (γ, β, θ, μ 1 ,σ 1 ,μ 2 ,σ 2 ,ω 1 ,ω 2 )

[0026] The parameters to be determined in the model of formula (1)-(4) are θ = (γ, β, θ, μ 1 ,σ 1 ,μ 2 ,σ 2 ,ω 1 ,ω 2 ) can be obtained by solving the following optimization problem:

[0027]

[0028] Where: F(·) represents the likelihood function; N represents solar radiation o observations; Represents the initial observation of solar radiation.

[0029] Step S2 specifically includes:

[0030] The second-order equivalent thermal parameter model is expressed as

[0031]

[0032] Where: T in Indicates indoor temperature; T out Indicates outdoor temperature; Tw Indicates the wall temperature; C in Represents the specific heat capacity of indoor air; C w is the specific heat capacity of the wall; R in Indicates the indoor air heat loss coefficient; R w Indicates the wall heat loss coefficient; Q a Indicates the amount of heat absorbed by the indoor air; P ac Indicates the rated power of the air conditioner; η cop Indicates the energy efficiency ratio of the air conditioner; μ ac Equal to 0 or 1, indicating that the air conditioner is turned off or on respectively; Q w Indicates the portion of indoor heat absorbed by the wall;

[0033] By participating in demand response through air conditioning load, the indoor temperature meets the constraints:

[0034] 18≤T in ≤22 (7)

[0035] Formula (7) indicates that the optimal scheduling of the virtual power plant keeps the indoor temperature change within the range of 18°C ​​to 22°C;

[0036] In addition, the operating cost of the i-th air conditioner at sampling time k is

[0037]

[0038] Where: k represents the discrete moment; P i ac Indicates the rated power of the air conditioner; π ac Indicates the electricity price, Equal to 0 or 1, indicating that the air conditioner is turned off or on.

[0039] Step S3 specifically includes:

[0040] For gas turbine models:

[0041] A gas turbine is a device that converts natural gas into electrical energy. The cost of the i-th gas turbine at sampling time k can be expressed as:

[0042]

[0043] Where: C gas_start Indicates the startup cost; The value is 0 or 1, indicating that the gas turbine is shut down or on, respectively; C gas_op Indicates the operating cost; Indicates the gas turbine output power; C gas_fuel represents the natural gas price; η gas represents the gas turbine conversion efficiency;

[0044] In addition, the gas turbine needs to meet the constraints:

[0045]

[0046]

[0047]

[0048] Where: and They represent the minimum and maximum output power of the gas turbine respectively; The value is 1 or 0, indicating that the gas turbine is shut down or started, respectively; Indicates the start and stop state variable of the gas turbine; The value is 1 or 0, indicating that the gas turbine is turned on or off, respectively;

[0049] For the battery energy storage system model:

[0050] At sampling time k, the i-th battery energy storage system satisfies the constraint:

[0051]

[0052]

[0053]

[0054] Where: The value is 0 or 1, indicating the battery discharge state or charging state respectively; The value is 0 or 1, indicating the battery charging state or discharging state respectively; .S storage,i (k) Indicates the battery charge state; and Respectively represent the minimum and maximum values ​​of the battery state of charge; and Respectively represent the charging power and discharging power of the battery; and Respectively represent the charging efficiency and discharging efficiency of the battery; A storage,i Indicates the capacity of the battery;

[0055] The cost of the battery is

[0056]

[0057] Where: π c (k) represents the charging price at time k; π d (k) represents the discharge electricity price at time k;

[0058] Constructing a virtual power plant optimization dispatch model:

[0059] The daily dispatch cycle of the virtual power plant is 24 hours, each dispatch period is 15 minutes, and there are 96 dispatch periods in total. The dispatch target is to optimize the operation of the gas turbine, air conditioner and battery to minimize the system operation cost. The operation cost of the virtual power plant can be expressed as:

[0060]

[0061] Where: N represents the finite prediction time interval; N gas 、N ac and N storage Respectively represent the number of gas turbines, air conditioners and batteries; P t Represents the power exchanged between the virtual power plant and the external power grid; π p Indicates the electricity selling price;

[0062] The optimal dispatch of virtual power plants needs to meet the following constraints:

[0063]

[0064]

[0065] Where: P d Indicates non-flexible electrical load; Indicates the maximum value of power exchanged between the virtual power plant and the external power grid;

[0066] The virtual power plant dispatch model can be expressed as:

[0067]

[0068] st(1)-(19)(21)

[0069] Where: represents the decision variable.

