A distribution network control method and related devices considering distributed generation.

Abstract

This application discloses a distribution network control method and related device considering distributed generation, belonging to the field of new energy distribution network control technology. The method introduces a combination of quantum compression technology and reinforcement learning strategy optimization to achieve efficient control of distribution networks containing distributed generation. By performing quantum compression on the state vectors of distribution network nodes, the dimensionality of the state space can be significantly reduced while maintaining key system information, thereby reducing the amount of subsequent control calculations and improving computational efficiency. By using compressed state vectors to generate converter power commands, fine-grained power control of distributed generation can be achieved, ensuring the stability of voltage and power at distribution network nodes. Combined with reinforcement learning strategy optimization based on compressed state vectors and converter power commands, intelligent control of SOP switching actions is achieved, enabling the distribution network to respond quickly and maintain safe and stable operation under dynamic load and distributed generation fluctuations.

Description

Technical Field

[0001] This invention pertains to the field of new energy distribution network control technology, and particularly relates to a distribution network control method and related devices that take into account distributed power sources. Background Technology

[0002] With the large-scale integration of distributed generation (DG) power sources, the operating characteristics of distribution networks have changed significantly. Unlike traditional power systems dominated by centralized power sources, DG power sources are characterized by their large number, dispersed locations, and high output uncertainty, resulting in more complex dynamic characteristics of distribution networks in terms of power flow distribution, voltage stability, and power fluctuations. This complexity presents new challenges to the operation, control, and dispatch of distribution networks.

[0003] To address the uncertainties and volatility of distributed generation (DG), existing technologies widely incorporate flexible control devices such as Soft Open Points (SOPs). SOPs regulate active and reactive power across different nodes in the distribution network, achieving voltage support, power flow optimization, and peak shaving / valley filling, making them a crucial means of improving distribution network flexibility. Existing SOP control strategies primarily focus on optimization control based on traditional power flow calculations and SOP coordination based on Model Predictive Control (MPC). However, when dealing with large-scale DG integration, the high-dimensional dynamic characteristics of DGs cause a surge in computation time for traditional MPCs, and the response speed of mechanical switches and voltage regulating transformers is insufficient to suppress voltage flicker caused by new energy fluctuations. Furthermore, manually preset reconfiguration rules cannot meet the rapid response requirements of DGs.

[0004] Therefore, there is an urgent need for a new technical solution that can improve the efficiency of state modeling and control strategies while ensuring global optimization, so as to better adapt to the operation requirements of distribution networks with distributed power sources. Summary of the Invention

[0005] Based on this, the present invention aims to propose a distribution network regulation method and related device that considers distributed power sources. It achieves real-time perception of high-dimensional states through quantum state compression, constructs a dynamic balance network by utilizing the power rapid mutual assistance capability of multi-port SOPs, and improves the absorption capacity of distributed power sources by autonomously mining topology optimization strategies through reinforcement learning to realize the action regulation of SOPs.

[0006] In a first aspect, the present invention provides a distribution network control method considering distributed generation, comprising:

[0007] Establish the state-space equations for the distribution network nodes. The state vectors of the state-space equations include the output of distributed generation sources.

[0008] Quantum compression of the state vector yields a compressed state vector;

[0009] Generate converter power commands corresponding to distribution network nodes based on the compression state vector;

[0010] Using real-time measurement data of distribution network nodes as input state and SOP switch action command as action vector, a reinforcement learning state space is established. The reinforcement learning state space strategy is optimized based on converter power command, and the SOP switch action command is output. The SOP switch action command is used to drive each SOP action in the distribution network.

[0011] Furthermore, establishing the state-space equations for distribution network nodes includes:

[0012] Acquire real-time measurement data of distribution network nodes, including the output of distributed power sources;

[0013] State-space equations for distribution network nodes are established using real-time measurement data as state vectors.

[0014] Furthermore, the state-space equations of the distribution network nodes are expressed as follows:

[0015] ,

[0016] in, Represents the state vector. Indicates the output vector. express The derivative, Represents the state matrix, Represents the control input matrix. Indicates control input, Represents the perturbation input matrix. This indicates a disturbance input. Indicates the output matrix. Indicates the node voltage amplitude. Indicates the phase angle of the node voltage. and These represent the active and reactive power outputs of distributed generation sources, respectively; n represents the number of nodes in the distribution network; and m represents the number of distributed generation sources in the distribution network.

