Energy transaction price-volume decision-making method

Abstract

The invention discloses an energy transaction volume-price decision-making method, and relates to the field of power system optimization scheduling, and the method comprises the steps: building a first model according to a prediction sample, and enabling the first model to be used for analyzing the uncertainty of an electric heating load and photovoltaic power generation; constructing a second model in combination with the first model and a value-at-risk theory, wherein the second model is used for reducing source load uncertainty; according to an output result of the first model, constructing a third model for balancing energy; and constructing a fourth model according to an output result of the first model, the second model and the third model, wherein the fourth model is used for obtaining a decision result. According to the method, on the basis of fully considering the uncertainty of the source load and the interests of the buyer and the seller, the established Nash bargaining model further realizes revenue maximization of the buyer and the seller under the condition of considering the interests of the buyer and the seller, the existing resources are efficiently utilized, and the additional cost is prevented from being increased.

Description

Technical Field

[0001] The present invention relates to the technical field of power system optimization dispatching, and in particular to an energy transaction quantity and price decision method. Background Art

[0002] With the development of information technology, traditional consumers can become prosumers by investing in local power generation. The park energy system can bring multiple prosumers together to share their energy resources. However, due to the volatility and intermittency of distributed power sources, the amount of electricity that buyers purchase from energy suppliers is uncertain.

[0003] In the current context of energy Internet and electricity market, there is a demand for mutual energy trading among park energy systems. For multi-energy systems in parks connected to the same distribution network area, a distributed power trading market with less interactive information, simple operation and decentralization is more attractive. At the same time, the formulation of energy prices is also related to distributed energy generation. In these aspects, Nash bargaining theory provides an effective method for energy price negotiations, which can ensure that the interests of buyers and sellers are fairly resolved. By applying the concepts of game theory, the negotiated price can balance the needs of both parties according to their bargaining power, thereby reaching a consensus and ultimately obtaining a sustainable power purchase agreement. Summary of the invention

[0004] In view of the problems existing in the above-mentioned technology, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that the existing energy trading method is difficult to maximize the interests of both buyers and sellers and is difficult to efficiently utilize existing resources.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a method for energy trading quantity and price decision-making, which includes: constructing a first model based on prediction samples to analyze the uncertainty of electric thermal load and photovoltaic power generation; constructing a second model in combination with the first model and risk value theory to reduce source-load uncertainty; constructing a third model based on the output results of the first model to balance energy; constructing a fourth model based on the output results of the first model, the second model and the third model to obtain a decision result.

[0007] As a preferred solution of the energy transaction quantity and price decision-making method described in the present invention, the prediction samples include historical data or simulation data on electric thermal load and photovoltaic power generation.

[0008] As a preferred solution of the energy transaction quantity and price decision-making method described in the present invention, the first model is a hybrid model of a convolutional neural network and a long short-term memory network.

[0009] As a preferred solution of the energy transaction quantity and price decision method of the present invention, the method of constructing the first model includes obtaining multiple feature graphs, accelerating the network convergence speed, modeling the input gate, the forget gate, and the output gate, obtaining the memory cell state, and obtaining the output information;

[0010] Among them, the method of obtaining multiple feature maps includes performing a convolution operation on the input feature map, and the expression is:

[0011]

[0012] In the formula, l represents the number of layers, represents the jth input quantity, represents the j-th output feature map, are the convolution kernel weights and bias values ​​respectively;

[0013] The method of accelerating the network convergence speed includes setting the BN layer, which is expressed as:

[0014]

[0015] In the formula, μ B , are the mean and variance respectively, ε is an infinitesimal quantity, x i ,y i They represent the input feature quantity and output feature quantity of the BN layer respectively, γ and β are the scale factor and motion factor respectively, and m is the number of samples in the current batch;

[0016] The model expressions of the input gate, forget gate and output gate are:

[0017]

[0018] In the formula, i t 、f t , o t They are input gate, forget gate, and output gate respectively; x t Represents preprocessed data; h t-1 Represents the hidden state of the previous moment; W xi , W hi 、b i They are the weight matrix of the input gate, the weight matrix of the hidden state, and the bias vector; W xf , W hf 、b f They are the weight matrix of the forget gate, the weight matrix of the hidden state, and the bias vector; W xo , W ho 、b oThey are the weight matrix of the output gate, the weight matrix of the hidden state, and the bias vector; σ represents the sigmoid function, which changes the output value to between 0 and 1;

[0019] The expression of the memory cell state is:

[0020] c t =f t c t-1 +i t tanh(W xc x t +W hc h t-1 +b c )

[0021] In the formula, c t-1 Indicates the cell state at the last moment, W xc , W hc 、b c Represents the weight matrix of the cell state, the weight matrix of the hidden state and the bias vector, and tanh represents the hyperbolic tangent function;

[0022] The expression of the output information is:

[0023] h t =o t *tanh(c t )

[0024] In the formula, h t Represents the output at the current moment.

