Water-wind-light system non-iterative scheduling method considering privacy protection and uncertainty

Abstract

The invention discloses a water-wind-light system non-iterative scheduling method giving consideration to privacy protection and uncertainty, and the method comprises the steps: firstly constructing a new energy uncertainty set based on a Wasserstein distance for the wind-light uncertainty in a complementary power generation system; secondly, in combination with a distribution robust optimization method and a chance constraint optimization method, a distribution robust chance constraint optimization model of the water-wind-light complementary power generation system is established; the model is converted into a single-layer mixed integer linear programming model through strong duality and other theories; then, technologies such as equivalent projection, row transformation, column transformation and affine transformation are applied, a privacy protection mechanism with a water-wind-light power generation main body, an encryption center and a cloud computing center as cores is established, non-iterative solution of the model is achieved on the premise that privacy of participants is protected, and a scheduling result of the water-wind-light complementary system is output. According to the method, non-iterative water-wind-light coordinated optimization scheduling is realized, and privacy protection of participants and uncertainty of new energy output are considered at the same time.

Description

Technical Field

[0001] The present invention belongs to the field of power dispatching, and relates to privacy protection and multi-energy complementarity of participating entities, and in particular to a non-iterative dispatching method for a water-wind-solar complementary system that takes into account both privacy protection and uncertainty. Background Art

[0002] As energy and environmental issues become increasingly prominent, renewable energy sources such as wind and solar have ushered in large-scale development and have gradually become one of the dominant energy sources in the power system. However, wind and solar have strong uncertainty, and direct grid connection will have an impact on the power grid and affect the robustness of the system. To this end, flexible power generation resources can be introduced to carry out multi-energy complementation, thereby solving the problem of wind and solar in the power system. Among flexible power generation resources, hydropower has the advantages of mature technology, large installed capacity, clean and pollution-free, and can serve as the main support for wind and solar grid connection. At present, the research on multi-energy complementary systems mainly focuses on two aspects:

[0003] On the one hand, there is the uncertainty problem of wind and solar power output. With the large-scale installation of wind power, photovoltaic and other power generation equipment, the impact of output uncertainty on the power grid has gradually increased, and advanced uncertainty optimization algorithms such as stochastic optimization and robust optimization have become research hotspots. Stochastic optimization generally takes the expected value as the objective function, the solution result is relatively conservative, and the probability distribution of uncertainty parameters needs to be preset. Robust optimization, on the other hand, solves the worst result in uncertain scenarios. Compared with traditional uncertainty algorithms, the distributed robust optimization algorithm based on Wasserstein distance changes the conservatism of the solution result by adjusting the confidence of the system, and the final solution model is a single-layer deterministic optimization model, which can be directly solved by commercial solvers.

[0004] On the other hand, there is the issue of privacy protection for alliance members. The proportion of wind and solar power generation in the electricity market is increasing, intensifying the competition with traditional power generation entities. In the water, wind and solar multi-energy complementary system, in order to maintain their own market advantages, effective privacy protection mechanisms have become a common need for all parties involved in water, wind and solar. In order to protect the privacy of each member of the alliance, the commonly used approach is to establish a multi-agent cooperative game model and perform distributed solution. A common method is to construct a cooperative game model based on Nash negotiation theory, convert it into two sub-problems through the mean inequality, and then use the cross-direction multiplier method for distributed solution. This type of method requires iterative information exchange between multiple subsystems, and has disadvantages such as poor convergence and weak scalability. Summary of the invention

[0005] Purpose of the invention: In order to solve the problems of wind and solar uncertainty and privacy protection of participating entities, the present invention proposes a non-iterative scheduling method for a water-wind-solar complementary system that takes into account both privacy protection and uncertainty.

[0006] Invention content: The non-iterative scheduling method of the water-wind-solar complementary system that takes into account both privacy protection and uncertainty is implemented as follows:

[0007] (1) Aiming at the uncertainty of wind and solar power in complementary power generation systems, a new energy uncertainty set based on Wasserstein distance is constructed;

[0008] (2) Combining the distributed robust optimization and chance-constrained optimization methods, a distributed robust chance-constrained optimization model for the hydro-wind-solar hybrid power generation system is established;

[0009] (3) The distributed robust chance constrained optimization model is transformed into a single-level mixed integer linear programming model through strong duality theory;

[0010] (4) Construct a privacy protection mechanism with the hydropower, wind and solar power generation entities, encryption center and cloud computing center as the core. Under the premise of protecting the privacy of the participating entities, the non-iterative solution of the model is realized and the scheduling results of the hydropower, wind and solar power complementary system are output.

