A data-driven distributed photovoltaic voltage operation control method and system

Abstract

The present invention discloses a distributed photovoltaic voltage operation control method and system based on data-driven. The control method includes the following steps: obtaining the time series fluctuation characteristic data of the voltage, current and power of the photovoltaic system, and completing the missing values ​​of the data through a generative adversarial network to obtain the completed data; extracting the voltage fluctuation amplitude, current change rate, and power peak frequency from the completed data, and constructing a voltage operation state mapping model based on the extracted voltage fluctuation amplitude, current change rate, and power peak frequency; based on the voltage operation state mapping model, constructing a voltage control strategy using an improved gray wolf algorithm. The present application solution does not rely on precise physical models, has strong adaptability, can quickly respond to photovoltaic output fluctuations, significantly improves the power supply quality and reliability of the distribution network, reduces operation and maintenance costs, and has good scalability. It can be widely used in distributed photovoltaic systems of different scales and types.

Description

Technical Field

[0001] The present invention belongs to the technical field of model-free adaptive control under voltage operation of distributed photovoltaic distribution networks, and in particular relates to a data-driven distributed photovoltaic voltage operation control method and system. Background Art

[0002] With the large-scale integration of distributed photovoltaics, traditional voltage control methods based on physical models are struggling to cope with its volatility and uncertainty. Traditional voltage control methods rely on physical models, but the acquisition of physical parameters is often inaccurate in scenarios with a large number of distributed photovoltaics. The volatility of photovoltaic output requires control equipment with rapid response capabilities, but existing equipment such as OLTCs and capacitor banks has slow adjustment speeds and is unable to cope with real-time voltage fluctuations. Furthermore, traditional voltage control methods have limitations in handling voltage over-limit issues under large-scale distributed photovoltaic integration, and cannot effectively ensure the stable operation of the distribution network. Summary of the Invention

[0003] In view of the technical problems pointed out in the above background technology, the purpose of the present invention is to provide a data-driven distributed photovoltaic voltage operation control method and system.

[0004] To achieve the purpose of the present invention, the technical solution provided by the present invention is as follows:

[0005] First aspect

[0006] This application provides a data-driven distributed photovoltaic voltage operation control method, comprising the following steps:

[0007] Step 1: Obtain the time-series fluctuation characteristic data of the voltage, current, and power of the photovoltaic system, and use the generative adversarial network to fill in the missing values ​​of the data to obtain the completed data;

[0008] Step 2: Extract the voltage fluctuation amplitude, current change rate, and power peak frequency from the completed data, and build a voltage operation status mapping model based on the extracted voltage fluctuation amplitude, current change rate, and power peak frequency;

[0009] Step 3: Based on the voltage operation state mapping model, a voltage control strategy is constructed using an improved grey wolf algorithm.

[0010] Second aspect

[0011] The present application provides a data-driven distributed photovoltaic voltage operation control system, comprising the following units: a data acquisition unit, a model building unit, and a strategy building unit;

[0012] The data acquisition unit is used to obtain the time series fluctuation characteristic data of the voltage, current and power of the photovoltaic system, and to fill in the missing values ​​of the data through a generative adversarial network to obtain the filled data;

[0013] The model building unit is used to extract the voltage fluctuation amplitude, current change rate, and power peak occurrence frequency from the completed data, and build a voltage operation state mapping model based on the extracted voltage fluctuation amplitude, current change rate, and power peak occurrence frequency;

[0014] The strategy construction unit is used to construct a voltage control strategy based on the voltage operation state mapping model using an improved grey wolf algorithm.

