Multi-port flexible interconnection power regulation and control method and system based on power prediction

Abstract

The invention provides a multi-port flexible interconnection power regulation and control method and system based on power prediction. The method comprises the steps that the voltage of each port of a multi-port flexible interconnection device is acquired; processing the obtained voltage of the port by using a phase locking link to obtain a positive sequence fundamental wave effective value of each voltage; calculating a difference value of voltages of different ports based on the positive sequence fundamental wave effective values of the voltages, inputting the difference value into a locally deployed neural network, and calculating a power difference value between the two ports based on a neural network model; the power difference value is input into a PI regulator to obtain a d-axis current reference value Id of the port, and the port of the converter controls the output current based on the d-axis reference value Id and further controls the output power.

Description

Technical Field

[0001] The present invention belongs to the field of power prediction related to power electronics, and in particular relates to a multi-port flexible interconnection power control method and system based on power prediction. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the rapid development of power electronics technology, renewable energy has received more and more attention. At present, distributed generation (DG), represented by wind power and solar power generation, is developing rapidly and connected to the distribution network in large quantities, which can effectively save energy and reduce emissions. However, the output of distributed generation fluctuates greatly and rapidly, resulting in rapid and drastic fluctuations in the voltage of the distribution network, which seriously threatens the safety of electricity use.

[0004] Various new power electronic devices represented by flexible interconnection devices (also known as intelligent soft switches, soft open points, SOP) have achieved rapid development, and their applications in the optimization operation and dispatching of active distribution networks have become more and more extensive, and have achieved good results. As a new type of power electronic equipment, SOP can not only replace the interconnection switches in traditional distribution networks, improve the "hard connection" of traditional distribution networks, but also realize power mutual assistance between different feeders. The energy storage system can not only "cut the peak and fill the valley" to achieve effective distribution of electric energy in different time periods, but also convert unstable electric energy into stable electric energy output to ensure the safe and stable operation of the power grid. Charging piles can realize local consumption of electric energy, which is expected to solve the problem of electric energy consumption after large-scale access of distributed photovoltaics.

[0005] However, in the existing technology, the multi-port flexible interconnection device needs to rely on the power data of the power supply area to formulate an operating strategy. Additional detection devices, energy management devices and communication systems are indispensable, which significantly increases the complexity of the system. How to make it operate independently without relying on external instructions is an urgent problem that needs to be solved.

[0006] Existing patents propose flexible interconnection port power control methods based on local sampling, etc. Such methods need to be based on the distribution network model, but the distribution network model is difficult to obtain in practice. In practical applications, multi-port flexible interconnection devices need to be connected to the existing distribution network. In order to facilitate installation, the nearest wiring method is often used, such as Figure 1 As shown, the connection positions of multiple ports and multiple distribution networks are not exactly the same, and there are multiple forms such as the beginning of the line, the middle of the line, and the end of the line. At the same time, there are differences in the parameters of multiple distribution networks, mainly including differences in line load power, location differences, differences in photovoltaic power generation capacity, location differences, and differences in equivalent parameters of distribution lines.

[0007] In addition, the neural networks mentioned in existing literature or patents are mainly used for the day-ahead scheduling of flexible interconnected devices, which has a long time scale and is based on the global information of the entire power grid. However, this patent aims to solve the problem of real-time power regulation of flexible interconnected devices, and cannot rely on the global information of the power grid, but can only rely on local information such as voltage sampling.

[0008] The above reasons will lead to deviations in the power control method based on local sampling, making it difficult to meet the needs of precise control. Summary of the invention

[0009] In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a multi-port flexible interconnection power control method based on power prediction, which can realize real-time power distribution among multiple substations.

[0010] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0011] In a first aspect, a multi-port flexible interconnection power control method based on power prediction is disclosed, comprising:

[0012] Obtaining the voltage of each port of the multi-port flexible interconnection device;

[0013] The voltage of the acquired port is processed by using a phase-locked link to obtain the positive sequence fundamental effective value of each voltage;

[0014] The difference of the voltages at different ports is calculated based on the positive sequence fundamental effective value of the above voltages and input into a locally deployed neural network, and the power difference between the two ports is calculated based on the neural network model;

[0015] The power difference is input into the PI regulator to obtain the d-axis current reference value I of the port d , the converter port is based on the d-axis reference value I d Controls the output current and further controls the output power.