[0017] The matrices are represented as follows:

[0018] ,

[0019] in, and These represent the active power imbalance and reactive power imbalance at the nodes, respectively. and These represent the active power and reactive power of the load, respectively. Indicates active power loss. V represents reactive power loss. i and V j Let i and j represent the node voltages, respectively. and Let i represent nodal conductance and nodal admittance, where i=j represents self-conductance or self-susceptance, and i≠j represents mutual conductance or mutual susceptance. This represents the phase angle difference between two nodes. and Let represent the active power response time constant and the reactive power response time constant of the distributed generation, respectively. and These represent the SOP's control gain on the active and reactive power of distributed generation, respectively. and Let SOP represent the active and reactive power regulation gains of the u-th distributed source, respectively, where u∈(1,m). Represents the identity matrix. The distributed generation output observation matrix is ​​represented by:

[0020] .

[0021] DG stands for Distributed Generation, and PMU stands for Phasor Measurement Unit.

[0022] Furthermore, quantum compression of the state vector yields a compressed state vector, including:

[0023] The quantum state of the state vector is obtained by encoding the state vector with qubits;

[0024] Quantum variational optimization of a quantum state generates a squeezed state of the state vector;

[0025] Compressed state vectors are obtained by compressing the state vectors.

[0026] Furthermore, the compressed state vector obtained by compressed state decoding of the state vector includes:

[0027] The probability amplitude distribution is obtained by projecting the compressed state of the state vector.

[0028] A compressed state vector is obtained by inverse encoding transformation of the probability amplitude distribution using a pre-trained quantum encoder.

[0029] Furthermore, generating converter power commands corresponding to distribution network nodes based on the compressed state vector includes:

[0030] The output of the distributed power source is obtained by decoding and filtering the compressed state vector.

[0031] Obtain the line communication gain and voltage droop coefficient of the distribution network;

[0032] The active power command of the converter is calculated based on the output of the distributed power source and the line communication gain, and the reactive power command of the converter is calculated based on the voltage reactive droop coefficient.

[0033] Furthermore, based on the converter power command, the reinforcement learning state space policy is optimized, and the output SOP switching action command includes:

[0034] High-order features are obtained by extracting features from the reinforcement learning state space based on the compressed state vector and the converter power command.

[0035] The higher-order state is obtained by enhancing the input state based on higher-order features;

[0036] Using the SOP switch action instruction as the optimization target and higher-order states as policy inputs, the reinforcement learning state space is optimized to output the SOP switch action instruction.

[0037] In a second aspect, the present invention provides a distribution network control device considering distributed power sources, comprising:

[0038] The distribution network state space establishment module is used to establish the state space equations of the distribution network nodes. The state vectors of the state space equations include the output of distributed generation sources.

[0039] The state vector quantum compression module is used to perform quantum compression on the state vector to obtain a compressed state vector;

[0040] The power command generation module is used to generate converter power commands corresponding to distribution network nodes based on the compression state vector;

[0041] The reinforcement learning optimization module is used to establish a reinforcement learning state space with real-time measurement data of distribution network nodes as input state and SOP switch action commands as action vectors. It optimizes the reinforcement learning state space strategy based on converter power commands and outputs SOP switch action commands, which are used to drive each SOP action in the distribution network.

[0042] Thirdly, the present invention provides an electronic device including a memory storing computer-executable instructions and a processor, wherein when the computer-executable instructions are executed by the processor, the device performs the steps of the distribution network control method considering distributed power sources provided in the first aspect.

[0043] Fourthly, the present invention provides a readable storage medium storing a computer-executable program that, when executed, can implement the various steps of the distribution network control method considering distributed power sources provided in the first aspect.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] This invention proposes a distribution network control method and related device considering distributed generation. The method incorporates quantum compression technology combined with reinforcement learning strategy optimization to achieve efficient control of distribution networks containing distributed generation. By quantum compression of the distribution network node state vectors, the state space dimension can be significantly reduced while maintaining key system information, thereby reducing subsequent control computation and improving computational efficiency. By generating converter power commands using compressed state vectors, fine-grained power control of distributed generation can be achieved, ensuring the stability of voltage and power at distribution network nodes. Combined with reinforcement learning strategy optimization based on compressed state vectors and converter power commands, intelligent control of SOP switching actions is achieved, enabling the distribution network to respond quickly and maintain safe and stable operation under dynamic load and distributed generation fluctuations. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0047] Figure 1 A flowchart illustrating the implementation of a distribution network control method considering distributed power sources, as provided in an embodiment of the present invention.