[0011] Furthermore, the scenery uncertainty set based on Wasserstein distance in step (1) specifically includes:

[0012] Uncertain variables include wind power output and photovoltaic output, and the actual output is:

[0013]

[0014] Output deviation sample set They are all generated by Latin hypercube sampling. The similarity between the two distributions is quantified by the Wasserstein distance, which is used to construct the uncertainty set. The Wasserstein distance is defined as:

[0015]

[0016] Among them, the uncertainty parameters θ and ξ follow the distribution P e and P N ; Π represents the joint distribution of θ and ξ; ||·|| represents the 1-norm;

[0017] New energy uncertainty set P:

[0018] P={P∈P(Ξ):W(P,P N )≤ε(M)}

[0019] Among them, Ξ is the support set of random variables, that is, ε(M) is the radius of the Wasserstein sphere:

[0020]

[0021] Among them, β is the confidence level; μ is the sample mean.

[0022] Furthermore, the hydro-wind-solar complementary power generation system described in step (2) includes an operation model of a hydropower entity, an operation model of a wind power entity, and an operation model of a photovoltaic entity.

[0023] Furthermore, the operation model of the hydropower entity is a mixed integer linear programming model, that is, a deterministic model, and its specific form is as follows:

[0024]

[0025] s P h，k，t =η k H k Q k，t

[0026] P h，k，t =η k H k Q k，t

[0027]

[0028] V k，T =V k，end

[0029]

[0030] S k，t ≥0

[0031]

[0032] Where T is the total number of time periods; N h is the number of hydropower stations; R k is the set of upstream reservoirs of hydropower station k; ψ g P h，k，t is the set of load periods g; the output of hydropower station k in period t; Q k，t is the power generation flow of hydropower station k in period t; S k，t V is the water discharge of hydropower station k in period t; k，t is the reservoir capacity of hydropower station k at time t; E h，t is the shared electricity of the hydropower entity; t is the electricity price in period t; k is the hydropower conversion coefficient of hydropower station k; t lk is the water stagnation time; and are the upper and lower limits of the output k of the hydropower station respectively; and are the upper and lower limits of the k storage capacity of the hydropower station respectively; and are the upper and lower limits of the power generation flow of hydropower station k; W k，t is the natural water inflow of hydropower station k in period t; τ g is the proportional coefficient of the g load period.

[0033] Furthermore, the operation model of the wind power entity is a distributed robust chance-constrained optimization model, that is, an uncertainty model, and its specific form is as follows:

[0034]

[0035] P w，m，t ≥0

[0036]

[0037] Among them, N w is the number of wind power stations; P w，m，t is the output of wind power station m in period t; E w，m，t is the shared electricity of wind power station m in period t; is the actual output of wind power station m during period t; w，m，t is the prediction error of wind power station m in period t; f pen penalizing electricity prices for renewable energy; is the predicted output of wind power station m during period t.

[0038] Furthermore, the operation model of the photovoltaic main body is a distributed robust chance constrained optimization model, that is, an uncertainty model, and its specific form is as follows:

[0039]

[0040] P v，n，t ≥0

[0041]

[0042] Among them, N v is the number of photovoltaic power stations; P v，n，t is the output of photovoltaic power station n in period t; E v，n，t is the shared electricity of photovoltaic power station n in period t; is the actual output of photovoltaic power station n in period t; v，n，t is the prediction error of photovoltaic power station n in period t; f pen penalizing electricity prices for renewable energy; is the predicted output of photovoltaic power station n in period t.

[0043] Furthermore, the implementation process of step (3) is as follows:

[0044] (31) Dual transformation of the objective function of the wind power and photovoltaic main operation models:

[0045] The objective functions of the wind power and photovoltaic main operation models are unified as follows:

[0046]

[0047] In the formula, the variables x, y and z in the first stage are located on the set χ; G(x,ξ) is a linear expression about x and ξ; for the second term, the strong duality theory is used to transform it into the following form:

[0048]

[0049] According to the convex transformation theory, it is further transformed into:

[0050]

[0051] At this point, the objective function is finally transformed into:

[0052]

[0053] (32) Robust chance constrained conversion:

[0054] Due to opportunity constraints, wind power entities and photovoltaic entities are unified as follows:

[0055]

[0056] By introducing the wind and solar output error ξ t The lower bound Rewrite the chance constraint as:

[0057]

[0058] In order to obtain a tight lower bound for the wind and solar output error, the following model is constructed and solved:

[0059]

[0060] Based on CVaR and strong duality theory, this type of model is reconstructed and finally converted into:

