An adaptive online optimization method for multi-energy collaborative power supply

Abstract

The present invention proposes an adaptive online optimization method for multi-energy collaborative power supply. By establishing a multi-energy collaborative power supply optimization objective function and constraint conditions, it not only considers the operating cost, but also comprehensively considers the pollution cost and energy loss, optimizes resource utilization, reduces investment and operating costs, and maximizes the consumption of clean energy such as wind power and photovoltaic power. A wind power prediction model and a photovoltaic power prediction model are established. Wind power is predicted using a prediction method based on a BP‑GA model or a support vector machine, and photovoltaic power is predicted using a K‑Cascaded‑Attention Network, thereby improving the accuracy of wind power and photovoltaic power prediction. Dynamic planning and scheduling of multiple energy sources are performed to obtain a multi-energy collaborative power supply plan, thereby improving energy utilization.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-energy collaborative power supply technology, and in particular to a multi-energy collaborative power supply adaptive online optimization method. Background Art

[0002] The construction of smart cities is inseparable from the development of advanced smart grids. Existing technologies have applied cloud and neural network technologies to smart grid construction, resulting in the development of numerous software systems and intelligent technologies. These technologies improve the timeliness and accuracy of grid information processing, ensure the safety and economic efficiency of power system operation, and promote the overall advancement of my country's smart grid technology.

[0003] Smart grid can be said to be the main development trend of the global power industry. It takes the power system of power generation, transmission, distribution, energy storage and consumption as the object, applies digital information technology, communication technology and automatic control technology to realize two-way exchange of information in all links from power generation to power consumption, systematically optimizes the production, transmission and use of electricity, and realizes the intelligence of the power system.

[0004] The key technologies of smart grids encompass many areas and fields, such as remote monitoring, cloud-based monitoring of the comprehensive status of grid equipment, smart grid information collection, fault diagnosis and troubleshooting of smart distribution networks, dynamic cloud-based dispatch and control of smart grids, and smart grids with multi-energy collaborative power supply. Multi-energy collaborative power supply smart grid technology, through the application of cloud computing and neural network technology, can quickly, timely, and accurately calculate power consumption at different times in the grid system and control power consumption based on the calculated results. However, current multi-energy collaborative power supply systems lack real-time online optimization, and cannot quickly and accurately reconstruct collaborative power supply strategies. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem that multi-energy collaborative power supply in the above-mentioned prior art cannot be optimized online in real time, and to provide a multi-energy collaborative power supply adaptive online optimization method to achieve the purpose of quickly reconstructing the collaborative power supply strategy.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] The present invention provides a multi-energy collaborative power supply adaptive online optimization method, comprising the following steps:

[0008] Step S1: Establishing a multi-energy collaborative power supply optimization objective function and constraint conditions, wherein the objective function includes a cost function F1, a pollution cost function F2, and a voltage deviation function F3;

[0009] Step S2: Solve the optimal solution of the objective function based on the wolf pack optimization algorithm, perform dynamic planning and scheduling on the multi-energy collaborative power supply, and obtain a multi-energy collaborative power supply plan.

[0010] Furthermore, the multiple energy sources include: thermal power generation, wind power generation, and photovoltaic power generation.

[0011] Furthermore, the multi-energy collaborative power supply optimization objective function and constraint conditions in step S1 include:

[0012] Step S11, establish cost function F1:

[0013]

[0014] In the above formula, F f 、F w 、F pv They represent the power generation costs of thermal power generation, wind power generation, and photovoltaic power generation, respectively. T represents the cycle. I, J, and Q are the number of thermal power generation, wind power generation, and photovoltaic power generation units, respectively. P i，t Represents the active power of the ith thermal power generating unit in the tth cycle, W j，t represents the power loss of the j-th wind turbine in the t-th cycle, PV q，t represents the power loss of the qth photovoltaic generator set in the tth cycle, a i 、b i 、c i d j 、e q All are coefficients.

[0015] Among them, the P i，t 、W j，t PV q，t Satisfy the following constraints:

[0016]

[0017] In the above formula, P t represents the load demand in the tth period, W t PV t are the power losses of wind power generation and photovoltaic power generation in the tth cycle respectively.