[0048] Figure 2 A schematic diagram of a distribution network control device considering distributed power sources is provided for an embodiment of the present invention;

[0049] Figure 3 This is an electronic device architecture diagram provided for an embodiment of the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] See Figure 1 One embodiment of the present invention provides a distribution network control method considering distributed generation, comprising the following steps:

[0052] Step S110. Establish the state space equations of the distribution network nodes. The state vectors of the state space equations include the output of distributed generation sources.

[0053] This step acquires real-time measurement data of each node in the distribution network, including node voltage amplitude, phase angle, and active and reactive power output of distributed generation. Using these measurement data as state vectors, the state-space equations of the distribution network nodes are established.

[0054] The operating state of a power distribution network is determined by variables such as node voltage, phase angle, and distributed generation output. Establishing state-space equations can model the dynamic relationship between the system's internal state and external inputs, providing fundamental data for compression processing and control strategies. The state-space form clearly defines the system's coupling characteristics, response patterns, and controllability, facilitating subsequent quantum compression and reinforcement learning processing.

[0055] Specifically, the state vector can include node voltage magnitude, node voltage phase angle, and active or reactive power output of distributed generation; the input vector reflects external effects, such as load power or control commands; the output vector is the observables of the system, such as node voltage, current, and network losses; and the dynamic coupling relationship between state and input / output is described by the system matrix.

[0056] Furthermore, the state-space equations for the distribution network nodes are established as follows:

[0057]

[0058] in, Represents the state vector. Indicates the output vector. express The derivative, Represents the state matrix, Represents the control input matrix. Indicates control input, Represents the perturbation input matrix. This indicates a disturbance input. Indicates the output matrix. Indicates the node voltage amplitude. Indicates the phase angle of the node voltage. and These represent the active and reactive power outputs of distributed generation sources, respectively; n represents the number of nodes in the distribution network; and m represents the number of distributed generation sources in the distribution network.

[0059] The matrices are represented as follows:

[0060]

[0061] in, and These represent the active power imbalance and reactive power imbalance at the nodes, respectively. and These represent the active power and reactive power of the load, respectively. It represents active power loss, which is the energy consumed by the current as it flows through the line resistance due to heat generation, directly leading to energy waste and economic loss; Reactive power loss refers to the loss that occurs when current exchanges energy across reactance (inductance / capacitance), affecting voltage stability and equipment utilization; V i and V j Let i and j represent the node voltages, respectively. and They represent nodal conductance and nodal admittance, respectively. When i=j, they represent self-conductance or self-susceptance, and when i≠j, they represent mutual conductance or mutual susceptance. This represents the phase angle difference between two nodes; and These represent the active power response time constant and the reactive power response time constant of the distributed generation, respectively. and These represent the SOP's control gain on the active and reactive power of distributed generation, respectively. and Let SOP represent the active and reactive power regulation gains of the u-th distributed power source, respectively, where u∈(1,m); Represents the identity matrix. The distributed generation output observation matrix is ​​represented by:

[0062]

[0063] DG stands for Distributed Generation, and PMU stands for Phasor Measurement Unit.

[0064] Furthermore, if the data dimension is high, the computational load can be reduced by grouping or sampling nodes. A predictive model can also be introduced to assist in generating state vectors and enhance the model's dynamic response capability.

[0065] Step S120. Perform quantum compression on the state vector to obtain a compressed state vector.

[0066] This step utilizes quantum compression technology to effectively compress the dimensionality of the state space while preserving as much key system information as possible, thus reducing the computational complexity of subsequent control calculations. When the state vector dimension is high, direct control computation is computationally intensive and inefficient. Quantum compression uses quantum superposition and quantum variational optimization to map high-dimensional states to low-dimensional quantum states, achieving information compression while preserving key system characteristics. The compressed state vector reduces computational load and provides a high-quality control input foundation.

[0067] Further, step S120 includes the following steps:

[0068] Step S121. Encode the state vector with qubits to obtain the quantum state of the state vector.

[0069] Quantum bit encoding maps the original state space to a superposition of qubit states, realizing the quantization encoding of classical data, and can convert the state vector... Mapped to A quantum bit state This process can be represented as , ,in To represent the calculation of the ground state, for example , , This represents the normalization coefficient.

[0070] Step S122. Perform quantum variational optimization on the quantum state to generate a squeezed state of the state vector.

[0071] Quantum variational optimization is often used to find the optimal solution for a certain Hamiltonian H, such as finding the minimum energy state in the distribution network state compression problem. Its main steps include generating trial states, Hamiltonian optimization, and iterative loops.