[0070] Step S4 specifically includes the following steps:

[0071] S4-1. At sampling time k, according to the current measured output of the system and the constraint condition formula (21), calculate the minimum value of the optimization target formula (20) within the future time interval N, and obtain the control input sequence u k ,…u N-1 ;

[0072] S4-2. Control the first value u of the input sequence k imposed on the accused;

[0073] S4-3. At sampling time k+1, the measured output at that time is used, and the above calculation process is re-executed to finally obtain the value of the scheduling decision variable u.

[0074] The beneficial effects of the present invention are:

[0075] 1. The existing virtual power plant optimization scheduling method is mainly based on linear regression theory and Kalman filter to predict photovoltaic output, which requires feedback control to correct the prediction error and cannot reflect the photovoltaic output at night. In addition, when the randomness of solar radiation is strong, the prediction method based on linear regression theory has low accuracy. The design method proposed in the present invention uses an extended Kalman filter to estimate solar radiation and uses nonlinear stochastic differential equation theory to predict solar radiation passing through the atmosphere. It can take into account the random nonlinear characteristics of photovoltaic nighttime output and improve the estimation accuracy.

[0076] 2. When considering the impact of air conditioning cooling load participating in demand response on photovoltaic consumption, the existing virtual power plant optimization scheduling method does not consider the impact of building walls on indoor temperature, resulting in inaccurate scheduling results. The design method proposed in this invention uses a second-order thermal parameter model to model indoor temperature, taking into account the impact of building wall temperature on indoor temperature, which can further improve the accuracy of scheduling results. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 Design a flow chart for the present invention. DETAILED DESCRIPTION

[0078] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not intended to limit the scope of use of the present invention.

[0079] like Figure 1 As shown, the present invention proposes a virtual power plant real-time online scheduling method that takes into account the nonlinear randomness of photovoltaic output to solve the problem of poor accuracy caused by the linear regression method in predicting photovoltaic output in the existing scheduling method. At the same time, the method proposed by the present invention takes into account the role of air-conditioning load flexibility in promoting photovoltaic power generation and consumption, and uses a model that takes into account the temperature of the building's exterior wall to model the indoor temperature to solve the problem of inaccurate scheduling results caused by ignoring the temperature of the building's exterior wall in the existing scheduling method. The specific implementation steps are:

[0080] Step S1. Constructing a photovoltaic output prediction model. The virtual power plant real-time dispatch design method proposed in the present invention adopts the theory of nonlinear stochastic differential equations to construct a photovoltaic output prediction model, taking into account the photovoltaic output at night, and can reflect the nonlinear random characteristics of photovoltaic output.

[0081] First, a stochastic differential equation model of solar radiation passing through the atmosphere is constructed. The model uses weather forecast information as the input of the prediction model to improve the prediction accuracy; then, the extended Kalman filter is used to estimate the solar radiation; finally, the parameters of the prediction model are determined by likelihood estimation. Specifically, the following steps are included:

[0082] S1-1. Constructing a model of solar radiation passing through the atmosphere

[0083] The photovoltaic output forecast can be expressed as:

[0084]

[0085] A(k+1)=A(k)+θ(μ 2 -A(k))dt+σ 2 dB 2 (2)

[0086] P solar (k) = γMX(k) (3)