[0025] 5. The energy transaction quantity and price decision method according to claim 4, characterized in that: the construction method of the second model includes, according to the predicted photovoltaic output and electric heat load information, using the beta distribution function to construct the uncertainty of the source and load in any period, and the expression is:

[0026]

[0027] Where Γ is the gamma function; γ t , β t is the probability shape parameter, X t Represents the photovoltaic power generation or electric heating load data predicted in time period t;

[0028] According to the predefined distribution function of the uncertainty of photovoltaic power generation and electric heating load, the second model is constructed, and its expression is:

[0029] VaR α =μ-σ·φ -1 (α)

[0030] In the formula, VaR αis the risk value calculated at the confidence level α, μ and σ are probability distribution functions, f(X t ), and φ -1 (α) is the quantile function under the confidence level α; the probability distribution function of the uncertainty of photovoltaic power generation or electric heating load within a day is as follows:

[0031] f(X1,…,X 24 )=f(X1)f(X2)…f(X 24 ).

[0032] As a preferred solution of the energy transaction quantity and price decision method of the present invention, the construction method of the third model includes constructing a battery energy storage model according to the balanced energy demand, and its expression is:

[0033] LT avg =LC DOD *DOD*E bat

[0034] In the formula, LT avg is the total throughput of the device over the entire battery life, DOD is the depth of discharge of the battery, LC DOD is the life cycle of the battery at a given discharge depth, E bat is the battery capacity;

[0035]

[0036] In the formula, C w is the life loss cost of the battery due to frequent charging, η dch is the battery discharge efficiency, C ini is the initial investment cost;

[0037]

[0038] In the formula, is the photovoltaic power generation of the next day predicted by the autoregressive model, ε t is white noise, φ i are the modeling parameters estimated using the Yule-Walker equation;

[0039] P t NET =P t PPA -P t PV

[0040] Where P t NET To consider the PV power generation surplus or deficit of the PPA at each moment, P t PPAis the power purchase agreement amount at time t, P t PV is the predicted power of photovoltaic power generation.

[0041] As a preferred solution of the energy transaction quantity and price decision method of the present invention, the construction method of the fourth model includes constructing a seller utility function, which is expressed as follows:

[0042] U s (e,P PV )=e·P PV -C(P PV )

[0043]

[0044] In the formula, e is the electricity price, P PV is the photovoltaic power purchase amount, C(P PV ) is the cost function;

[0045] Construct the buyer's utility function, which is expressed as:

[0046] U B (e,P PV )=V(P PV )-e·P PV

[0047] In the formula, V(P PV ) is the value function;

[0048] The fourth model is constructed according to the seller utility function and the buyer utility function, and its expression is:

[0049] e * = argmax(U S (e,P PV )-d s )(U B (e,P PV )-d b )

[0050] In the formula, d=(d S ,d B ) is the divergence point, d S d B are the clearing prices for sellers and buyers respectively;

[0051] Solving the fourth model, the expression of its first-order optimal condition is as follows:

[0052]

[0053] d S =(τ-r s )·PPV -C(P PV )

[0054] d b =V(P PV )-(τ-r b )·P PV

[0055] In the formula, r s 、r b are the risks faced by the seller and the buyer respectively, and τ is the average electricity price predicted during the agreement period.

[0056] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of the above method when executing the computer program.

[0057] A computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the above method when executed by a processor.

[0058] The beneficial effects of the present invention are as follows: the Nash bargaining model constructed by the present invention fully considers the uncertainty of source and load and the interests of both parties, further realizes the maximization of the benefits of both parties in consideration of their respective interests, efficiently utilizes existing resources, and avoids increasing additional costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0060] Figure 1 This is the overall flow chart of the energy trading quantity and price decision-making method.

[0061] Figure 2 This is a comparison chart of the relationship between photovoltaic output prediction and transaction power in scenario 1 in Example 3.