[0072] Specifically, first use parameterized quantum circuits By acting on the initial quantum state (usually a state of zero), a candidate quantum state is obtained. This state is "tunable," i.e., a parameterized quantum circuit. It is an adjustable quantum gate sequence, dependent on the parameter vector. This is equivalent to generating a candidate solution in quantum space.

[0073] Parameterized quantum circuits The mathematical form of can be expressed as follows:

[0074]

[0075] in, This represents the Hamiltonian of the l-th layer. Represents a fixed nonparametric quantum gate. Let L represent the parameter vector of the l-th layer, and L represent the number of iteration layers.

[0076] Taking the state vector given in the embodiment of the present invention as an example, the trial state can be represented as:

[0077]

[0078] denoted by tensor product, and k represents the number of qubits.

[0079] tentative state It is a quantum state The parameterized approximation is achieved by optimizing the parameter vector. Minimize the expected energy (e.g., Hamiltonian H) using different parameter vectors. Different superposition states will be generated, covering different possible system configurations or compressed states. The trial state retains the key features of the original state, but the dimensionality is reduced.

[0080] Furthermore, the optimization process uses the Hamiltonian; the trial state is input into the Hamiltonian, the expected energy value is calculated, and the parameters are adjusted using a classical optimizer. To minimize the expected energy value, update the trial state.

[0081] Specifically, the Hamiltonian optimization process can be expressed as:

[0082]

[0083] in, This represents the Pauli-Z operator, which acts on a qubit to drive the quantum state to the ground state. Evolution, achieving state alignment; As a penalty, it ensures that critical grid information (such as voltage stability) is preserved in the compressed state. Indicates the weighting coefficient. This indicates the output reference value, reflecting the ideal operating condition (e.g., standard voltage).

[0084] The physical constraints of the power distribution system (such as voltage deviation) are transformed into the energy function of the quantum system through the Hamiltonian. The optimal solution of the parameter vector corresponds to the dimension-reduced state that satisfies the constraints.

[0085] The optimal parameters are obtained after optimization and convergence. The corresponding trial state is the optimal compressed state. Therefore, the compressed state of the final output state vector in this embodiment can be expressed as: .

[0086] Step S123. Compressed state decoding of the state vector yields the compressed state vector.

[0087] Specifically, this step involves projecting the compressed state of the state vector to obtain the probability amplitude distribution, and then using a pre-trained quantum encoder to perform an inverse encoding transformation of the probability amplitude distribution to obtain the compressed state vector.

[0088] The projection measurement process can be represented as follows:

[0089]

[0090] This represents a probability distribution.

[0091] The following inverse encoding transformation is performed using a pre-trained quantum encoder:

[0092]

[0093] Decoder represents a pre-trained quantum encoder that compresses the state vector. .

[0094] Step S130. Generate converter power commands corresponding to distribution network nodes based on the compression state vector.

[0095] Compressed state vectors provide crucial system state information. Generating converter power commands based on compressed state vectors enables precise adjustment of distributed power source output, achieving safe and stable control of node voltage and power, and is a core component of control decision-making.

[0096] Further, step S130 includes the following steps:

[0097] Step S131. Decode and filter the compressed state vector to obtain the output of the distributed power source.

[0098] Specifically, decoding filtering refers to recovering distributed generation output information from compressed state vectors. These compressed state vectors contain crucial information about the distribution network nodes, but their dimensionality is typically much lower than the original state vectors. Therefore, a decoding process is necessary to recover the specific output values ​​of the distributed generation. Decoding filtering not only extracts the real-time output of the power source but also eliminates noise interference, improves the accuracy of output commands, and thus achieves more precise power regulation.

[0099] The decoding process can be represented as follows:

[0100]

[0101] Furthermore, Kalman filtering can be performed during filtering to eliminate quantum measurement noise, and a sliding window FFT can be used for node voltage amplitude. Extract the fundamental frequency component to suppress harmonic interference.

[0102] Furthermore, the filter design needs to be adjusted according to the real-time dynamic characteristics of the power grid. For example, a stronger filter can be used in areas with large load fluctuations. A more preferred implementation can add a feedback mechanism to dynamically adjust the parameters of the filtering algorithm to adapt to real-time changes under different load conditions.

[0103] Step S132. Obtain the line communication gain and voltage droop coefficient of the distribution network.

[0104] Voltage and power in a distribution network are closely related, especially in the control of distributed generation sources and converters. Line communication gain determines the power coordination between nodes, while the voltage droop factor affects the reactive power regulation capability of the converter. Obtaining these two parameters helps achieve precise control of voltage and power flow in the distribution network.