[0061]

[0062] The lower bound of the wind and solar output error is is 0;

[0063] (33) Constructing a joint dispatching model of water, wind and solar power:

[0064] The main operation models of hydropower, wind power and photovoltaic power are transformed into compact matrix forms. The main operation model of hydropower is a mixed integer nonlinear programming problem, and the main operation models of wind power and photovoltaic power are linear programming problems. The compact models of each subject are expressed as:

[0065]

[0066] Where i is the index of the power generation entity, including hydropower stations, wind power stations and photovoltaic power stations; x i is a linear decision variable; i is an integer decision variable; z is an integer decision variable, that is, {E h，t , E w，n，t , E v，n，t};

[0067] For hydropower stations, x i = {P h，k，t , Q k,t , S k，t , V k，t}、y i = {I h，k，t}; For a wind power station, x i = {P w，m，t ,λ m ,s m，k}, For photovoltaic power stations; For photovoltaic power stations, x i = {P v，n，t ,λ n ,s n，k}, In order to achieve the multi-energy complementarity of water, wind and solar, the joint scheduling model P0 was established:

[0068]

[0069] In the formula, Ez=0 is the constraint between the interaction variables, which means that the sum of the interaction power of each subject in each period is 0, specifically:

[0070]

[0071] Furthermore, the implementation process of step (4) is as follows:

[0072] A decentralized strategy is adopted to share the system management rights among the hydropower, wind power and photovoltaic power generation entities, the encryption center and the cloud computing center; the hydropower, wind power and photovoltaic power generation entities are composed of multiple nodes, including hydropower, wind power and photovoltaic entities, and the encryption methods are elimination, column transformation, row transformation and affine transformation; the encryption center is a public asset on the user side, which can only mechanically execute pre-set instructions, and its operation and maintenance are supervised by all entities on the user side; the encryption methods are column transformation, row transformation and affine transformation; the cloud computing center solves various optimization models through the Internet and with computing resources paid by usage.

[0073] Furthermore, the optimization model is specifically solved as follows:

[0074] Preprocessing of hydropower, wind and solar power generation data: Each subject randomly selects the variable x to be eliminated i , and characterize the safety domain through FM P i Perform row transformation, that is and The transformed security domain is right Perform column transformation, that is and The transformed security domain is right Perform an affine transformation, that is and The transformed security domain is Each entity will separately The coefficient matrix E is encrypted and transmitted to the encryption center;

[0075] Encryption-centric encryption: Building the model and convert it into a compact form UP = {min A up x up +a up sB up x up +C up y up ≤ b up};in, Perform row transformation on UP, that is and The transformed model is UP-H. The column transformation of UP-H is performed, that is, and The transformed model is UP-L; UP-L is affine transformed, that is, and The transformed model Encrypt and transmit the model UP-S to the cloud computing center;

[0076] Cloud computing center computing: Cloud computing solves the UP-S model and sets the optimal solution Encrypted transmission to the encryption center;

[0077] Encryption Center Decryption: Decryption, that is in And compare the result z* with Encrypted and transmitted to the corresponding power generation entity i;

[0078] Distributed solution by participating entities: Each entity solves the model {maxω i stω i =A i x i +a i , B i x i +C i y i +D i z * ≤ b i}, that is, the solution of coordinated optimization of water, wind and solar power scheduling is completed.

[0079] Beneficial effects: Compared with the prior art, the beneficial effects of the present invention are: the water-wind-solar distributed robust chance constraint method proposed in the present invention can process the uncertainty factors in the system through scene generation, strong duality theory, etc., and the final model is a single-layer model, which is easy to solve; the proposed privacy protection mechanism can realize the solution of the single-layer distributed robust model on the basis of realizing the privacy protection of each subject; at the same time, the method does not require iterative information interaction between water, wind and solar subjects; the present invention realizes non-iterative water, wind and solar coordinated optimization scheduling, while taking into account the privacy protection of participating subjects and the uncertainty of new energy output. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 is a flow chart of the present invention;

[0081] Figure 2 This is the distribution map of the water-wind-solar complementary power generation system;

[0082] Figure 3 This is the flow chart of the privacy protection mechanism. DETAILED DESCRIPTION

[0083] The technical solution of the invention is described in detail below with reference to the accompanying drawings.