[0016] As a further technical solution, when the flexible interconnect has multiple ports, one port is fixed to adopt the DC voltage control mode, and the voltages of the remaining ports are respectively combined with the port voltages controlled by the DC voltage to form a neural network parameter input pair, which is not affected by the number of ports.

[0017] As a further technical solution, the training process of the locally deployed neural network is:

[0018] Obtaining the secondary side voltages of different transformers, and taking the difference between the secondary side voltages of two different transformers to obtain the voltage difference;

[0019] The voltage difference is normalized to a floating point number between 0 and 1;

[0020] The normalized voltage difference and the current time corresponding to the voltage difference are used as inputs of the neural network;

[0021] The input voltage difference is mapped to the current time to obtain a mapping value of time and voltage, and the power value at the current moment is inferred based on the mapping value.

[0022] As a further technical solution, the secondary voltage of the transformer of the distribution network is expressed as:

[0023]

[0024] Among them, V m Represents the transformer secondary voltage of distribution network m, I m Represents the secondary current of the transformer in the distribution network m, V g It represents the voltage value of 10kV distribution network normalized to 380V distribution network, Z ij Represents the impedance parameters between different nodes, C mg Indicates the influence coefficient of the upper-level power grid.

[0025] As a further technical solution, the current time corresponding to the voltage difference represents that the current moment is the tth second of the day.

[0026] As a further technical solution, the neural network includes an Embedding layer, a linear layer, and a fully connected network;

[0027] The Embedding layer maps the integer value t to a floating point vector, which is the mapping value of time, and is used to increase the complexity of time information;

[0028] The linear layer also maps the voltage value v to a floating point vector, the mapping value of the voltage value;

[0029] The mapping value of time and the mapping value of voltage are concatenated into a vector as the input of the fully connected neural network;

[0030] The fully connected network has three layers, including two hidden layers and an activation layer, which outputs the real-time power difference.

[0031] As a further technical solution, the parameters of the Embedding layer are a matrix E 86400×16 , get the mapping t of t map The process is as follows:

[0032] t map =[E t,0 ,E t,1 ,E t,2 ,…,E t,15 ](t=0,1,…,86399)

[0033] Where E t,j Represents the matrix E 86400×16 The element at row t, column j.

[0034] As a further technical solution, a linear layer is used to map the voltage value v to a 1x16 floating point vector;

[0035]

[0036] Matrix E 86400×16 as well as They are all learnable parameters and are continuously updated during network training.

[0037] In the second aspect, a multi-port flexible interconnection power control system based on power prediction is disclosed, including:

[0038] The voltage acquisition module is configured to: acquire the voltage of each port of the multi-port flexible interconnection device;

[0039] The positive sequence fundamental wave effective value acquisition module is configured to: obtain the positive sequence fundamental wave effective value of each voltage by processing the voltage of the acquired port using a phase-locked link;

[0040] The power difference prediction module is configured to: calculate the difference of voltages at different ports based on the positive sequence fundamental effective value of each voltage and input the difference into a locally deployed neural network, and calculate the power difference between the two ports based on the neural network model;

[0041] The power control module is configured as follows: the power difference is input into the PI regulator to obtain the d-axis current reference value I of the port d , the converter port is based on the d-axis reference value I d Controls the output current and further controls the output power.

[0042] One or more of the above technical solutions have the following beneficial effects:

[0043] The multi-port flexible interconnection device in the present invention can estimate the power difference based on the local voltage sampling information and the built-in neural network model, without the need to install a power measurement unit on each transformer. At the same time, it also avoids the data loss problem caused by the long-distance transmission of power data. The above-mentioned local sampling voltage is the AC port voltage collected by the flexible interconnection device.

[0044] The neural network model in the present invention can better obtain the parameter differences of the distribution network, which can not only avoid the cumbersome process of modeling, but also improve the applicability in different distribution networks, and is more conducive to the reasonable distribution of the power of multi-port flexible interconnection.