[0105] Specifically, line communication gain describes the efficiency or enhancement of information transmission between nodes in a distribution network, and is used to determine the response relationship of power adjustment between nodes. It can be calculated and expressed as follows:

[0106]

[0107] Indicates the line communication gain. This indicates a communication delay.

[0108] The voltage droop factor describes the response capability of a converter in a distribution network to voltage changes. It is typically related to reactive power regulation capability; the larger the droop factor, the stronger the reactive power regulation capability of the converter in response to voltage changes. The setting calculation of the droop factor can be expressed as follows:

[0109]

[0110] This represents the voltage reactive power droop factor. Indicates the maximum reactive power capacity at SOP. Indicates the voltage dead zone threshold. This represents the attenuation coefficient.

[0111] Step S133. Calculate the active power command of the converter based on the output of the distributed power source and the line communication gain, and calculate the reactive power command of the converter based on the voltage reactive droop coefficient.

[0112] The active power command of the converter determines the power balance of the system, while the reactive power command directly affects the voltage level of the distribution network. Through proper calculation, the stability and optimized operation of the power grid under different operating conditions can be ensured.

[0113] The active power at each moment is obtained by solving the following equation:

[0114]

[0115] in, This represents the active power of line ij. Denotes the set of adjacent SOPs of distribution network node i. This represents the active power fluctuation of distributed generation. , This represents the active power output of the distributed power source at the previous moment.

[0116] The active power command for the SOP port converter is expressed as:

[0117]

[0118] This indicates the active power command for the f-th SOP port. This indicates the total active capacity of the SOP.

[0119] Furthermore, when the transmission power of port 1 accounts for 70% of the total, then the allocation... .

[0120] The reactive power command for the SOP port is generated as follows:

[0121]

[0122] Indicates voltage deviation. Indicates SOP voltage. This represents the reference voltage, sat represents the limiting function, and Qmax represents the preset maximum value of reactive power.

[0123] Step S140. Using real-time measurement data of distribution network nodes as input state and SOP switch action command as action vector, establish reinforcement learning state space, optimize reinforcement learning state space strategy based on converter power command, and output SOP switch action command. SOP switch action command is used to drive each SOP action in distribution network.

[0124] SOP (Start of Production) switching actions directly affect the topology and power flow distribution of the distribution network. Through reinforcement learning strategy optimization, the system can autonomously decide on switching actions in complex and dynamic environments, achieving intelligent control and ensuring the stable and efficient operation of the distribution network. This step establishes a reinforcement learning state space using real-time node measurement data as the state and SOP switching commands as the actions.

[0125] Specifically, the reinforcement learning state space is represented as:

[0126]

[0127]

[0128] Let represent the initial state at time t. This represents the node voltage magnitude at time t. This represents the amplitude of the line current at time t. This represents the output of the distributed power source at time t. Let represent the topological adjacency matrix at time t. Let represent the action vector at time t, with For example, this means the action command for the first SOP switch sw1.

[0129] Further, step S140 includes the following process:

[0130] Step S141. Extract features from the reinforcement learning state space based on the compressed state vector and the converter power command to obtain high-order features.

[0131] The effectiveness of reinforcement learning often depends on the quality and complexity of the input states. To improve the learning effect of the policy, it is necessary to extract features from the original state space, especially extracting high-order features that can reflect the dynamic behavior of the system from the compressed state vector and converter power commands. These high-order features can better capture the complex state changes of the distribution network, thereby enabling reinforcement learning algorithms to make more intelligent decisions.

[0132] Higher-order features typically refer to more abstract features that can effectively represent the behavior of a system, obtained from the original state space through combination, aggregation, and other methods, such as time series features and interdependency features.

[0133] Specifically, the initial features are obtained by decoding the compressed state vector and concatenating it with the converter active power command, as shown below:

[0134]

[0135] QuantumDecoder stands for Quantum Decoder. This indicates vector concatenation.

[0136] Feature extraction is performed using a pre-trained graph neural network, and the high-order features are output as follows:

[0137]

[0138] This represents the activation function. Degree matrix, Represents the augmented adjacency matrix. This represents the trainable weight matrix of the l-th layer. This represents the feature matrix of the l-th layer node. This represents the symmetric normalized Laplace operator.

[0139] Furthermore, output three layers of higher-order features. .

[0140] Step S142. Enhance the input state based on higher-order features to obtain a higher-order state.