[0084] like Figure 1 The present invention proposes a non-iterative scheduling method for a water-wind-solar complementary system that takes into account both privacy protection and uncertainty, and the specific steps are as follows:

[0085] Step 1: Input the information of the water-wind-solar complementary power generation system; for the uncertainty of wind and solar in the complementary power generation system, construct a new energy uncertainty set based on Wasserstein distance, specifically:

[0086] Uncertain variables include wind power output and photovoltaic output. The actual output can be expressed as the sum of the predicted output and the output deviation, that is:

[0087]

[0088] Output deviation sample set All are generated by Latin hypercube sampling. Due to the limited number of samples, the empirical distribution obtained based on the sample set is difficult to fully reflect the true probability distribution. To solve this problem, the Wasserstein distance is used to quantify the similarity between the two distributions, and this is used to construct the uncertainty set. Among them, the Wasserstein distance is defined as:

[0089]

[0090] Among them, the uncertainty parameters θ and ξ follow the distribution P e and P N ; Π represents the joint distribution of θ and ξ; ||·|| represents the 1-norm.

[0091] The new energy uncertainty set P can be expressed as:

[0092] P={P∈P(Ξ):W(P,P N )≤ε(M)}

[0093] Among them, Ξ is the support set of random variables, that is, ε(M) is the radius of the Wasserstein sphere, and the calculation formula is:

[0094]

[0095] Among them, β is the confidence level; μ is the sample mean.

[0096] Step 2: Combine the distributed robust optimization and opportunity constrained optimization methods to establish a distributed robust opportunity constrained optimization model for the hydro-wind-solar hybrid power generation system, including:

[0097] 2.1: Construct the operation model of the hydropower entity. The model is a deterministic model, and its specific form is as follows:

[0098]

[0099] s P h，k， t=η k H kQ k，t

[0100] P h，k，t =η k H k Q k，t

[0101]

[0102] V k，T =V k，end

[0103]

[0104] S k，t ≥0

[0105]

[0106] Where T is the total number of time periods; N h is the number of hydropower stations; R k is the set of upstream reservoirs of hydropower station k; ψ g P h，k，t is the set of load periods g; the output of hydropower station k in period t; Q k，t is the power generation flow of hydropower station k in period t; S k，t V is the water discharge of hydropower station k in period t; k，t is the reservoir capacity of hydropower station k at time t; E h，t is the shared electricity of the hydropower entity; t is the electricity price in period t; k is the hydropower conversion coefficient of hydropower station k; t lk is the water stagnation time; are the upper / lower limits of the hydropower station k output respectively; are the upper / lower limits of the k storage capacity of the hydropower station, respectively; are the upper and lower limits of the power generation flow of hydropower station k; W k，t is the natural water inflow of hydropower station k in period t; τ g is the proportional coefficient of the g load period.

[0107] 2.2: Construct the operation model of the wind power entity, which is a distributed robust opportunity constrained optimization model. The specific form is as follows:

[0108]

[0109] P w，m，t ≥0

[0110]

[0111] Among them, N wis the number of wind power stations; P w，m，t is the output of wind power station m in period t; E w，m，t is the shared electricity of wind power station m in period t; is the actual output of wind power station m during period t; w，m，t is the prediction error of wind power station m in period t; f pen penalizing electricity prices for renewable energy; is the predicted output of wind power station m during period t.

[0112] 2.3: Construct the operation model of the photovoltaic entity, which is a distributed blue stick opportunity constraint optimization model, and its specific form is as follows:

[0113]

[0114] P v，n，t ≥0

[0115]

[0116] Among them, N v is the number of photovoltaic power stations; P v，n，t is the output of photovoltaic power station n in period t; E v，n，t is the shared electricity of photovoltaic power station n in period t; is the actual output of photovoltaic power station n in period t; v，n，t is the prediction error of photovoltaic power station n in period t; f pen penalizing electricity prices for renewable energy; is the predicted output of photovoltaic power station n in period t.

[0117] Step 3: Convert the distributed robust chance constrained optimization model into a single-level mixed integer linear programming model through theories such as strong duality, including:

[0118] 3.1: Dual transformation of the objective function. The specific process is:

[0119] The objective functions of the wind power and photovoltaic main operation models can be unified into the following form:

[0120]

[0121] In the formula, the variables x, y and z in the first stage are located on the set χ; G(x,ξ) is a linear expression about x and ξ. For the second term, the strong duality theory can be transformed into the following form:

[0122]

[0123] According to the convex transformation theory, it is further transformed into:

[0124]

[0125] At this point, the objective function is finally transformed into:

[0126]

[0127] 3.2: Robust opportunity constraint conversion, the specific process is:

[0128] Due to opportunity constraints, the wind power entity and the photovoltaic entity can be written uniformly as:

[0129]

[0130] First, by introducing the random variable ξ t The lower limit Rewrite the chance constraint as:

[0131]

[0132] Then, to obtain a tight lower bound on the random variable, the following model can be constructed and solved:

[0133]

[0134] Based on CVaR and strong duality theory, this type of model can be reconstructed and finally converted into:

[0135]

[0136] Note: The output of the photovoltaic body is close to 0 at night or during periods of extremely weak light, which can be ignored. Can be directly set to 0.