[0015] Compared with the prior art, the present invention has the following beneficial effects:

[0016] The present application provides a data-driven distributed photovoltaic voltage operation control method and system with significant beneficial effects. It innovatively combines generative adversarial networks for data completion to ensure the integrity and accuracy of input data, laying a solid foundation for subsequent analysis; by extracting key feature quantities and constructing a mapping model, it accurately reflects the voltage operation status and provides a reliable basis for the control strategy. The improved gray wolf algorithm is used to optimize the voltage control strategy to achieve efficient regulation of the voltage of the distributed photovoltaic system, effectively solve the voltage over-limit problem, and ensure the stable operation of the distribution network. This method does not rely on precise physical models, has strong adaptability, can quickly respond to photovoltaic output fluctuations, significantly improves the power supply quality and reliability of the distribution network, reduces operation and maintenance costs, and has good scalability. It can be widely used in distributed photovoltaic systems of different sizes and types, promoting the efficient use of new energy and the development of smart grids. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic flow chart of a data-driven distributed photovoltaic voltage operation control method according to an embodiment of the present invention;

[0018] Figure 2 Flowchart of the improved grey wolf algorithm provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following is a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts are within the scope of protection of the present invention.

[0020] Example 1

[0021] like Figure 1and Figure 2 As shown, this embodiment provides a data-driven distributed photovoltaic voltage operation control method, including the following steps:

[0022] Step 1: Obtain the time-series fluctuation characteristic data of the voltage, current, and power of the photovoltaic system, and use the generative adversarial network to fill in the missing values ​​of the data to obtain the completed data;

[0023] Among them, in step 1, the missing values ​​of the data are filled by generating an adversarial network, specifically: the missing values ​​of the data are filled by the generator G and the discriminator D of the generative adversarial network, wherein,

[0024] The goal of the generator G is to make the discriminator D unable to distinguish between generated data and real data. The objective function of the generator G is expressed as:

[0025]

[0026] Among them, G is the generator, which maps the input noise to the model of generated data; z is the noise distribution The input noise vector sampled in ; is the distribution of input noise, usually Gaussian distribution or uniform distribution; D is the discriminator, which determines whether the input data is real data or generated data model; The data generated by the generator; Generate data for the discriminator pair The closer the value is to 1, the more likely it is real data, and the closer it is to 0, the more likely it is generated data; Represents the expectation of the input noise z on the distribution; Indicates that the generator hopes to maximize this value, so that the discriminator thinks that the probability of the generated data being not real data is as small as possible;

[0027] The goal of the discriminator D is to correctly distinguish between real data and generated data. The objective function of the discriminator D is expressed as:

[0028]

[0029] in, is real data sampled from the real data distribution; is the distribution of real data; Represents real data In distribution The expectation on logD(x) indicates that the discriminator hopes to maximize this value, so that the discriminator thinks that the real data The greater the probability of being real data, the better; log(1−D(G(z))) indicates that the discriminator hopes to maximize this value, that is, to make the discriminator think that the probability of generated data G(z) is not real data is as large as possible;

[0030] In the generative adversarial network, the adversarial process between the generator G and the discriminator D is jointly represented by the following objective function:

[0031]

[0032] Among them, V(D,G) is the joint objective function of the generator and the discriminator. The generator tries to minimize V(D,G), while the discriminator tries to maximize V(D,G).

[0033] Step 2: Extract the voltage fluctuation amplitude, current change rate, and power peak frequency from the completed data, and build a voltage operation status mapping model based on the extracted voltage fluctuation amplitude, current change rate, and power peak frequency;

[0034] Specifically, in step 2, the voltage fluctuation amplitude is extracted by the following method: :

[0035]

[0036] in, is the mean voltage; is the number of data points; is the voltage at time t.

[0037] Specifically, in step 2, the current change rate is extracted by the following method: :

[0038]

[0039] in, is the current at time t.

[0040] Specifically, in step 2, the power peak frequency is extracted by:

[0041]

[0042] in, is the number of power peaks; is the power data recording period, is the frequency of power peak occurrence.

[0043] Specifically, the voltage operating state mapping model:

[0044]

[0045] in, , , is the weight coefficient; is the residual term, is the voltage operating state model, is the mathematical expectation of the current change rate.

[0046] Step 3: Based on the voltage operation state mapping model, a voltage control strategy is constructed using an improved grey wolf algorithm.