[0141] State augmentation refers to introducing additional information or dimensions into the original state space or extracted higher-order features to provide more comprehensive input and help reinforcement learning models make better decisions. The purpose of state augmentation is to enhance the model's adaptability to complex and dynamic power distribution network environments by supplementing information that is not fully expressed in the environment. Through state augmentation, the expressive power of the state can be improved, enabling reinforcement learning algorithms to better learn the global information and details of the system. The resulting higher-order states contain more information and can more accurately represent the system's operating state.

[0142] Furthermore, the extracted high-order features are combined with other possible information (such as historical states, control history, etc.) to generate an enhanced state vector. The input state can be enhanced through convolution operations, weighted averaging, and the addition of additional global information, thereby improving the reinforcement learning model's ability to predict the future environment. The high-order state can be further processed through certain mathematical formulas or neural network layers before being passed to the policy network.

[0143] Furthermore, the higher-order states are represented as:

[0144]

[0145] in, Indicates a higher-order state. Represents higher-order features, This represents the vector flattening operation. Represents the historical topology action window, This indicates a vector concatenation operation.

[0146] Step S143. Optimize the reinforcement learning state space with the SOP switch action instruction as the optimization target and the higher-order state as the policy input, and output the SOP switch action instruction.

[0147] In reinforcement learning, policy optimization is the core task. By inputting higher-order states into the policy network, reinforcement learning can find a balance between exploration and exploitation, thereby generating the optimal SOP (Start of Operation) switching action command. This command directly affects the topology adjustment and power flow of the distribution network, ensuring that the system operates in an optimal state. Through continuous training and optimization, the system will learn how to output the best switching action command according to different state environments, realizing intelligent control of the distribution network.

[0148] Specifically, the policy network is the core model in reinforcement learning. It receives the environment state as input and outputs an action, which in this embodiment of the invention is the SOP switch action instruction. The objective function of the reinforcement learning algorithm aims to maximize the cumulative reward or minimize the system cost.

[0149] Higher-order states are used as inputs to reinforcement learning and passed to the policy network for training. The reinforcement learning model continuously optimizes the policy through algorithms (such as DDPG, A3C, etc.) so that the output SOP switching action command can maximize long-term rewards (such as stability, energy efficiency, etc.). Finally, the topology of the distribution network is adjusted according to the switching action command output by the policy to drive the SOP action of each node.

[0150] Furthermore, the output of the policy network is represented as:

[0151]

[0152] in, This indicates the SOP switch operation command. This represents the Actor network structure in the policy network. Indicates a higher-order state. The trainable parameters represent the structure of the Actor network. This represents random noise.

[0153] Taking the DDPG strategy as an example, the reward function for strategy optimization can be set as follows:

[0154]

[0155]

[0156] Used to adjust the total active power loss of the system. ; Indicates the SOP efficiency coefficient; Indicates an immediate reward. P' represents the voltage deviation weighting factor. loss P represents loss The updated value; This represents the mean of the absolute values ​​of the node voltages deviating from the per-unit value of 1.0; Indicates the network loss penalty coefficient; This indicates a penalty for violating safety constraints; This represents the over-limit indicator function, which can be set to 1 when voltage / current / topology constraints are violated, and 0 otherwise.

[0157] The network parameter update of the Actor network structure in the policy optimization model is expressed as:

[0158]

[0159] This represents the gradient of the Actor's policy with respect to the policy. This represents the policy objective function (expected cumulative reward). Represents the mathematical expectation operator. This represents the gradient of the Critic network structure with respect to the action. This represents the gradient of the Actor's policy with respect to its own parameters. This represents the action value function of the Critic network. This represents the parameter vector of the Actor policy network. This represents the parameter vector of the Critic policy network.

[0160] Furthermore, the SOP switch action instruction can be represented as a mixed-integer programming model as follows:

[0161]

[0162] This indicates the SOP switch action command to be solved. This represents the continuous actions generated by the DDPG policy network. Represents the 2-norm distance. Indicates the number of isolated islands (usually) (Main network single radiation) and These represent the minimum and maximum values ​​of the standard voltage, respectively.

[0163] The disclosed method can be implemented using various types of devices. Therefore, the present invention also discloses an apparatus corresponding to the above method, and specific embodiments are given below for detailed description.

[0164] like Figure 2 As shown, one embodiment of the present invention provides a distribution network control device considering distributed power sources, comprising:

[0165] The distribution network state space establishment module 202 is used to establish the state space equations of the distribution network nodes. The state vectors of the state space equations include the output of distributed generation.