[0018] P in thermal power generation i，t Satisfy the following constraints:

[0019] P i min ≤P i，t ≤P i max (3)

[0020] In the above formula, P i min and P i maxRepresents the minimum and maximum active power of the i-th thermal power generating unit.

[0021] W in wind power generation j，t Satisfy the following constraints:

[0022] W j，t =|W j，t ′-W j，t ″|≤δ (4)

[0023] In the above formula, W j，t ′、W j，t ″ represent the actual value of wind power generation and the predicted value of wind power generation, respectively, and δ represents the confidence level.

[0024] Photovoltaic power generation q，t Satisfy the following constraints:

[0025] PV q，t =|PV q，t '-PV q，t ”|≤δ (5)

[0026] In the above formula, PV q，t '、PV q，t ” represent the actual value of photovoltaic power generation and the predicted value of photovoltaic power generation, and δ represents the confidence level.

[0027] The climbing rate satisfies the following constraints:

[0028]

[0029] In the above formula, UR i and DR i They represent the upper and lower ramp rate limits of the i-th thermal power generating unit respectively.

[0030] Step S12, constructing the pollution cost function F2:

[0031] in,

[0032] In the above formula, F2 is the pollution cost function, α i , β i , γ i is the pollution cost coefficient of the i-th thermal power generating unit.

[0033] Step S13, constructing a voltage deviation function F3:

[0034]

[0035] In the above formula, P is the number of transmission lines and load buses, V p，t is the bus voltage of the transmission line and load bus p in the tth cycle.

[0036] Bus voltage V p，t Satisfy the following constraints:

[0037]

[0038] In the above formula, Indicates the minimum and maximum values ​​of the voltage amplitude.

[0039] In step S14, the cost function F1, the pollution cost function F2 and the voltage deviation function F3 are combined to establish the multi-energy collaborative power supply optimization objective function as follows:

[0040]

[0041] Furthermore, the wind power generation power prediction value in step S11 is predicted by a prediction method based on a BP-GA model or a support vector machine.

[0042] Furthermore, the photovoltaic power generation power prediction value in step S11 is predicted by the following steps:

[0043] Step S21: Collect sample data, including solar irradiance, backplane temperature, ambient temperature and humidity, and corresponding power, and determine the sample data set, which is expressed as {(x n ,y n )|n=1, 2, 3,...N}, x n represents the input feature, y n Represents x n Corresponding power, the sample data set includes a training set and a test set;

[0044] Step S22: Construct a photovoltaic power generation prediction model, the prediction model is a K-Cascaded-attention network, the K-Cascaded-attention network is a cascade neural network combined with an attention mechanism, the K-Cascaded-attention network includes K levels, the first level includes an input layer, a hidden layer, a BN layer, a Relu layer, a hidden layer, a BN layer, a Relu layer, a fully connected layer, and finally a softmax layer to obtain a prediction result f1(x n ); The structure of each level from level 2 to level K includes: input layer, hidden layer, channel attention module layer, BN layer, Relu layer, hidden layer, BN layer, Relu layer, fully connected layer, and finally the prediction result is f after the softmax layer. k (x n ), the input of the first stage is x n The input of the 2nd to Kth level is the output of the previous fully connected layer and x n, the output of the K-Cascaded-Attention Network is: Among them, τ k is the weight of the k-th level, K is the number of levels of the K-Cascaded-Attention Network;

[0045] Step S23: training the K-Cascaded-Attention Network based on the training set obtained in step S21;

[0046] Step S24: Utilize the K-Cascaded-Attention Network trained in step S23 to predict photovoltaic power generation and obtain a prediction result.

[0047] Furthermore, in step S2, solving the optimal solution of the objective function based on the wolf pack optimization algorithm is specifically solving the optimal solution of formula (10) based on the wolf pack optimization algorithm.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] (1) The present invention provides an adaptive online optimization method for multi-energy collaborative power supply. By establishing a multi-energy collaborative power supply optimization objective function and constraint conditions, it not only takes into account the operating cost, but also comprehensively considers the pollution cost and energy loss, optimizes the utilization of resources, reduces investment and operating costs, and maximizes the consumption of clean energy such as wind power and photovoltaics.