[0087] Where: k represents discrete time; X(k) represents the ratio of the actual value of solar radiation intensity to its maximum value M; γ represents radiation efficiency; N(k) represents the input of solar radiation forecast data; A(k) represents the speed at which X(k) converges to N(k); P solar (k) represents the observed value of solar radiation; δ represents a small constant; B 1 and B 2 represents mutually independent standard Wiener processes; dt represents the micro-increment of time; γ, β, θ, μ 1 , σ 1 , μ 2 , σ 2 ,ω 1 and ω 2 Indicates parameters to be determined;

[0088] S1-2. Constructing a solar radiation estimation model

[0089] Estimated solar radiation intensity It can be expressed as

[0090]

[0091] Where: represents the predicted value of solar radiation at time k; express variance estimate; represents the predicted value of solar radiation intensity at time k; express variance estimate; represents the predicted value of solar radiation intensity at time k+1; express Variance estimation value; M represents the maximum solar radiation intensity; K represents the Kalman gain; L represents the differential operator g=σ 1X(k)(1-X(k)); (·) T represents transpose;

[0092] S1-3. Estimation of the parameters to be determined in the model of equations (1)-(4) θ = (γ, β, θ, μ 1 ,σ 1 ,μ 2 ,σ 2 ,ω 1 ,ω 2 )

[0093] The parameters to be determined in the model of formula (1)-(4) are θ = (γ, β, θ, μ 1 ,σ 1 ,μ 2 ,σ 2 ,ω 1 ,ω 2 ) can be obtained by solving the following optimization problem:

[0094]

[0095] Where: F(·) represents the likelihood function; N represents solar radiation o observations; Represents the initial observation of solar radiation.

[0096] Step S2. Construct an air conditioning cooling load model. The second-order equivalent thermal parameter model is used to model the temperature inside the building during air conditioning cooling, which can reflect the impact of the building's exterior wall temperature on the building's interior temperature and improve the accuracy of the scheduling results. Specifically, it includes:

[0097] The second-order equivalent thermal parameter model is expressed as

[0098]

[0099] Where: T in Indicates indoor temperature; T out Indicates outdoor temperature; T w Indicates the wall temperature; C in Represents the specific heat capacity of indoor air; C w is the specific heat capacity of the wall; R in Indicates the indoor air heat loss coefficient; R w Indicates the wall heat loss coefficient; Q a Indicates the amount of heat absorbed by the indoor air; P ac Indicates the rated power of the air conditioner; η cop Indicates the energy efficiency ratio of the air conditioner; μ ac Equal to 0 or 1, indicating that the air conditioner is turned off or on respectively; Q w Indicates the portion of indoor heat absorbed by the wall.

[0100] By participating in demand response through air conditioning load, the indoor temperature meets the constraints:

[0101] 18≤T in ≤22 (7)

[0102] Formula (7) indicates that the optimal scheduling of the virtual power plant keeps the indoor temperature change within the range of 18℃ to 22℃.

[0103] In addition, the operating cost of the i-th air conditioner at sampling time k is

[0104]

[0105] Where: P i ac Indicates the rated power of the air conditioner; π ac (k) represents the electricity price at time k, Equal to 0 or 1, indicating that the air conditioner is turned off or on.

[0106] Step S3. Based on steps S1 and S2, the gas turbine model and the battery energy storage system model are considered, and the virtual power plant optimization scheduling model is constructed with the minimum operation cost of the virtual power plant as the optimization goal. Specifically, it includes:

[0107] (1) For gas turbine models:

[0108] A gas turbine is a device that converts natural gas into electrical energy. The cost of the i-th gas turbine at sampling time k can be expressed as:

[0109]

[0110] Where: C gas_start Indicates the startup cost; The value is 0 or 1, indicating that the gas turbine is shut down or on, respectively; C gas_op Indicates the operating cost; Indicates the gas turbine output power; C gas_fuel represents the natural gas price; η gas Represents the gas turbine conversion efficiency.