[0137] 3.3: Construct a joint dispatch model for water, wind and solar power. The specific process is as follows:

[0138] To facilitate further processing, the main operating models of water, wind and light are converted into a compact matrix form.

[0139] The main operation model of hydropower is a mixed integer nonlinear programming problem, and the main operation models of wind power and photovoltaic power are transformed into linear programming problems through theories such as strong duality and conditional risk value. Therefore, the compact model of each entity can be expressed as:

[0140]

[0141] Where i is the index of the power generation entity, including hydropower stations, wind power stations and photovoltaic power stations; x i is a linear decision variable; i is an integer decision variable; z is an integer decision variable, that is, {E h，t , E w，n，t , E v，n，t}.

[0142] For hydropower stations, x i = {P h，k，t , Q k，t , S k，t , V k，t}、y i = {I h，k，t}; For a wind power station, x i = {P w，m，t ,λ m ,s m，k}, For photovoltaic power stations; For photovoltaic power stations, x i = {P v，n，t ,λ n ,s n，k}, Among them, λ m / λ m 、s m，k / s n，k Compared with λ and s in the model (13), k Have similar properties.

[0143] In order to achieve the multi-energy complementarity of water, wind and solar, the joint scheduling model P0 was established:

[0144]

[0145] In the formula, Ez=0 is the constraint between the interaction variables, which means that the sum of the interaction power of each subject in each period is 0, specifically:

[0146]

[0147] Step 4: Apply equivalent projection, row transformation, column transformation, affine transformation and other technologies to establish a privacy protection mechanism with hydropower, wind and solar power generation entities, encryption centers and cloud computing centers as the core to solve the model.

[0148] The application of elimination, row transformation, column transformation and affine transformation techniques can be used to change the objective function and feasible domain of the original problem, but will not affect the final scheduling results of the model. The details are as follows:

[0149] 1) Elimination:

[0150] For the linear decision variables within the model, all or part of the variables can be eliminated through the Fourier-Motzkin elimination method to achieve the effect of privacy protection. Taking the water-wind-solar coordinated optimization scheduling model P0 as an example, the specific process is as follows:

[0151] First, each issuing entity uses the Fourier-Motzkin elimination method to characterize the security domain Φ i :

[0152]

[0153] Among them, Ω i is the feasible domain of the original problem of subject i; is a linear decision variable that does not need / needs elimination, satisfying x i ={x i，d , x i，s}.

[0154] Then, an equivalent model of the coordinated optimization scheduling model P0 is constructed:

[0155]

[0156] 2) Row transformation and column transformation:

[0157] The general linear programming problem LP0 has the following form:

[0158]

[0159] in, A 0 is a H×L matrix, that is, A 0 =(p 1 , p 2 , ..., p H ) T or (e 1 , e 2 , ..., e L );C 0 is an L-dimensional row vector, that is d 0 is a real number; b 0 is an H-dimensional column vector,

[0160] For the row (constraint) transformation of model LP0, the converted model LP0-R is:

[0161]

[0162] Among them, I H =(h1, h2, ..., hH), which is the rearrangement of the row index {1, 2, ..., H}; and Represents the matrix A 0 With b 0 The row transformation of

[0163] For the column (variable) transformation of model LP0, the converted model LP0-C is:

[0164]

[0165] Among them, I L =(l1, l2, ..., lL), which is the rearrangement of column indices {1, 2, ..., L}; and Represents the matrix C 0 With A 0 The row transformation of X L represents the new variable after the decision variables are rearranged, that is,

[0166] The column transformation of model LP0 is a reversible transformation, and the solution set X of model LP0 0，* The solution set of model LP0-C can be Perform the inverse transformation to get X 0，* Among them, the process of inverse transformation is:

[0167] 3) Affine transformation:

[0168] The new optimization problem obtained by affine transformation of the convex optimization problem is still a convex optimization problem that is easy to solve. Taking model LP0 as an example, first, the decision variable X s Make the following definitions:

[0169] X S =K -1 X 0 +Δ

[0170] Where K is an L×L real matrix; Δ is an L-dimensional column matrix.