[0045] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0047] Figure 1 It is a multi-port flexible interconnection device connected to the power distribution network in the present invention;

[0048] Figure 2 It is the overall architecture of the multi-port flexible interconnected autonomous control system based on the Shenjiang network in the present invention;

[0049] Figure 3 is the neural network structure adopted in the present invention;

[0050] Figure 4 This is a block diagram of the flexible interconnection output power adaptive control in the present invention. DETAILED DESCRIPTION

[0051] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0052] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0053] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0054] The multi-port flexible interconnection device mainly includes a power converter and a controller for regulating voltage and current, and the specific control method is common knowledge.

[0055] Embodiment 1

[0056] This embodiment discloses a multi-port flexible interconnection power control method based on power prediction, including:

[0057] Obtaining the voltage of each port of the multi-port flexible interconnection device;

[0058] The voltage of the acquired port is processed by using a phase-locked link to obtain the positive sequence fundamental effective value of each voltage;

[0059] The difference of the voltages at different ports is calculated based on the positive sequence fundamental effective value of the above voltages and input into a locally deployed neural network, and the power difference between the two ports is calculated based on the neural network model;

[0060] The power difference is input into the PI regulator to obtain the d-axis current reference value I of the port d , the converter port is based on the d-axis reference value I d Controls the output current and further controls the output power.

[0061] Voltage based Power Difference Prediction (V-PDP) is a multi-port power prediction based on real-time voltage information, including the following processes: Figure 4 As shown:

[0062] Step 1: The multi-port flexible interconnection device measures the voltage of each port, and obtains the positive sequence fundamental effective value of each voltage through a phase-locked link, and calculates the difference Δv based on the effective value. ij The power difference between the two ports is calculated based on the neural network model. ij .

[0063] The flexible interconnection device has been connected to the power grid, so the local sampled voltage is the AC port voltage collected by the flexible interconnection device. The above difference is the difference between two ports, port i and port j.

[0064] The advantage of this step is that the voltage of the upper power grid fluctuates, and its fluctuation value will directly affect the judgment of the power value. By making a difference, the influence of the upper power grid fluctuation on the power difference judgment can be offset, thereby reducing the prediction error of the neural network.

[0065] Step 2: The power difference △P ij The d-axis current reference value I of this port is obtained by inputting it into the PI regulator d , the converter port is based on the d-axis reference value I d Controls the output current and further controls the output power.

[0066] Step 3: Power Difference based Power Regulation, PD-PR, multi-port power regulation based on predicted power difference: When the flexible interconnect device has multiple ports, one port is fixed to adopt DC voltage control mode, and the voltages of the remaining ports are respectively combined with the port voltages controlled by DC voltage to form a neural network parameter input pair. The parameter input pair Δv ij Not affected by the number of ports.

[0067] For example, if the flexible interconnection device has three ports, it can connect to three distribution networks at the same time, and each port has the same status and independent control.

[0068] The neural network is trained based on the meter data and the multi-port flexible interconnect voltage data (Power-Voltagebased Neural Network, PV-NN) to obtain the trained neural network.

[0069] Power data at the outlet of the low-voltage transformer. These data are used as power data for offline training. Power data is the data of the neural network, so the power data used is used to calculate the loss function of the neural network.

[0070] The training process includes the following steps:

[0071] Step (2-1) Data preprocessing based on multi-port flexible interconnect voltage difference:

[0072] In practice, the voltage of a 10kv distribution network is affected by the upper power grid and all lower power grids. According to the superposition theorem, it can be expressed by the following formula:

[0073]

[0074] Among them, V m Represents the transformer secondary voltage of distribution network m, I m Represents the secondary current of the transformer in the distribution network m, V g It represents the voltage value of 10kV distribution network normalized to 380V distribution network, Z ij Represents the impedance parameters between different nodes, C mg Indicates the influence coefficient of the upper power grid.

[0075] N represents the total number of nodes in the low-voltage distribution network. ij represents any number between 1 and N, and m represents the node number, from 1 to N.

[0076] Considering that the flexible interconnection devices are often installed at close distances, the influence of the upper power grid is close. The influence of the upper power grid can be eliminated by making a difference:

[0077] Δv ij =v i -v j (2)

[0078] Δv ij is the voltage difference, vi and vj are the voltage values ​​of ports i and j.

[0079] Considering that the input variable of the neural network is Δv ij ,Further, the power is differentiated as the output of the neural network.