[0050] (2) Establish wind power prediction models and photovoltaic power prediction models, predict wind power through prediction methods based on BP-GA model or support vector machine, and predict photovoltaic power through K-Cascaded-attention network, which improves the accuracy of wind power and photovoltaic power prediction.

[0051] (3) Dynamically plan and dispatch multiple energy sources to obtain a multi-energy coordinated power supply plan and improve energy utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the specific implementation of the present invention or the technical solutions in the prior art, the drawings required for describing the specific implementation or the prior art will be briefly introduced below. Obviously, the drawings described below are only one implementation of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 The figure is a flow chart of an adaptive online optimization method for multi-energy collaborative power supply.

[0054] Figure 2 This is the K-Cascaded-Attention Network structure diagram. DETAILED DESCRIPTION

[0055] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the technical solutions in the specific implementation methods of the present invention are clearly and completely described below to further illustrate the present invention. Obviously, the specific implementation methods described are only part of the implementation methods of the present invention, rather than all styles.

[0056] The present invention is further described in detail below with reference to the accompanying drawings:

[0057] As attached Figure 1 As shown, the present invention provides a multi-energy collaborative power supply adaptive online optimization method, comprising the following steps:

[0058] Step S1: Establishing a multi-energy collaborative power supply optimization objective function and constraint conditions, wherein the objective function includes a cost function F1, a pollution cost function F2, and a voltage deviation function F3;

[0059] Step S2: Solve the optimal solution of the objective function based on the wolf pack optimization algorithm, perform dynamic planning and scheduling on the multi-energy collaborative power supply, and obtain a multi-energy collaborative power supply plan.

[0060] Furthermore, the multiple energy sources include: thermal power generation, wind power generation, and photovoltaic power generation.

[0061] Furthermore, the multi-energy collaborative power supply optimization objective function and constraint conditions in step S1 include:

[0062] Step S11, establish cost function F1:

[0063]

[0064] In the above formula, F f 、F w 、F pv They represent the power generation costs of thermal power generation, wind power generation, and photovoltaic power generation, respectively. T represents the cycle. I, J, and Q are the number of thermal power generation, wind power generation, and photovoltaic power generation units, respectively. P i，t Represents the active power of the ith thermal power generating unit in the tth cycle, W j，t represents the power loss of the j-th wind turbine in the t-th cycle, PV q，t represents the power loss of the qth photovoltaic generator set in the tth cycle, a i 、b i 、c i d j 、e q All are coefficients.

[0065] Among them, the P i，t 、Wj，t PV q，t Satisfy the following constraints:

[0066]

[0067] In the above formula, P t represents the load demand in the tth period, W t PV t are the power losses of wind power generation and photovoltaic power generation in the tth cycle respectively.

[0068] P in thermal power generation i，t Satisfy the following constraints:

[0069] P i min ≤P i，t ≤P i max (3)

[0070] In the above formula, P i min and P i max Represents the minimum and maximum active power of the i-th thermal power generating unit.

[0071] W in wind power generation j，t Satisfy the following constraints:

[0072] W j，t =|W j，t ′-W j，t ″|≤δ (4)

[0073] In the above formula, W j，t ′、W j，t ″ represent the actual value of wind power generation and the predicted value of wind power generation, respectively, and δ represents the confidence level.

[0074] Photovoltaic power generation q，t Satisfy the following constraints:

[0075] PV q，t =|PV q，t '-PV q，t ”|≤δ (5)

[0076] In the above formula, PV q，t '、PV q，t ” represent the actual value of photovoltaic power generation and the predicted value of photovoltaic power generation, and δ represents the confidence level.

[0077] The climbing rate satisfies the following constraints:

[0078]

[0079] In the above formula, UR i and DR i They represent the upper and lower ramp rate limits of the i-th thermal power generating unit respectively.

[0080] Step S12, constructing the pollution cost function F2:

[0081] in,

[0082] In the above formula, F2 is the pollution cost function, α i , β i , γ i is the pollution cost coefficient of the i-th thermal power generating unit.