[0111] In addition, the gas turbine needs to meet the constraints:

[0112]

[0113]

[0114]

[0115] Where: and They represent the minimum and maximum output power of the gas turbine respectively; The value is 1 or 0, indicating that the gas turbine is shut down or started, respectively; The value is 1 or 0, indicating that the gas turbine is turned on or off, respectively; Indicates the start and stop state variable of the gas turbine; The value is 1 or 0, indicating that the gas turbine is turned on or off.

[0116] (2) For the battery energy storage system model:

[0117] At sampling time k, the i-th battery energy storage system satisfies the constraint:

[0118]

[0119]

[0120]

[0121] Where: The value is 0 or 1, indicating the battery discharge state or charging state respectively; The value is 0 or 1, indicating the battery charging state or discharging state respectively; S storage,i (k) Indicates the battery charge state; and Respectively represent the minimum and maximum values ​​of the battery state of charge; and Respectively represent the charging power and discharging power of the battery; and Respectively represent the charging efficiency and discharging efficiency of the battery; A storage,i Indicates the capacity of the battery.

[0122] (3) The cost of the battery is

[0123]

[0124] Where: π c (k) represents the charging price at time k; π d (k) represents the discharge electricity price at time k.

[0125] (4) Constructing a virtual power plant optimization dispatch model:

[0126] The virtual power plant has a daily dispatch cycle of 24 hours, each dispatch period is 15 minutes, and there are 96 dispatch periods in total. The dispatch goal is to optimize the operation of gas turbines, air conditioners and batteries, so as to minimize the system operation cost. The operating cost of the virtual power plant can be expressed as:

[0127]

[0128] Where: N represents the finite prediction time interval; N gas 、N ac and N storage Respectively represent the number of gas turbines, air conditioners and batteries; P t Represents the power exchanged between the virtual power plant and the external power grid; π p Indicates the electricity selling price.

[0129] The optimal dispatch of virtual power plants needs to meet the following constraints:

[0130]

[0131]

[0132] Where: P d Indicates non-flexible electrical load; Indicates the maximum value of power exchanged between the virtual power plant and the external power grid.

[0133] The virtual power plant dispatch model can be expressed as:

[0134]

[0135] st(1)-(19)(21)

[0136] Where: represents the decision variable.

[0137] Step S4. Solve the optimization scheduling model in real time. The design method proposed in the present invention uses a rolling optimization algorithm to solve the virtual power plant optimization scheduling model described by equations (20)-(21). The algorithm calculation process can be expressed as:

[0138] S4-1. At sampling time k, according to the current measured output of the system and the constraint condition formula (21), calculate the minimum value of the optimization target formula (20) within the future time interval N, and obtain the control input sequence u k ,…u N-1 ;

[0139] S4-2. Control the first value u of the input sequence k imposed on the accused;

[0140] S4-3. At sampling time k+1, the measured output at that time is used, and the above calculation process is re-executed to finally obtain the value of the scheduling decision variable u.

[0141] The present invention provides a virtual power plant optimization scheduling method that takes into account the nonlinear randomness of photovoltaic output, which can take into account the random nonlinear characteristics of photovoltaic night output and improve the estimation accuracy. In addition, the design method proposed by the present invention also takes into account the influence of air conditioning cooling load participating in demand response on photovoltaic consumption, and adopts a model that takes into account the influence of building wall temperature on building temperature to model, which further improves the accuracy of scheduling results.

[0142] To further illustrate the technical effect of the present invention, a typical virtual power plant is used as an example to explain the actual effect of the present invention. The virtual power plant contains 1 gas turbine, 1 photovoltaic generator set, 1 battery energy storage system and 26 air conditioners. The maximum output of the gas turbine is 600kW, the startup cost coefficient is 0.1, the operating cost coefficient is 0.036, the efficiency is 0.95, and the natural gas price is 0.15 USD / m 3 The battery capacity is 2MWh, the charge / discharge power is 200kW, and the charge / discharge efficiency is 0.95. The rated power of the air conditioner is 28.7kW, the energy efficiency ratio is 3.52, the equivalent specific heat capacity of indoor air is 5086.3kJ / ℃, the air heat transfer coefficient is 0.612, the wall thermal conductivity is 0.3, and the equivalent specific heat capacity of the wall is 208820kJ / ℃.