[0171] Then, substituting the above equation into model LP0, we can get model LP0-A:

[0172]

[0173] Among them, A S =A 0 K; C S =C 0 K; b S =b 0 -A 0 KΔ;d S =d 0 -C 0 KΔ.

[0174] Combination Figure 2 The privacy protection process diagram in the water-wind-solar coordinated optimization scheduling model P0, the main solution process is as follows:

[0175] Preprocessing of main data of hydropower, wind and solar power generation:

[0176] 1) Each subject randomly selects the variables to be eliminatedx i , and characterize the safety domain through FM

[0177] 2) For Φ i Perform row transformation, that is and The transformed security domain is

[0178] 3) Yes Perform column transformation, that is and The transformed security domain is

[0179] 4) Yes Perform an affine transformation, that is and The transformed security domain is

[0180] 5) Each entity will separately The coefficient matrix E is encrypted and transmitted to the encryption center.

[0181] Encryption Center Encryption:

[0182] 1) Build the model and convert it into a compact form UP = {min A up x up +a up sB up x up +C up y up ≤ b up}.in,

[0183] 2) Perform row transformation on UP, that is and The transformed model is denoted as UP-H.

[0184] 3) Perform column transformation on UP-H, that is and The transformed model is denoted as UP-L.

[0185] 4) Perform affine transformation on UP-L, that is and The transformed model

[0186]

[0187] 5) Encrypt the model UP-S and transmit it to the cloud computing center.

[0188] Cloud computing center computing: Cloud computing solves the UP-S model and sets the optimal solution Encrypted transmission to the encryption center.

[0189] Encryption Center Decryption: Decryption, that is in And the result z * With Δ i * Encrypted and transmitted to the corresponding power generation entity i.

[0190] Distributed solution by participating entities: Each entity solves the model {maxω i stω i =A i x i +a i , B i x i +C i y i +D i z * ≤ b i}, that is, the solution of coordinated optimization of water, wind and solar power scheduling is completed.

[0191] Planning area such as Figure 3 As shown in the figure, the water-wind-solar complementary power generation system includes 3 hydropower stations, 10 wind power stations and 4 photovoltaic power stations. Among them, the installed capacity of hydropower stations H1~H3 is 600MW, 600MW and 3300MW respectively; the installed capacity of wind power stations W1~W6 is 400MW, and the installed capacity of wind power stations W8~10 is 700MW; the installed capacity of photovoltaic power stations is 300MW, 300MW, 500MW and 300MW respectively. The scheduling cycle is 1 day, the scheduling period is 1h, and the confidence of the chance constraint and the Wasserstein sphere is 0.9.

[0192] Table 1 The impact of different confidence levels on the benefits of the water-wind-solar hybrid system

[0193]

[0194]

[0195] Table 1 shows the impact of different confidence levels on the revenue of the water-wind-solar complementary system. It can be seen that as the confidence level continues to increase, that is, as the system risk continues to decrease, the revenue decreases from the initial 59.9798 million yuan to 57.6098 million yuan. In addition, as the confidence level increases from 0.50 to 0.90, the reduction in system revenue continues to increase. Among them, in the process of increasing the confidence level from 0.50 to 0.70, the system revenue decreased by 666,200 yuan, and in the process of increasing the confidence level from 0.70 to 0.90, the system revenue decreased by 1.7038 million yuan. This shows that while the system risk is reduced, it also leads to a rapid decrease in revenue. Therefore, the conservatism of the scheduling model should be appropriately set, that is, the opportunity constraint and Wasserstei n Confidence of the ball.

[0196] Table 2 Income before and after multi-energy complementarity (ten thousand yuan)

[0197]

[0198] Table 2 shows the benefits before and after the multi-energy complementarity of water, wind and solar. It can be seen that the benefits of the water, wind and solar complementary system increased by 27.80%, or 12.5301 million yuan. Among them, the benefits of hydropower stations 1 to 3 increased by 3,800 yuan; the benefits of wind power stations 1 to 10 all increased by no less than 805,100 yuan, a total increase of 36.7158 million yuan, accounting for 93.22% of the total gain; the benefits of photovoltaic power stations 1 to 4 increased by 104,100 yuan, 190,400 yuan, 304,400 yuan and 247,100 yuan respectively. This shows that the multi-energy complementarity of water, wind and solar has promoted the absorption of renewable energy in the power system, thereby increasing the benefits of each entity, especially wind power and photovoltaic entities.

[0199] The above is only an implementation case of the present invention, and its description is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of this application, which all belong to the protection scope of this application. Therefore, the protection scope of the patent of this application shall be based on the attached claims.