[0047] Wherein, the step 3 specifically includes the following:

[0048] Step 3.1: Algorithm initialization, , , As the initial parameter of the individual position of the gray wolf, its value is dynamically adjusted through the optimization process to adapt to the real-time working conditions. The gray wolf position update formula is:

[0049]

[0050] in, For prey location, is the dynamic weight vector, is the distance factor;

[0051] Step 3.2: As input, construct the objective function of voltage control optimization to minimize the comprehensive deviation between the reference voltage and the actual voltage:

[0052]

[0053] in, is the reference voltage, and A is the weight coefficient, which is used to balance the comprehensive impact of the real-time voltage deviation and the state predicted by the mapping model. is the voltage at time t, where t ranges from 1 second to T seconds.

[0054] Step 3.3: Through the above iterative optimization process, the position of the gray wolf individuals is continuously updated, thereby gradually optimizing the objective function.

[0055] Example 2

[0056] Corresponding to the above method, this embodiment provides a data-driven distributed photovoltaic voltage operation control system, comprising the following units: a data acquisition unit, a model building unit, and a strategy building unit;

[0057] The data acquisition unit is used to obtain the time series fluctuation characteristic data of the voltage, current and power of the photovoltaic system, and to fill in the missing values ​​of the data through a generative adversarial network to obtain the filled data;

[0058] The model building unit is used to extract the voltage fluctuation amplitude, current change rate, and power peak occurrence frequency from the completed data, and build a voltage operation state mapping model based on the extracted voltage fluctuation amplitude, current change rate, and power peak occurrence frequency;

[0059] The strategy construction unit is used to construct a voltage control strategy based on the voltage operation state mapping model using an improved grey wolf algorithm.

[0060] The method of filling missing values ​​of data by generating an adversarial network is as follows: filling missing values ​​of data by generating an adversarial network generator G and a discriminator D, wherein:

[0061] The goal of the generator G is to make the discriminator D unable to distinguish between generated data and real data. The objective function of the generator G is expressed as:

[0062]

[0063] Among them, G is the generator, which maps the input noise to the model of generated data; z is the noise distribution The input noise vector sampled in ; is the distribution of input noise, usually Gaussian distribution or uniform distribution; D is the discriminator, which determines whether the input data is real data or generated data model; The data generated by the generator; Generate data for the discriminator pair The closer the value is to 1, the more likely it is real data, and the closer it is to 0, the more likely it is generated data; Represents the expectation of the input noise z on the distribution; Indicates that the generator hopes to maximize this value, so that the discriminator thinks that the probability of the generated data being not real data is as small as possible;

[0064] The goal of the discriminator D is to correctly distinguish between real data and generated data. The objective function of the discriminator D is expressed as:

[0065]

[0066] in, is real data sampled from the real data distribution; is the distribution of real data; Represents real data In distribution The expectation on logD(x) indicates that the discriminator hopes to maximize this value, so that the discriminator thinks that the real data The greater the probability of being real data, the better; log(1−D(G(z))) indicates that the discriminator hopes to maximize this value, that is, to make the discriminator think that the probability of generated data G(z) is not real data is as large as possible;

[0067] In the generative adversarial network, the adversarial process between the generator G and the discriminator D is jointly represented by the following objective function:

[0068]

[0069] Among them, V(D,G) is the joint objective function of the generator and the discriminator. The generator tries to minimize V(D,G), while the discriminator tries to maximize V(D,G).

[0070] Among them, the voltage fluctuation amplitude is extracted by the following method :

[0071]

[0072] in, is the mean voltage; is the number of data points; is the voltage at time t.

[0073] Finally, it should be noted that the above embodiments are merely examples and illustrations of the present invention and are not intended to limit the present invention to the described embodiments. Furthermore, those skilled in the art will appreciate that the present invention is not limited to the above embodiments and that various variations and modifications may be made based on the teachings of the present invention, all of which fall within the scope of the present invention.