[0143] In order to verify the effectiveness of the online optimization scheduling method for the virtual power plant proposed in the present invention, a comparative analysis was conducted between the method proposed in the present invention and the existing online scheduling method for predicting photovoltaic output based on linear regression theory.

[0144] When the photovoltaic output prediction based on linear regression theory is used for optimal scheduling, the total operating cost of the virtual power plant is US$2412.4. When the online optimal scheduling method proposed in the present invention is used, the total operating cost of the virtual power plant is US$2813.5, an increase of 16.6%. Therefore, the existing virtual power plant optimization scheduling method cannot fully reflect the impact of photovoltaic output uncertainty; while the method proposed in the present invention can reflect the impact of photovoltaic output uncertainty on the operating cost of the virtual power plant, thereby revealing the operating risks brought to virtual power plant operators by the randomness of photovoltaic output.

[0145] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A real-time online scheduling method for a virtual power plant taking into account the nonlinear randomness of photovoltaic output, characterized by: The method is implemented based on the following steps: S1. Construct a photovoltaic output prediction model: construct a nonlinear stochastic differential equation model of solar radiation passing through the atmosphere, use weather forecast information as the input of the prediction model; then use the extended Kalman filter to estimate the solar radiation; determine the parameters of the prediction model through likelihood estimation; S2. Construct air conditioning cooling load model: Use the second-order equivalent thermal parameter model to model the temperature inside the building when the air conditioner is cooling; S3. Considering the gas turbine model and the battery energy storage system model, taking the minimization of the virtual power plant operation cost as the optimization goal, a virtual power plant optimization scheduling model is constructed; S4. Use the rolling optimization algorithm to solve the virtual power plant optimization scheduling model in real time.

2. The real-time online scheduling method for a virtual power plant taking into account the nonlinear randomness of photovoltaic output according to claim 1 is characterized by: Step S1 specifically includes the following steps: S1-1. Constructing a model of solar radiation passing through the atmosphere The photovoltaic output forecast can be expressed as: A(k+1)=A(k)+θ(μ2-A(k))dt+σ2dB2 (2) P solar (k)=γMX(k) (3) Where: k represents discrete time; X(k) represents the ratio of the actual value of solar radiation intensity to its maximum value M; γ represents radiation efficiency; N(k) represents the input of solar radiation forecast data; A(k) represents the speed at which X(k) converges to N(k); P solar (k) represents the observed value of solar radiation; δ represents a small constant; B1 and B2 represent independent standard Wiener processes; dt represents a micro-increment of time; γ, β, θ, μ1, σ1, μ2, σ2, ω1 and ω2 represent parameters to be determined; S1-2. Constructing a solar radiation estimation model Estimated solar radiation intensity It can be expressed as Where: represents the predicted value of solar radiation at time k; express variance estimate; represents the predicted value of solar radiation intensity at time k; express variance estimate; represents the predicted value of solar radiation intensity at time k+1; express Variance estimation value; M represents the maximum solar radiation intensity; K represents the Kalman gain; L represents the differential operator g = σ1X(k)(1-X(k)); (·) T represents transpose; S1-3. Estimation of the parameters to be determined in the model of equations (1)-(4) θ = (γ, β, θ, μ1, σ1, μ2, σ2, ω1, ω2) The parameters to be determined in the model of formulas (1)-(4) can be obtained by solving the following optimization problem: Where: F(·) represents the likelihood function; N represents solar radiation o observations; Represents the initial observation of solar radiation.