[0083] Step S13, constructing a voltage deviation function F3:

[0084]

[0085] In the above formula, P is the number of transmission lines and load buses, V p，t is the bus voltage of the transmission line and load bus p in the tth cycle.

[0086] Bus voltage V p，t Satisfy the following constraints:

[0087]

[0088] In the above formula, Indicates the minimum and maximum values ​​of the voltage amplitude.

[0089] In step S14, the cost function F1, the pollution cost function F2 and the voltage deviation function F3 are combined to establish the multi-energy collaborative power supply optimization objective function as follows:

[0090]

[0091] Furthermore, the wind power generation power prediction value in step S11 is predicted by a prediction method based on a BP-GA model or a support vector machine.

[0092] Furthermore, the photovoltaic power generation power prediction value in step S11 is predicted by the following steps:

[0093] Step S21: Collect sample data, including solar irradiance, backplane temperature, ambient temperature and humidity, and corresponding power, and determine the sample data set, which is expressed as {(x n ,y n )|n=1, 2, 3,...N}, x n represents the input feature, y n Represents xn Corresponding power, the sample data set includes a training set and a test set;

[0094] Step S22: Construct a photovoltaic power generation prediction model. The model structure is shown in the attached Figure 2 The prediction model is a K-Cascaded-attention network, which is a cascade neural network combined with an attention mechanism. The K-Cascaded-attention network includes K levels. The first level includes an input layer, a hidden layer, a BN layer, a Relu layer, a hidden layer, a BN layer, a Relu layer, a fully connected layer, and finally a softmax layer to obtain the prediction result f1(x n ); The structure of each level from level 2 to level K includes: input layer, hidden layer, channel attention module layer, BN layer, Relu layer, hidden layer, BN layer, Relu layer, fully connected layer, and finally the prediction result is f after the softmax layer. k (x n ), the input of the first stage is x n The input of the 2nd to Kth level is the output of the previous fully connected layer and x n , the output of the K-Cascaded-Attention Network is: Among them, τ k is the weight of the k-th level, K is the number of levels of the K-Cascaded-Attention Network;

[0095] Step S23: training the K-Cascaded-Attention Network based on the training set obtained in step S21;

[0096] Step S24: Utilize the K-Cascaded-Attention Network trained in step S23 to predict photovoltaic power generation and obtain a prediction result.

[0097] Furthermore, in step S2, solving the optimal solution of the objective function based on the wolf pack optimization algorithm is specifically solving the optimal solution of formula (10) based on the wolf pack optimization algorithm.

[0098] Compared with the prior art, the present invention has the following beneficial effects: the present invention provides an adaptive online optimization method for multi-energy collaborative power supply. By establishing a multi-energy collaborative power supply optimization objective function and constraint conditions, it not only takes into account operating costs, but also comprehensively considers pollution costs and energy losses, optimizes resource utilization, reduces investment and operating costs, and maximizes the consumption of clean energy such as wind power and photovoltaics. A wind power prediction model and a photovoltaic power prediction model are established. Wind power is predicted using a prediction method based on a BP-GA model or a support vector machine. Photovoltaic power is predicted using a K-Cascaded-Attention Network, thereby improving the accuracy of wind power and photovoltaic power prediction. Dynamic planning and scheduling of multiple energy sources are performed to obtain a multi-energy collaborative power supply plan, thereby improving energy utilization.

[0099] The above describes the main technical features and basic principles of the present invention and the related advantages. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments and that the present invention can be implemented in other specific forms without departing from the concept or essential characteristics of the present invention. Therefore, from all perspectives, the above-mentioned specific embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the foregoing description, and it is intended that all changes that come within the meaning and range of equivalents of the claims be included within the present invention.