[0080] Step (2-2) Normalization of voltage difference

[0081] The input of the network is the current time t and the voltage difference Δv at the current time ij , where time t is an integer between 0 and 86399, representing the current time as the tth second of the day (00:00:00-23:59:59). Voltage Δv ij It is normalized to a 32-bit floating point number between 0 and 1. The normalization method is the maximum and minimum value normalization method, as follows:

[0082]

[0083] Where Δv ij，mean is the average value, Δv ij，max , Δv ij，min It is 1.2 times the historical maximum voltage value and the historical minimum voltage value.

[0084] Step (2-3) is based on the input mapping of voltage difference and time series:

[0085] See attached Figure 3 As shown in the figure, the neural network contains 3 layers of neural networks, with 32, 64, and 32 neurons respectively. In order to make the neural network better perceive time information, an Embedding layer is added to the network to map the integer value t (t = 0, 1, ..., 86399) to a 1×16 floating point vector, increasing the 1-dimensional data to 16 dimensions, thereby increasing the complexity of time information. The Embedding parameter is an 86400×16 matrix E 86400×16 , get the mapping t of t map The process is as follows:

[0086] t map =[E t,0 ,E t,1 ,E t,2 ,…,E t,15 ](t=0,1,…,86399) (4)

[0087] Where E t,j Represents the matrix E 86400×16 The element at row t, column j.

[0088] In step (2-4), in order to prevent time information from having an absolute influence on network calculations, a linear layer is used to map the voltage difference v into a 1x16 floating-point vector.

[0089]

[0090] The above matrix E 86400×16 as well as They are all learnable parameters and are continuously updated during network training.

[0091] Steps (2-5) are based on the construction of a fully connected neural network structure based on voltage difference and time series:

[0092] Based on the mapping value of time and voltage, the power value at the current moment is inferred. map and voltage value mapping v map Concatenated into a 1×32 vector as the input of the fully connected neural network

[0093]

[0094] The fully connected network has three layers, of which the sizes of the two hidden layers are 64 and 32 respectively. The network uses the ReLU function as the activation layer, which is simple to calculate and has good gradient transfer characteristics.

[0095] Steps (2-6) Offline learning and field deployment:

[0096] The forward calculation process of the neural network is as follows:

[0097]

[0098] in It is a trainable parameter that is continuously updated during the network training process. After the network training is completed, the above-mentioned learnable parameters are:

[0099]

[0100] By substituting it into the Linux system deployed on site, the real-time power difference can be realized according to the voltage difference.

[0101] Embodiment 2

[0102] The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the program.

[0103] Embodiment 3

[0104] The purpose of this embodiment is to provide a computer-readable storage medium.

[0105] A computer-readable storage medium stores a computer program, which executes the steps of the above method when executed by a processor.

[0106] Embodiment 4

[0107] The overall structure of the system is as follows Figure 2 As shown, the purpose of this embodiment is to provide a multi-port flexible interconnection power control system based on power prediction, including:

[0108] The voltage acquisition module is configured to: acquire the voltage of each port of the multi-port flexible interconnection device;

[0109] The positive sequence fundamental wave effective value acquisition module is configured to: obtain the positive sequence fundamental wave effective value of each voltage by processing the voltage of the acquired port using a phase-locked link;

[0110] The power difference prediction module is configured to: calculate the difference of voltages at different ports based on the positive sequence fundamental effective value of each voltage and input the difference into a locally deployed neural network, and calculate the power difference between the two ports based on the neural network model;

[0111] The power control module is configured as follows: the power difference is input into the PI regulator to obtain the d-axis current reference value I of the port d , the converter port is based on the d-axis reference value I d Controls the output current and further controls the output power.

[0112] A neural network model is constructed based on low time resolution power sampling data and local voltage sampling data of flexible interconnection devices. Then, the neural network is deployed in the multi-port flexible interconnection device controller to construct a substation power difference prediction system and method based on the output results of the neural network. Finally, a system and method for regulating the multi-port power output based on the predicted substation power difference is constructed to achieve real-time power distribution among multiple substations. Specifically, by dynamically solving the power output of multiple ports, dynamic mutual assistance of power among multiple substations can be achieved, and the power-surplus substations are transmitted to the power-deficient substations, ultimately achieving power distribution.