3. The real-time online scheduling method for a virtual power plant taking into account the nonlinear randomness of photovoltaic output according to claim 1 is characterized in that: Step S2 specifically includes: The second-order equivalent thermal parameter model is expressed as Where: T in Indicates indoor temperature; T out Indicates outdoor temperature; T w Indicates the wall temperature; C in Represents the specific heat capacity of indoor air; C w is the specific heat capacity of the wall; R in Indicates the indoor air heat loss coefficient; R w Indicates the wall heat loss coefficient; Q a Indicates the amount of heat absorbed by the indoor air; P ac Indicates the rated power of the air conditioner; η cop Indicates the energy efficiency ratio of the air conditioner; μ ac Equal to 0 or 1, indicating that the air conditioner is turned off or on respectively; Q w Indicates the portion of indoor heat absorbed by the wall; By participating in demand response through air conditioning load, the indoor temperature meets the constraints: 18≤T in ≤22 (7) Formula (7) indicates that the optimal scheduling of the virtual power plant keeps the indoor temperature change within the range of 18°C ​​to 22°C; In addition, the operating cost of the i-th air conditioner at sampling time k is Where: k represents discrete time; P i ac Indicates the rated power of the air conditioner; π ac Indicates the electricity price, Equal to 0 or 1, indicating that the air conditioner is turned off or on.

4. The real-time online scheduling method for a virtual power plant taking into account the nonlinear randomness of photovoltaic output according to claim 3 is characterized by: Step S3 specifically includes: For gas turbine models: A gas turbine is a device that converts natural gas into electrical energy. The cost C of the i-th gas turbine at sampling time k is i gas It can be expressed as: Where: C gas_start Indicates the startup cost; The value is 0 or 1, indicating that the gas turbine is shut down or on, respectively; C gas_op Indicates the operating cost; Indicates the gas turbine output power; C gas_fuel represents the natural gas price; η gas represents the gas turbine conversion efficiency; In addition, the gas turbine needs to meet the constraints: Where: and They represent the minimum and maximum output power of the gas turbine respectively; The value is 1 or 0, indicating that the gas turbine is shut down or started, respectively; Indicates the start and stop state variable of the gas turbine; The value is 1 or 0, indicating that the gas turbine is turned on or off, respectively; For the battery energy storage system model: At sampling time k, the i-th battery energy storage system satisfies the constraint: Where: The value is 0 or 1, indicating the battery discharge state or charging state respectively; The value is 0 or 1, indicating the battery charging state or discharging state respectively; S storage,i (k) Indicates the battery charge state; and Respectively represent the minimum and maximum values ​​of the battery state of charge; and Respectively represent the charging power and discharging power of the battery; and Respectively represent the charging efficiency and discharging efficiency of the battery; A storage,i Indicates the capacity of the battery; The cost of the battery is Where: π c (k) represents the charging price at time k; π d (k) represents the discharge electricity price at time k; Constructing a virtual power plant optimization dispatch model: The daily dispatch cycle of the virtual power plant is 24 hours, each dispatch period is 15 minutes, and there are 96 dispatch periods in total. The dispatch target is to optimize the operation of the gas turbine, air conditioner and battery to minimize the system operation cost. The operation cost of the virtual power plant can be expressed as: Where: N represents the finite prediction time interval; N gas 、N ac and N storage Respectively represent the number of gas turbines, air conditioners and batteries; P t Represents the power exchanged between the virtual power plant and the external power grid; π p Indicates the electricity selling price; The optimal dispatch of virtual power plants needs to meet the following constraints: Where: P d Indicates non-flexible electrical load; Indicates the maximum value of power exchanged between the virtual power plant and the external power grid; The virtual power plant dispatch model can be expressed as: st(1)-(19)(21) Where: represents the decision variable.

5. The real-time online scheduling method for a virtual power plant taking into account the nonlinear randomness of photovoltaic output according to claim 4 is characterized in that: Step S4 specifically includes the following steps: S4-1. At sampling time k, according to the current measured output of the system and the constraint condition formula (21), calculate the minimum value of the optimization target formula (20) within the future time interval N, and obtain the control input sequence u k ,…u N-1 ; S4-2. Control the first value u of the input sequence k imposed on the accused; S4-3. At sampling time k+1, the measured output at that time is used, and the above calculation process is re-executed to finally obtain the value of the scheduling decision variable u.

Images