[0166] The state vector quantum compression module 204 is used to perform quantum compression on the state vector to obtain a compressed state vector;

[0167] The power command generation module 206 is used to generate converter power commands corresponding to distribution network nodes based on the compression state vector.

[0168] The reinforcement learning optimization module 208 is used to establish a reinforcement learning state space with real-time measurement data of distribution network nodes as input state and SOP switch action command as action vector. Based on the converter power command, the reinforcement learning state space strategy is optimized and the SOP switch action command is output. The SOP switch action command is used to drive each SOP action in the distribution network.

[0169] The device provided in this application embodiment has the same implementation principle and technical effect as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the aforementioned method embodiment.

[0170] The methods and related apparatuses mentioned in the above embodiments are described with reference to the method flowcharts and / or structural diagrams provided in the embodiments of this application. Specifically, each block of the method flowchart and / or structural diagram, as well as combinations of blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, generate instructions for implementing the process. Figure 1 A schematic diagram of one or more processes and / or structures. Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 A schematic diagram of one or more processes and / or structures. Figure 1 The functions specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 A process or multiple processes and / or structures illustrate the steps of the functions specified in one or more boxes.

[0171] The following embodiments illustrate the application of this method to a computer device. It is understood that the computer device can be any device with computing and processing capabilities, including but not limited to servers or personal laptops. In one embodiment, the computer device can be an application server, which can be a server used to run the application under test.

[0172] See Figure 3This document illustrates a hardware block diagram of an electronic device intended to represent various forms of digital computers, such as laptops, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present application described and / or claimed herein.

[0173] like Figure 3 As shown, the electronic device includes: at least one processor 1, at least one communication interface 2, at least one memory 3, and at least one communication bus 4;

[0174] In this embodiment of the application, the number of processor 1, communication interface 2, memory 3, and communication bus 4 is at least one, and processor 1, communication interface 2, and memory 3 communicate with each other through communication bus 4;

[0175] Processor 1 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention.

[0176] Memory 3 may include high-speed RAM, and may also include non-volatile memory, such as at least one disk storage device;

[0177] The memory stores a program, which the processor can call. The program is used to implement the various processing steps of the aforementioned power distribution network control scheme that considers distributed power sources.

[0178] This invention also provides a readable storage medium storing a computer program thereon, which, when executed by a processor, implements various processing flows of the power distribution network control scheme considering distributed power sources provided in any possible implementation of the above embodiments and / or in combination with the embodiments.

[0179] The invention has been described in particular detail above with respect to possible scenarios, and those skilled in the art will recognize that the invention can be practiced through other embodiments. Specific naming of components, capitalization of terms, attributes, data structures, or any other programming or structural aspects are not mandatory or important, and the mechanisms or features of implementing the invention may have different names, forms, or procedures. The system can be implemented through a combination of hardware and software (as described), entirely through hardware elements, or entirely through software elements. The specific division of functions among the various system components described herein is merely exemplary and not mandatory; rather, the functions performed by a single system component can be performed by multiple components, or the functions performed by multiple components can be performed by a single component.

[0180] Those skilled in the art should understand that the various steps of the disclosed methods can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using device-executable program code, which can then be stored in a storage device for execution by the computing device. Alternatively, they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Therefore, the embodiments disclosed in this invention are not limited to any specific hardware and software combination.

[0181] The programs (also referred to as programs, software, software applications, or code) executable by these computing devices include machine instructions of a programmable processor and can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. As used herein, the terms “machine-readable medium” and “computer-readable medium” refer to any computer program product, device, and / or apparatus (e.g., disk, optical disk, memory, programmable logic device (PLD)) used to provide machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term “machine-readable signal” refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0182] Certain aspects of this invention include the process steps and instructions described herein in algorithmic form. It should be noted that the process steps and instructions of this invention can be implemented in software, firmware, and / or hardware, and when implemented in software, they can be downloaded, stored on various operating systems and operated from said platforms.

[0183] Those skilled in the art will understand that the structures shown in the figures are merely block diagrams of some structures related to the present application and do not constitute a limitation on the terminal device to which the present application is applied. Specific terminal devices may include more or fewer components than those shown in the figures, or combine certain components, or have different component arrangements.