[0062] Figure 3 This is a battery energy storage charge state curve diagram for scenario 1 in Example 3.

[0063] Figure 4 This is a comparison chart of the relationship between photovoltaic output prediction and transaction power under scenario 2 in Example 3.

[0064] Figure 5 This is a battery energy storage charge state curve diagram under scenario 2 in Example 3. DETAILED DESCRIPTION

[0065] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0066] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0067] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an embodiment that is mutually exclusive with other embodiments, either individually or selectively.

[0068] Example 1

[0069] Reference Figure 1 , which is the first embodiment of the present invention, and provides an energy transaction quantity and price decision method, the energy transaction quantity and price decision method comprising:

[0070] S1: Based on the forecast samples, the first model is constructed to analyze the uncertainty of electric thermal load and photovoltaic power generation.

[0071] Specifically, the prediction samples are historical data or simulated data about electric heating load and photovoltaic power generation. The first model is a hybrid model of convolutional neural network (CNN) and long short-term memory network (LSTM). The construction process of the first model is as follows:

[0072] S1-1: First, perform convolution operation on the input feature map to obtain multiple input feature maps, which are expressed as:

[0073]

[0074] In the formula, l represents the number of layers, represents the jth input quantity, represents the j-th output feature map, are the convolution kernel weights and bias values ​​respectively.

[0075] S1-2: To speed up the network convergence, set the BN layer, and the calculation is as follows:

[0076]

[0077] In the formula, μ B , are the mean and variance respectively, ε is an infinitesimal quantity, xi ,y i They represent the input feature and output feature of the BN layer respectively, γ and β are the scale factor and motion factor respectively, and m is the number of samples in the current batch.

[0078] S1-3: Model the input gate, forget gate, and output gate. The expression is:

[0079]

[0080] In the formula, i t 、f t , o t They are input gate, forget gate, and output gate respectively; x t Represents preprocessed data; h t-1 Represents the hidden state of the previous moment; W xi , W hi 、b i They are the weight matrix of the input gate, the weight matrix of the hidden state, and the bias vector; W xf , W hf 、b f They are the weight matrix of the forget gate, the weight matrix of the hidden state, and the bias vector; W xo , W ho 、b o They are the weight matrix of the output gate, the weight matrix of the hidden state, and the bias vector; σ represents the sigmoid function, which changes the output value to between 0 and 1.

[0081] S1-4: The expression of memory cell state is:

[0082] c t =f t c t-1 +i t tanh(W xc x t +W hc h t-1 +b c )

[0083] In the formula, c t-1 Indicates the cell state at the last moment, W xc , W hc 、b c Represents the weight matrix of the cell state, the weight matrix of the hidden state, and the bias vector, and tanh represents the hyperbolic tangent function.

[0084] S1-5: Output information, the expression is:

[0085] h t =o t *tanh(ct )

[0086] In the formula, h t Represents the output at the current moment.

[0087] S:1-6: Evaluate prediction accuracy:

[0088]

[0089] In the formula, MAE is the mean absolute error, which represents the absolute difference between the predicted value and the measured value, MAPE represents the mean absolute percentage error, RMSE is the root mean square error, n is the number of samples, Fi is the predicted value, Obi is the observed value, and the smaller the value, the higher the accuracy of the prediction.

[0090] S2: The second model is constructed by combining the first model and risk value theory to reduce source load uncertainty.

[0091] S2-1: First, based on the predicted photovoltaic output and electric heat load information, the uncertainty of the source and load in any period is constructed using the β distribution function, and its expression is:

[0092]

[0093] Where Γ is the gamma function; γ t , β t is the probability shape parameter, X t Represents the photovoltaic power generation or electric heating load size data predicted during period t.

[0094] S2-2: Based on the predefined distribution function of the uncertainty of photovoltaic power generation and electric heating load, the second model is constructed, and its expression is:

[0095] VaR α =μ-σ·φ -1 (α)

[0096] In the formula, VaR α is the risk value calculated at the confidence level α, μ and σ are probability distribution functions, f(X t ), and φ -1 (α) is the quantile function under the confidence level α; the confidence level indicates the probability that the actual photovoltaic power output or electric thermal load exceeds the risk value. The economic losses caused by the uncertainty of photovoltaic power generation and electric thermal load can be effectively reduced through prediction and probability confidence. Therefore, the probability distribution function of the uncertainty of photovoltaic power generation or electric thermal load within a day can be obtained as follows:

[0097] f(X1,…,X 24 )=f(X1)f(X2)…f(X 24 ).