[0113] Embodiment 5

[0114] The purpose of this embodiment is to provide a computer program product containing instructions, which, when running on a computer, enables the computer to execute the methods and functions involved in any of the above embodiments.

[0115] The steps involved in the apparatus of the above embodiment correspond to the method embodiment 1, and the specific implementation method can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0116] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0117] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A multi-port flexible interconnection power control method based on power prediction, characterized in that: include: Obtaining the voltage of each port of the multi-port flexible interconnection device; The voltage of the acquired port is processed by using a phase-locked link to obtain the positive sequence fundamental effective value of each voltage; The difference of the voltages at different ports is calculated based on the positive sequence fundamental effective value of the above voltages and input into a locally deployed neural network, and the power difference between the two ports is calculated based on the neural network model; The power difference is input into the PI regulator to obtain the d-axis current reference value I of the port d , the converter port is based on the d-axis reference value I d Controls the output current and further controls the output power.

2. The multi-port flexible interconnection power control method based on power prediction as claimed in claim 1 is characterized in that: When the flexible interconnect has multiple ports, one port is fixed to adopt the DC voltage control mode, and the voltages of the remaining ports and the port voltages controlled by the DC voltage form a neural network parameter input pair, which is not affected by the number of ports.

3. The multi-port flexible interconnection power control method based on power prediction as claimed in claim 1 is characterized in that: The training process of the locally deployed neural network is: Obtaining the secondary side voltages of different transformers, and taking the difference between the secondary side voltages of two different transformers to obtain the voltage difference; The voltage difference is normalized to a floating point number between 0 and 1; The normalized voltage difference and the current time corresponding to the voltage difference are used as inputs of the neural network; The input voltage difference is mapped to the current time to obtain a mapping value of time and voltage, and the power value at the current moment is inferred based on the mapping value.

4. The multi-port flexible interconnection power control method based on power prediction according to claim 1 is characterized in that: The secondary voltage of the transformer in the distribution network is expressed as: Among them, V m Represents the transformer secondary voltage of distribution network m, I m Represents the secondary current of the transformer in the distribution network m, V g It represents the voltage value of 10kV distribution network normalized to 380V distribution network, Z ij Represents the impedance parameters between different nodes, C mg Indicates the influence coefficient of the upper-level power grid.

5. The multi-port flexible interconnection power control method based on power prediction as claimed in claim 1 is characterized in that: The neural network includes an Embedding layer, a linear layer, and a fully connected network; The Embedding layer maps the integer value t to a floating point vector, which is the mapping value of time, and is used to increase the complexity of time information; The linear layer also maps the voltage value v to a floating point vector, the mapping value of the voltage value; The mapping value of time and the mapping value of voltage are concatenated into a vector as the input of the fully connected neural network; The fully connected network has three layers, including two hidden layers and an activation layer, which outputs the real-time power difference.

6. The multi-port flexible interconnection power control method based on power prediction according to claim 1 is characterized in that: The parameter of the Embedding layer is a matrix E 86400×16 , get the mapping t of t map The process is as follows: t map =[And t,0 ,AND t,1 ,AND t,2 ,…,AND t,15 ](t=0,1,…,86399) (4) Where E t,j Represents the matrix E 86400×16 The element at row t and column j; Use a linear layer to map the voltage value v to a floating-point vector; Matrix E 86400×16 as well as They are all learnable parameters and are continuously updated during network training.

7. A multi-port flexible interconnection power control system based on power prediction, characterized in that: include: The voltage acquisition module is configured to: acquire the voltage of each port of the multi-port flexible interconnection device; The positive sequence fundamental wave effective value acquisition module is configured to: obtain the positive sequence fundamental wave effective value of each voltage by processing the voltage of the acquired port using a phase-locked link; The power difference prediction module is configured to: calculate the difference of voltages at different ports based on the positive sequence fundamental effective value of each voltage and input the difference into a locally deployed neural network, and calculate the power difference between the two ports based on the neural network model; The power control module is configured as follows: the power difference is input into the PI regulator to obtain the d-axis current reference value I of the port d , the converter port is based on the d-axis reference value I d Controls the output current and further controls the output power.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 6 are performed.

Images