[0184] In the description of this specification, the use of terms such as "one embodiment," "some embodiments," "example," "specific example," or "possible design," etc., refers to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0185] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A distribution network control method considering distributed generation, characterized in that, include: Establish the state space equations for the distribution network nodes, wherein the state vectors of the state space equations include the output of distributed generation sources; The state vector is encoded by qubits to obtain a quantum state of the state vector. The quantum state is then subjected to quantum variational optimization to generate a compressed state of the state vector. The compressed state of the state vector is then subjected to projection measurement to obtain a probability amplitude distribution. The probability amplitude distribution is then inversely encoded and transformed using a pre-trained quantum encoder to obtain a compressed state vector. Generate converter power commands corresponding to distribution network nodes based on the compression state vector; Using real-time measurement data of distribution network nodes as input states and SOP switch action commands as action vectors, a reinforcement learning state space is established. Based on the compressed state vectors and converter power commands, features are extracted from the reinforcement learning state space to obtain higher-order features. The input states are then enhanced based on these higher-order features to obtain higher-order states. The higher-order states are used as policy inputs to optimize the reinforcement learning state space and output SOP switch action commands. These SOP switch action commands are used to drive each SOP action in the distribution network.

2. The method according to claim 1, characterized in that, The state-space equations for establishing distribution network nodes include: Acquire real-time measurement data of distribution network nodes, including the output of distributed power sources; The state-space equations of the distribution network nodes are established using the real-time measurement data as state vectors.

3. The method according to claim 2, characterized in that, The state-space equations of the distribution network nodes are expressed as follows: ， in, Represents the state vector. Indicates the output vector. express The derivative of Represents the state matrix, Represents the control input matrix. Indicates control input, Represents the perturbation input matrix. This indicates a disturbance input. Indicates the output matrix. Indicates the node voltage amplitude. Indicates the phase angle of the node voltage. and These represent the active and reactive power outputs of distributed generation sources, respectively; n represents the number of nodes in the distribution network; and m represents the number of distributed generation sources in the distribution network. The matrices are represented as follows: ， in, and These represent the active power imbalance and reactive power imbalance at the nodes, respectively. and These represent the active power and reactive power of the load, respectively. Indicates active power loss. Represents reactive power loss; V represents the node voltage amplitude. and denoted as nodal conductance and nodal susceptance respectively, with subscripts i, j, and k indicating the node number, where i=j indicates self-conductance or self-susceptance, and i≠j indicates mutual conductance or mutual susceptance; This represents the phase angle difference between two nodes. and Let represent the active power response time constant and the reactive power response time constant of the distributed generation, respectively. and These represent the SOP's control gain on the active and reactive power of distributed generation, respectively. and Let SOP represent the active and reactive power regulation gains of the u-th distributed source, respectively, where u∈(1,m). Represents the identity matrix. The distributed generation output observation matrix is ​​represented by: ， DG stands for Distributed Generation, and PMU stands for Phasor Measurement Unit.

4. The method according to claim 1, characterized in that, The step of generating converter power commands corresponding to distribution network nodes based on the compression state vector includes: The output power of the distributed power source is obtained by decoding and filtering the compressed state vector. Obtain the line communication gain and voltage droop coefficient of the distribution network; The active power command of the converter is calculated based on the output of the distributed power source and the line communication gain, and the reactive power command of the converter is calculated based on the voltage reactive droop coefficient.

5. A distribution network control device considering distributed generation, characterized in that, include: The distribution network state space establishment module is used to establish the state space equations of the distribution network nodes, wherein the state vectors of the state space equations include the output of distributed generation sources; The state vector quantum compression module is used to encode the state vector with qubits to obtain the quantum state of the state vector, perform quantum variational optimization on the quantum state to generate the compressed state of the state vector, perform projection measurement on the compressed state of the state vector to obtain the probability amplitude distribution, and use a pre-trained quantum encoder to perform inverse encoding transformation on the probability amplitude distribution to obtain the compressed state vector. A power command generation module is used to generate converter power commands corresponding to distribution network nodes based on the compression state vector. The reinforcement learning optimization module is used to establish a reinforcement learning state space with real-time measurement data of distribution network nodes as input state and SOP switch action commands as action vectors. Based on the compressed state vector and converter power commands, features are extracted from the reinforcement learning state space to obtain higher-order features. Based on the higher-order features, the input state is enhanced to obtain higher-order states. The higher-order states are used as policy inputs to optimize the reinforcement learning state space and output SOP switch action commands. The SOP switch action commands are used to drive each SOP action in the distribution network.

6. An electronic device, characterized in that, It includes a memory storing computer-executable instructions and a processor, which, when executed by the processor, causes the device to perform the distribution network control method considering distributed power sources as described in any one of claims 1 to 4.

7. A readable storage medium, characterized in that, It stores a computer-executable program that, when executed, can implement the distribution network control method considering distributed generation as described in any one of claims 1 to 4.

Images