[0098] S3: Constructing a third model according to the output result of the first model to balance energy.

[0099] Specifically, the third model construction method includes constructing a battery energy storage model according to the balanced energy demand, and its expression is:

[0100] LT avg =LC DOD *DOD*E bat

[0101] In the formula, LT avg is the total throughput of the device over the entire battery life, DOD is the depth of discharge of the battery, LC DOD is the life cycle of the battery at a given discharge depth, E bat is the battery capacity;

[0102]

[0103] In the formula, C w is the life loss cost of the battery due to frequent charging, η dch is the battery discharge efficiency, taking 90%, C ini is the initial investment cost;

[0104]

[0105] In the formula, is the photovoltaic power generation of the next day predicted by the autoregressive model, ε t is white noise, φ i are the modeling parameters estimated using the Yule-Walker equation;

[0106] P t NET =P t PPA -P t PV

[0107] Where P t NET To consider the PV power generation surplus or deficit of the PPA at each moment, P t PPA is the power purchase agreement amount at time t, P t PV is the predicted power of photovoltaic power generation.

[0108] Battery energy storage state of charge related constraints:

[0109] The energy storage state of charge related constraints include the energy storage state of charge constraints at each moment, the maximum and minimum state of charge constraints, the first and last state of charge constraints of each day, and the charge and discharge constraints, which are given by the following formulas:

[0110] (1) Constraints on the state of charge of energy storage at each moment:

[0111]

[0112] Where, SoC t is the state of charge of the energy storage at time t, η c , η d are the energy storage charging and discharging efficiency, P t G2B , P t P2B are the power from the grid to the energy storage and the power from the energy storage to the grid, respectively, es is the battery capacity.

[0113] (2) Maximum and minimum state of charge constraints:

[0114]

[0115] In the formula, They are the minimum and maximum energy storage power respectively.

[0116] (3) Constraints on the charge state at the beginning and end of each day:

[0117] SoC1=SoC 24

[0118] In the formula, SoC1, SoC 24 They are the state of charge at the first moment and the state of charge at the last moment of energy storage respectively.

[0119] (4) Charge and discharge constraints:

[0120] 0≤P t G2B ⊥P t P2B ≥0

[0121] 0≤P t G2B ⊥P t P2B ≤0

[0122] 0≤P t P2B ⊥P t B2G ≤0

[0123] The first formula ensures that the battery energy storage can be charged from photovoltaic or power grid, while the second and third formulas ensure that the battery energy storage can be charged and discharged in any specific time period; at the same time, it also ensures that the energy storage can only be in two states at each moment, either charging or discharging, and cannot be charged or discharged at the same time, but can be neither charged nor discharged at the same time.

[0124] S4: constructing a fourth model according to the output result of the first model, the second model and the third model to obtain a decision result.

[0125] S4-1: Construct the seller’s (park energy supplier) utility function.

[0126] The utility function can be used to express the supplier's satisfaction and utility with the energy purchase agreement, which is related to various factors, such as the price of photovoltaic power generation, the sales volume of photovoltaic power generation and the stability of the demand side. Its basic form is as follows:

[0127] U s (e,P PV )=e·P PV -C(P PV )

[0128]

[0129] In the formula, e is the electricity price, P PV is the photovoltaic power purchase amount, C(P PV ) is the cost function. The goal of the park energy supplier is to provide the maximum price when negotiating the power purchase agreement with the demand side to maximize the utility. The first-order derivative conditions obtained by taking the first-order derivative of the above formula are as follows:

[0130]

[0131] To ensure that the marginal benefit of the supplier's utility is greater than 0, its price e should follow the following constraints:

[0132] e≥C′(P PV )

[0133] S4-2: Construct the buyer (demand side) utility function.