Claims

1. A non-iterative scheduling method for a water-wind-solar hybrid system that takes into account both privacy protection and uncertainty, characterized in that: include: (1) Aiming at the uncertainty of wind and solar power in complementary power generation systems, a new energy uncertainty set based on Wasserstein distance is constructed; (2) Combining the distributed robust optimization and chance-constrained optimization methods, a distributed robust chance-constrained optimization model for the hydro-wind-solar hybrid power generation system is established; (3) The distributed robust chance constrained optimization model is transformed into a single-level mixed integer linear programming model through strong duality theory; (4) Construct a privacy protection mechanism with the hydropower, wind and solar power generation entities, encryption center and cloud computing center as the core. Under the premise of protecting the privacy of the participating entities, the non-iterative solution of the model is realized and the scheduling results of the hydropower, wind and solar power complementary system are output.

2. The non-iterative scheduling method for a water-wind-solar complementary system taking into account privacy protection and uncertainty according to claim 1 is characterized in that: The scenery uncertainty set based on Wasserstein distance in step (1) specifically includes: Uncertain variables include wind power output and photovoltaic output, and the actual output is: Output deviation sample set They are all generated by Latin hypercube sampling. The similarity between the two distributions is quantified by the Wasserstein distance, which is used to construct the uncertainty set. The Wasserstein distance is defined as: Among them, the uncertainty parameters θ and ξ follow the distribution P e and P N ; Π represents the joint distribution of θ and ξ; ||·|| represents the 1-norm; New energy uncertainty set P: P={P∈P(Ξ):W(P,P N )≤ε(M)} Among them, Ξ is the support set of random variables, that is, ε(M) is the radius of the Wasserstein sphere: Among them, β is the confidence level and μ is the sample mean.

3. The non-iterative scheduling method for a water-wind-solar complementary system taking into account privacy protection and uncertainty according to claim 1 is characterized in that: The hydro-wind-solar complementary power generation system described in step (2) has a robust opportunity-constrained optimization model including an operation model of the hydropower entity, an operation model of the wind power entity, and an operation model of the photovoltaic entity.

4. The non-iterative scheduling method for a water-wind-solar complementary system taking into account both privacy protection and uncertainty according to claim 3 is characterized in that: The operation model of the hydropower entity is a mixed integer linear programming model, that is, a deterministic model, and its specific form is as follows: s.t.P h，k，t =η k H k Q k,t P h，k，t =η k H k Q k，t V k，T =V k,end S k,t ≥0 Where T is the total number of time periods; N h is the number of hydropower stations; R k is the set of upstream reservoirs of hydropower station k; ψ g P h，k，t is the set of load periods g; the output of hydropower station k in period t; Q k，t is the power generation flow of hydropower station k in period t; S k，t V is the water discharge of hydropower station k in period t; k，t is the reservoir capacity of hydropower station k at time t; E h，t The shared electricity of the hydropower entity; t is the electricity price in period t; k is the hydropower conversion coefficient of hydropower station k; t lk is the water stagnation time; and are the upper and lower limits of the output k of the hydropower station respectively; and are the upper and lower limits of the k storage capacity of the hydropower station respectively; and are the upper and lower limits of the power generation flow of hydropower station k; W k，t is the natural water inflow of hydropower station k in period t; τ g is the proportional coefficient of the g load period.

5. The non-iterative scheduling method for a water-wind-solar complementary system taking into account both privacy protection and uncertainty according to claim 3 is characterized in that: The operation model of the wind power entity is a distributed robust chance-constrained optimization model, that is, an uncertainty model, and its specific form is as follows: P w,m,t ≥0 Among them, N w is the number of wind power stations; P w，m，t is the output of wind power station m in period t; E w，m，t is the shared electricity of wind power station m in period t; is the actual output of wind power station m during period t; w，m，t is the prediction error of wind power station m in period t; f pen penalizing electricity prices for renewable energy; is the predicted output of wind power station m during period t.

6. The non-iterative scheduling method for a water-wind-solar complementary system taking into account both privacy protection and uncertainty according to claim 3 is characterized in that: The operation model of the photovoltaic entity is a distributed robust chance-constrained optimization model, that is, an uncertainty model, and its specific form is as follows: P v,n,t ≥0 Among them, N v is the number of photovoltaic power stations; P v，n，t is the output of photovoltaic power station n in period t; E v，n，t is the shared electricity of photovoltaic power station n in period t; is the actual output of photovoltaic power station n in period t; v，n，t is the prediction error of photovoltaic power station n in period t; f pen penalizing electricity prices for renewable energy; is the predicted output of photovoltaic power station n in period t.