Claims

1. A data-driven distributed photovoltaic voltage operation control method, characterized in that: The steps include: Step 1: Obtain the time-series fluctuation characteristic data of the voltage, current, and power of the photovoltaic system, and use the generative adversarial network to fill in the missing values ​​of the data to obtain the completed data; Step 2: Extract the voltage fluctuation amplitude, current change rate, and power peak frequency from the completed data, and build a voltage operation status mapping model based on the extracted voltage fluctuation amplitude, current change rate, and power peak frequency; Step 3: Based on the voltage operation state mapping model, a voltage control strategy is constructed using an improved grey wolf algorithm; The voltage operation status mapping model: V state =as V +βE[ΔI]+γf P +∈ Among them, α, β, γ are weight coefficients; ∈ is the residual term, V state is the voltage operating state, E[ΔI] is the mathematical expectation of the current change rate; the voltage fluctuation amplitude σ V , f P is the frequency of power peak occurrence; The step 3 specifically includes the following: Step 3.1: Initialize the algorithm. α, β, and γ are used as the initial parameters of the individual gray wolf positions. Their values ​​are dynamically adjusted through the optimization process to adapt to the real-time working conditions. The gray wolf position update formula is: in, For prey location, is the dynamic weight vector, is the distance factor; Step 3.2: V state As input, construct the objective function of voltage control optimization to minimize the comprehensive deviation between the reference voltage and the actual voltage: Among them, V ref is the reference voltage, A is the weight coefficient, which is used to balance the comprehensive impact of the real-time voltage deviation and the state predicted by the mapping model, V t is the voltage at time t, where t ranges from 1 second to T seconds; Step 3.3: Through the above iterative optimization process, the position of the gray wolf individuals is continuously updated, thereby gradually optimizing the objective function.

2. A data-driven distributed photovoltaic voltage operation control method according to claim 1, characterized in that: In step 1, the missing values ​​of the data are filled by generating an adversarial network, specifically: the missing values ​​of the data are filled by the generator G and the discriminator D of the generative adversarial network, wherein, The goal of the generator G is to make the discriminator D unable to distinguish between generated data and real data. The objective function of the generator G is expressed as: Among them, G is a generator that maps input noise to a model of generated data; z is a model derived from the noise distribution P Z (z) is the input noise vector sampled in; P Z (z) is the distribution of input noise, usually Gaussian distribution or uniform distribution; D is the discriminator, which determines whether the input data is real data or generated data; G(z) is the data generated by the generator; D(G(z)) is the judgment result of the discriminator on the generated data G(z). The closer the value is to 1, the more likely it is real data, and the closer it is to 0, the more likely it is generated data. Represents the expectation of the input noise z on the distribution; log(1-D(G(z))) indicates that the generator hopes to maximize this value, so that the discriminator thinks that the probability of the generated data being not real data is as small as possible; The goal of the discriminator D is to correctly distinguish between real data and generated data. The objective function of the discriminator D is expressed as: Where x is the real data sampled from the real data distribution; P data (x) is the distribution of real data; Indicates that the real data x is distributed in P data (x); logD(x) means that the discriminator hopes to maximize this value, so that the discriminator thinks that the probability of the real data x is the real data is as large as possible; log(1-D(G(z))) means that the discriminator hopes to maximize this value, that is, the discriminator thinks that the probability of the generated data G(z) is not the real data is as large as possible; In the generative adversarial network, the adversarial process between the generator G and the discriminator D is jointly represented by the following objective function: Among them, V(D,G) is the joint objective function of the generator and the discriminator. The generator tries to minimize V(D,G), while the discriminator tries to maximize V(D,G).

3. The data-driven distributed photovoltaic voltage operation control method according to claim 1, characterized in that: In step 2, the voltage fluctuation amplitude σ is extracted as follows V : Among them, μ V is the voltage mean; N is the number of data points; V t is the voltage at time t.

4. A data-driven distributed photovoltaic voltage operation control method according to claim 3, characterized in that: In step 2, the current change rate ΔI is extracted as follows t : ΔI t =I t+1 -I t Among them, I t is the current at time t.

5. The data-driven distributed photovoltaic voltage operation control method according to claim 4, characterized in that: In step 2, the power peak frequency is extracted as follows: Where M is the number of power peaks; T is the power data recording period, f P is the frequency of power peak occurrence.