[0134] Similar to the supplier model, the utility function on the demand side represents the satisfaction with the PPA, which is related to the PV power generation, the price paid for PV power generation, the reliability of PV power generation and the size of the electric thermal load. Its basic form is as follows:

[0135] U B (e,P PV )=V(P PV )-e·P PV

[0136] In the formula, V(P PV) is the value function, which is related to the photovoltaic power generation and the reliability of photovoltaic power generation. The goal of the demand side is to obtain the power purchase agreement at the lowest possible power purchase price to maximize the utility. By taking the first-order derivative of the utility function, the first-order optimality condition is as follows:

[0137]

[0138] S4-3: Construct the fourth model based on the seller's utility function and the buyer's utility function, and its expression is:

[0139] e * = argmax(U S (e,P PV )-d s )(U B (e,P PV )-d b )

[0140] The divergence point can reflect the worst case scenario for both parties when the PPA negotiation fails, that is, there are market risks on both the energy supply side and the demand side, but the risks borne by both parties on the future market electricity price are different, as shown in the following formula:

[0141] d S =(τ-r s )·P PV -C(P PV )

[0142] d b =V(P PV )-(τ-r b )·P PV

[0143] In the formula, r s 、r b are the risks faced by the seller and the buyer respectively, and τ is the average electricity price predicted during the agreement period. The solution to Nash negotiation can be obtained by solving the first-order optimality condition, that is, setting the derivative of the product with respect to the price to 0 and solving it. Its first-order optimality condition is as follows:

[0144]

[0145] Following the above steps, we finally get a mixed integer linear programming problem. By solving it on the MATLAB platform, we finally get the optimal quantity and price decision plan.

[0146] Example 2

[0147] The second embodiment of the present invention is different from the previous embodiment in that:

[0148] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0149] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as based on a computer, including a processor or other device that can fetch instructions from the instruction execution system, device or apparatus and execute instructions), or in conjunction with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with these instruction execution systems, devices or apparatuses.

[0150] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0151] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0152] Example 3

[0153] Reference Figure 2 to Figure 5 , which is the third embodiment of the present invention, and this embodiment is based on embodiment 1. This embodiment provides two different weather and energy market scenarios, and uses these two different scenarios to test the feasibility of the method.

[0154] (1) Scenario 1 is a typical day of profit. Figure 2 In the figure, we can see the balance demand of the day. On a typical day with surplus, photovoltaic power generation is usually higher than demand. However, from the 6th to the 10th hour, photovoltaic power generation is lower than demand, so energy storage is needed to achieve energy balance through discharge. From the 10th to the 16th hour, photovoltaic power generation is higher than demand. At this time, energy storage needs to be charged to achieve power balance. The status of battery energy storage device in day-ahead economic dispatch is as follows: Figure 3 , the energy storage related parameters are shown in Table 1 below:

[0155] Table 1 Battery energy storage related

[0156]

[0157] (2) Scenario 2 is a typical deficit day, that is, power generation deficit in all periods. Obviously, energy must be purchased from the market to make up for the shortfall in photovoltaic power generation. Figure 4 In the example, the battery energy storage system is effectively charged from the grid during the 20th to 24th hour when the spot price is lower than the PPA price. Since this is a deficit example where the forecasted generation is less than the contracted generation, the impact of PV on the battery is zero. However, the BESS (battery energy storage system) plays a vital role by releasing its stored energy to meet the contracted power demand with the buyer with the PPA (power purchase agreement). This enables the agreed power to be delivered in the event of a deficit in PV generation, ensuring that contractual obligations are met. In this case, the optimal size of the BESS is 50 kWh, which means that a larger storage system is needed when the park's PV generators are producing insufficient power.

[0158] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for energy transaction quantity and price decision-making, characterized by: include, Based on the forecast samples, the first model is constructed to analyze the uncertainty of electric heating load and photovoltaic power generation; Combining the first model with the risk value theory to construct a second model for reducing source load uncertainty; Constructing a third model based on the output result of the first model for balancing energy; A fourth model is constructed according to the output result of the first model, the second model and the third model to obtain a decision result.

2. The energy transaction quantity and price decision method according to claim 1, characterized in that: The prediction samples include historical data or simulation data on electric thermal load and photovoltaic power generation.

3. The energy transaction quantity and price decision method according to claim 2, characterized in that: The first model is a hybrid model of a convolutional neural network and a long short-term memory network.