7. The non-iterative scheduling method for a water-wind-solar complementary system taking into account both privacy protection and uncertainty according to claim 1, characterized in that: The implementation process of step (3) is as follows: (31) Dual transformation of the objective function of the wind power and photovoltaic main operation models: The objective functions of the wind power and photovoltaic main operation models are unified as follows: In the formula, the variables x, y and z in the first stage are located on the set χ; G(x,ξ) is a linear expression about x and ξ; for the second term, the strong duality theory is used to transform it into the following form: According to the convex transformation theory, it is further transformed into: At this point, the objective function is finally transformed into: (32) Robust chance constrained transformation: Due to opportunity constraints, wind power entities and photovoltaic entities are unified as follows: By introducing the wind and solar output error ξ t The lower bound Rewrite the chance constraint as: In order to obtain a tight lower bound for the wind and solar output error, the following model is constructed and solved: Based on CVaR and strong duality theory, this type of model is reconstructed and finally converted into: (33) Constructing a joint dispatching model for water, wind and solar power: The main operation models of hydropower, wind power and photovoltaic power are transformed into compact matrix forms. The main operation model of hydropower is a mixed integer nonlinear programming problem, and the main operation models of wind power and photovoltaic power are linear programming problems. The compact models of each subject are expressed as: Where i is the index of the power generation entity, including hydropower stations, wind power stations and photovoltaic power stations; x i is a linear decision variable; i is an integer decision variable; z is an integer decision variable, that is, {E h，t ,E w，n，t ,E v，n，t }; For hydropower stations, x i = {P h，k，t ,Q k，t ,S k，t ,V k，t }、y i = {I h，k，t }; For a wind power station, x i = {P w，m，t ,λ m ,s m，k }, For photovoltaic power stations; For photovoltaic power stations, x i = {P v，n，t ,λ n ,s n，k }, In order to achieve the multi-energy complementarity of water, wind and solar, the joint scheduling model P0 was established: In the formula, Ez=0 is the constraint between the interaction variables, which means that the sum of the interaction power of each subject in each period is 0, specifically:

8. The non-iterative scheduling method for a water-wind-solar complementary system taking into account both privacy protection and uncertainty according to claim 1, characterized in that: The implementation process of step (4) is as follows: A decentralized strategy is adopted to share the system management rights among the hydropower, wind power and photovoltaic power generation entities, the encryption center and the cloud computing center; the hydropower, wind power and photovoltaic power generation entities are composed of multiple nodes, including hydropower, wind power and photovoltaic entities, and the encryption methods are elimination, column transformation, row transformation and affine transformation; the encryption center is a public asset on the user side, which can only mechanically execute pre-set instructions, and its operation and maintenance are supervised by all entities on the user side; the encryption methods are column transformation, row transformation and affine transformation; the cloud computing center solves various optimization models through the Internet and with computing resources paid by usage.

9. The non-iterative scheduling method for a water-wind-solar complementary system taking into account both privacy protection and uncertainty according to claim 7, characterized in that: The lower bound of the wind and solar output error is is 0.

10. The non-iterative scheduling method for a water-wind-solar complementary system taking into account both privacy protection and uncertainty according to claim 8, characterized in that: The solution of the optimization model is specifically as follows: Preprocessing of hydropower, wind and solar power generation data: Each subject randomly selects the variables to be eliminated x i , and characterize the safety domain through FM P i Perform row transformation, that is and The transformed security domain is right Perform column transformation, that is and The transformed security domain is right Perform an affine transformation, that is and The transformed security domain is Each entity will separately The coefficient matrix E is encrypted and transmitted to the encryption center; Encryption-centric encryption: Building the model and convert it into a compact form UP = {min A up x up +a up sB up x up +C up y up ≤ b up };in, Perform row transformation on UP, that is and The transformed model is UP-H. The column transformation of UP-H is performed, that is, and The transformed model is UP-L; UP-L is affine transformed, that is, and The transformed model Encrypt and transmit the model UP-S to the cloud computing center; Cloud computing center computing: Cloud computing solves the UP-S model and sets the optimal solution Encrypted transmission to the encryption center; Encryption Center Decryption: Decryption, that is in And the result z * and Encrypted and transmitted to the corresponding power generation entity i; Participating entities distribute the solution: Each entity solves the model {max ω i st ω i =A i x i +a i , B i x i +C i y i +D i z * ≤ b i }, that is, the solution of coordinated optimization of water, wind and solar power scheduling is completed.

Images