6. A data-driven distributed photovoltaic voltage operation control system, characterized in that: It includes the following units: data collection unit, model building unit and strategy building unit; The data acquisition unit is used to obtain the time series fluctuation characteristic data of the voltage, current and power of the photovoltaic system, and to fill in the missing values ​​of the data through a generative adversarial network to obtain the filled data; The model building unit is used to extract the voltage fluctuation amplitude, current change rate, and power peak occurrence frequency from the completed data, and build a voltage operation state mapping model based on the extracted voltage fluctuation amplitude, current change rate, and power peak occurrence frequency; The strategy construction unit is configured to construct a voltage control strategy based on the voltage operation state mapping model using an improved grey wolf algorithm; The voltage operation status mapping model: V state =as V +βE[ΔI]+γf P +∈ Among them, α, β, γ are weight coefficients; ∈ is the residual term, V state is the voltage operating state, E[ΔI] is the mathematical expectation of the current change rate; the voltage fluctuation amplitude σ V , f P is the frequency of power peak occurrence; The strategy building unit is used to perform the following steps: Step 3.1: Initialize the algorithm. α, β, and γ are used as the initial parameters of the individual gray wolf positions. Their values ​​are dynamically adjusted through the optimization process to adapt to the real-time working conditions. The gray wolf position update formula is: in, For prey location, is the dynamic weight vector, is the distance factor; Step 3.2: V state As input, construct the objective function of voltage control optimization to minimize the comprehensive deviation between the reference voltage and the actual voltage: Among them, V ref is the reference voltage, A is the weight coefficient, which is used to balance the comprehensive impact of the real-time voltage deviation and the state predicted by the mapping model, V t is the voltage at time t, where t ranges from 1 second to T seconds; Step 3.3: Through the above iterative optimization process, the position of the gray wolf individuals is continuously updated, thereby gradually optimizing the objective function.

7. A data-driven distributed photovoltaic voltage operation control system according to claim 6, characterized in that: The method of filling missing values ​​of data by generating an adversarial network is as follows: filling missing values ​​of data by generating an adversarial network generator G and a discriminator D, wherein: The goal of the generator G is to make the discriminator D unable to distinguish between generated data and real data. The objective function of the generator G is expressed as: Among them, G is a generator that maps input noise to a model of generated data; z is a model derived from the noise distribution P Z (z) is the input noise vector sampled in; P Z (z) is the distribution of input noise, usually Gaussian distribution or uniform distribution; D is the discriminator, which determines whether the input data is real data or generated data; G(z) is the data generated by the generator; D(G(z)) is the judgment result of the discriminator on the generated data G(z). The closer the value is to 1, the more likely it is real data, and the closer it is to 0, the more likely it is generated data. Represents the expectation of the input noise z on the distribution; log(1-D(G(z))) indicates that the generator hopes to maximize this value, so that the discriminator thinks that the probability of the generated data being not real data is as small as possible; The goal of the discriminator D is to correctly distinguish between real data and generated data. The objective function of the discriminator D is expressed as: Where x is the real data sampled from the real data distribution; P data (x) is the distribution of real data; Indicates that the real data x is distributed in P data (x); logD(x) means that the discriminator hopes to maximize this value, so that the discriminator thinks that the probability of the real data x is the real data is as large as possible; log(1-D(G(z))) means that the discriminator hopes to maximize this value, that is, the discriminator thinks that the probability of the generated data G(z) is not the real data is as large as possible; In the generative adversarial network, the adversarial process between the generator G and the discriminator D is jointly represented by the following objective function: Among them, V(D,G) is the joint objective function of the generator and the discriminator. The generator tries to minimize V(D,G), while the discriminator tries to maximize V(D,G).

8. The data-driven distributed photovoltaic voltage operation control system according to claim 6, characterized in that: The voltage fluctuation amplitude σ is extracted by V : Among them, μ V is the voltage mean; N is the number of data points; V t is the voltage at time t.

Images