[0100] In addition, it should be understood that although this specification is described according to various implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each implementation method can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A multi-energy collaborative power supply adaptive online optimization method, characterized in that: The following steps are involved: Step S1: Establishing a multi-energy collaborative power supply optimization objective function and constraint conditions, wherein the objective function includes a cost function F1, a pollution cost function F2, and a voltage deviation function F3; Step S2: Solve the optimal solution of the objective function based on the wolf pack optimization algorithm, perform dynamic planning and scheduling on the multi-energy collaborative power supply, and obtain a multi-energy collaborative power supply plan; The multi-energy collaborative power supply optimization objective function and constraint conditions in step S1 include: Step S11, establish cost function F1: （1）； In the above formula, They represent the power generation costs of thermal power generation, wind power generation, and photovoltaic power generation, respectively; T represents the cycle; I, J, and Q represent the number of thermal power generation, wind power generation, and photovoltaic power generation units, respectively. represents the active power of the ith thermal power generating unit in the tth cycle, represents the power loss of the j-th wind turbine in the t-th cycle, represents the power loss of the qth photovoltaic generator set in the tth cycle, All are coefficients; Among them, the 、 、 Satisfy the following constraints: （2）； In the above formula, represents the load demand in period t, 、 are the power losses of wind power generation and photovoltaic power generation in the tth cycle respectively; Thermal power generation Satisfy the following constraints: （3）； In the above formula, represents the minimum and maximum active power of the i-th thermal power generating unit; Wind power generation Satisfy the following constraints: (4) ； In the above formula 、 They represent the actual value of wind power generation and the predicted value of wind power generation respectively. Indicates confidence; Photovoltaic power generation Satisfy the following constraints: （5）； In the above formula They represent the actual value of photovoltaic power generation and the predicted value of photovoltaic power generation respectively. Indicates confidence; The climbing rate satisfies the following constraints: （6）； In the above formula, and They represent the upper and lower ramp rate limits of the i-th thermal power generating unit respectively; Step S12, constructing the pollution cost function F2: in, (7); In the above formula, F2 is the pollution cost function, is the pollution cost coefficient of the i-th group of thermal power generating units; Step S13, constructing a voltage deviation function F3: （8）； In the above formula, P is the number of transmission lines and load buses, is the bus voltage of the transmission line and load bus p in the tth cycle; Bus voltage Satisfy the following constraints: （9）； In the above formula, 、 Indicates the minimum and maximum values ​​of the voltage amplitude; In step S14, the cost function F1, the pollution cost function F2 and the voltage deviation function F3 are combined to establish the multi-energy collaborative power supply optimization objective function as follows: (10)； The photovoltaic power generation power prediction value in step S11 is predicted by the following steps: Step S21: Collect sample data, including solar irradiance, backplane temperature, ambient temperature and humidity, and corresponding power, and determine the sample data set, which is expressed as , represents the input features, express Corresponding power, the sample data set includes a training set and a test set; Step S22: Construct a photovoltaic power generation prediction model, the prediction model is a K-Cascaded-attention network, the K-Cascaded-attention network is a cascade neural network combined with an attention mechanism, the K-Cascaded-attention network includes K levels, the first level includes an input layer, a hidden layer, a BN layer, a Relu layer, a hidden layer, a BN layer, a Relu layer, a fully connected layer, and finally a softmax layer to obtain a prediction result. The structure of each level from level 2 to level K includes: input layer, hidden layer, channel attention module layer, BN layer, Relu layer, hidden layer, BN layer, Relu layer, fully connected layer, and finally the prediction result is obtained through the softmax layer. , the input of the first stage is The input of the 2nd to Kth level is the output of the previous level fully connected layer and , the output of the K-Cascaded-Attention Network is: ,in, is the weight of the k-th level, K is the number of levels of the K-Cascaded-Attention Network; Step S23: training the K-Cascaded-Attention Network based on the training set obtained in step S21; Step S24: Utilize the K-Cascaded-Attention Network trained in step S23 to predict photovoltaic power generation and obtain a prediction result.

2. The method for adaptive online optimization of multi-energy collaborative power supply according to claim 1, characterized in that: The multiple energy sources include: thermal power generation, wind power generation, and photovoltaic power generation.

3. The method for adaptive online optimization of multi-energy collaborative power supply according to claim 1, characterized in that: The wind power generation power prediction value in step S11 is predicted by a prediction method based on a BP-GA model or a support vector machine.

Images