4. The energy transaction quantity and price decision method according to claim 3, characterized in that: The method for constructing the first model includes obtaining multiple feature maps, accelerating the network convergence speed, modeling the input gate, the forget gate, and the output gate, obtaining the memory cell state, and obtaining output information; Among them, the method of obtaining multiple feature maps includes performing a convolution operation on the input feature map, and the expression is: In the formula, l represents the number of layers, represents the jth input quantity, represents the j-th output feature map, are the convolution kernel weights and bias values ​​respectively; The method of accelerating the network convergence speed includes setting the BN layer, which is expressed as: In the formula, μ B , are the mean and variance respectively, ε is an infinitesimal quantity, x i ,y i They represent the input feature quantity and output feature quantity of the BN layer respectively, γ and β are the scale factor and motion factor respectively, and m is the number of samples in the current batch; The model expressions of the input gate, forget gate and output gate are: In the formula, i t 、f t , o t They are input gate, forget gate, and output gate respectively; x t Represents preprocessed data; h t-1 Represents the hidden state of the previous moment; W xi , W hi 、b i They are the weight matrix of the input gate, the weight matrix of the hidden state, and the bias vector; W xf , W hf 、b f They are the weight matrix of the forget gate, the weight matrix of the hidden state, and the bias vector; W xo , W ho 、b o They are the weight matrix of the output gate, the weight matrix of the hidden state, and the bias vector; σ represents the sigmoid function, which changes the output value to between 0 and 1; The expression of the memory cell state is: c t =f t c t-1 +i t tanh(W xc x t +W hc h t-1 +b c ) In the formula, c t-1 Indicates the cell state at the last moment, W xc , W hc 、b c Represents the weight matrix of the cell state, the weight matrix of the hidden state and the bias vector, and tanh represents the hyperbolic tangent function; The expression of the output information is: h t =o t *tanh(c t ) In the formula, h t Represents the output at the current moment.

5. The energy transaction quantity and price decision method according to claim 4, characterized in that: The construction method of the second model includes constructing the uncertainty of the source and load in any period of time using the beta distribution function according to the predicted photovoltaic output and electric heat load information, and the expression is: Where Γ is the gamma function; γ t , β t is the probability shape parameter, X t Represents the photovoltaic power generation or electric heating load data predicted in time period t; According to the predefined distribution function of the uncertainty of photovoltaic power generation and electric heating load, the second model is constructed, and its expression is: VaR α =μ-σ·φ -1 (a) In the formula, VaR α is the risk value calculated at the confidence level α, μ and σ are probability distribution functions, f(X t ), and φ -1 (α) is the quantile function under the confidence level α; the probability distribution function of the uncertainty of photovoltaic power generation or electric heating load within a day is as follows: f(X1,...,X 24 )=f(X1)f(X2)...f(X 24 )。 6. The energy transaction quantity and price decision method according to claim 5, characterized in that: The method for constructing the third model includes constructing a battery energy storage model according to the balanced energy demand, and the expression is: LT avg =LC DOD *DOD*E bat In the formula, LT avg is the total throughput of the device over the entire battery life, DOD is the depth of discharge of the battery, LC DOD is the life cycle of the battery at a given discharge depth, E bat is the battery capacity; In the formula, C w is the life loss cost of the battery due to frequent charging, η dch is the battery discharge efficiency, C ini is the initial investment cost; In the formula, is the photovoltaic power generation of the next day predicted by the autoregressive model, ε t is white noise, φ i are the modeling parameters estimated using the Yule-Walker equation; P t NET =P t PPA -P t PV Where P t NET To consider the PV power generation surplus or deficit of the PPA at each moment, P t PPA is the power purchase agreement amount at time t, P t PV is the predicted power of photovoltaic power generation.

7. The energy transaction quantity and price decision method according to claim 6, characterized in that: The method for constructing the fourth model includes constructing a seller utility function, which is expressed as follows: U s (e,P PV )=e·P PV -C(P PV ) In the formula, e is the electricity price, P PV is the photovoltaic power purchase amount, C(P PV ) is the cost function; Construct the buyer's utility function, which is expressed as: U B (e,P PV )=V(P PV )-e·P PV In the formula, V(P PV ) is the value function; The fourth model is constructed according to the seller utility function and the buyer utility function, and its expression is: e * =argmax(U S (e,P PV )-d s )(U B (e,P PV )-d b ) In the formula, d=(d S ,d B ) is the divergence point, d S ,d B are the clearing prices for sellers and buyers respectively; Solving the fourth model, the expression of its first-order optimal condition is as follows: d S (τ-r s )·P PV -C(P PV ) d b =V(P PV )-(τ-r b )·P PV In the formula, r s 、r b are the risks faced by the seller and the buyer respectively, and τ is the average electricity price predicted during the agreement period.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images