Distributed photovoltaic power station power regulation method based on multi-channel data distribution framework

Abstract

The invention relates to the technical field of distributed photovoltaic power station power regulation, provides a distributed photovoltaic power station power regulation method based on a multichannel data distribution framework, and realizes data distribution considering both data transmission efficiency and transmission reliability. The method comprises the following steps: when data interaction about adjusting the power of a distributed photovoltaic power station is to be carried out between multiple MPPT voltage devices and a power adjusting system, a self-adaptive data distribution module inputs multi-channel transmission state data collected by a communication node into a deep Q network to obtain a coding mode of each channel; the adaptive data distribution module obtains an inter-channel distribution proportion according to the relative advantages and disadvantages of the transmission quality between the channels; and the communication node allocates the to-be-interacted data to the corresponding channels according to the inter-channel allocation proportion, encodes the corresponding to-be-interacted data according to the encoding modes of the corresponding channels, and transmits the encoded to-be-interacted data through the corresponding channels.

Description

Technical Field

[0001] The present application relates to the technical field of power regulation of distributed photovoltaic power stations, and in particular to a power regulation method of distributed photovoltaic power stations based on a multi-channel data distribution framework. Background Art

[0002] As power grids gradually become intelligent, digital, and networked, cyber-physical converged energy systems are becoming a trend in future grid development. Distributed photovoltaic power stations, as a new component of future grids, not only support the stable operation of the grid system but also effectively alleviate environmental pollution. However, the intermittent and fluctuating performance of distributed photovoltaic power stations can affect the stability of the grid system, necessitating power regulation.

[0003] However, there are some problems with traditional power regulation methods. Traditional technologies do not take into account that data transmission efficiency and reliability are affected by the underlying data distribution framework, resulting in the power regulation of distributed photovoltaic power stations being unable to fully utilize the advantages of data distribution technology. Summary of the Invention

[0004] Based on this, it is necessary to provide a distributed photovoltaic power station power regulation method based on a multi-channel data distribution framework to address the above technical problems.

[0005] The present application provides a distributed photovoltaic power station power regulation method based on a multi-channel data distribution framework, the method comprising:

[0006] When data exchange regarding power regulation of a distributed photovoltaic power station is to be conducted between the multi-MPPT transformer and the power regulation system, an adaptive data distribution module is used to input multi-channel transmission status data collected by the communication node into the deep Q network to obtain a coding method for each channel; the reward function used to train the deep Q network is related to coding efficiency and transmission reliability, so that the deep Q network outputs a coding method with high coding efficiency when the channel transmission status is good, and outputs a coding method with high transmission reliability when the channel transmission status is poor;

[0007] The adaptive data distribution module is further configured to obtain an allocation ratio between channels based on the relative quality of transmission between channels;

[0008] The communication node is used to distribute the data to be interacted to the corresponding channels according to the distribution ratio between the channels, encode the corresponding data to be interacted according to the encoding method of the corresponding channel, and transmit it between the multi-MPPT transformer and the power regulation system through the corresponding channel.

[0009] In the method provided by the present application, when data interaction regarding regulating the power of a distributed photovoltaic power station is to be carried out between a multi-MPPT transformer and a power regulation system, an adaptive data distribution module is used to input the multi-channel transmission status data collected by the communication node into a deep Q network to obtain a coding method for each channel; the reward function used to train the deep Q network is related to coding efficiency and transmission reliability, so that the deep Q network outputs a coding method with high coding efficiency when the transmission state of the channel is good, and outputs a coding method with high transmission reliability when the transmission state of the channel is poor; the adaptive data distribution module is also used to obtain a distribution ratio between channels based on the relative quality of transmission between channels; the communication node is used to distribute the data to be interacted to the corresponding channel according to the channel distribution ratio, encode the corresponding data to be interacted according to the coding method of the corresponding channel, and then transmit it between the multi-MPPT transformer and the power regulation system through the corresponding channel. This application adopts a deep Q-network (DQN) model to dynamically determine the encoding method of the channel according to the channel transmission state, and obtain the channel allocation ratio based on the relative quality of transmission between channels. The channel allocation ratio is used as one of the data distribution strategies to allocate the data to be interacted to different channels and encode them using the corresponding encoding methods to achieve adaptive optimization of data transmission; and the reward function used to train the deep Q-network is related to the coding efficiency and transmission reliability, so that the deep Q-network outputs a coding method with high coding efficiency when the channel transmission state is good, and outputs a coding method with high transmission reliability when the channel transmission state is poor, thereby achieving data distribution that takes into account both data transmission efficiency and transmission reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0011] Figure 1 FIG1 is an application environment diagram of a distributed photovoltaic power station power regulation method based on a multi-channel data distribution framework in one embodiment;

[0012] Figure 2 1 is a flow chart of a distributed photovoltaic power station power regulation method based on a multi-channel data distribution framework in one embodiment;

[0013] Figure 3 A schematic diagram of a flow chart for generating a control instruction in one embodiment;

[0014] Figure 4 Schematic diagram of a fuzzy control strategy flow in one embodiment;

[0015] Figure 5 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0017] The distributed photovoltaic power station power regulation method based on the multi-channel data distribution framework provided in this application involves Figure 1 Devices shown, such as MPPT transformers, communication nodes, adaptive data distribution modules and power regulation systems.

[0018] MPPT stands for "Maximum Power Point Tracking" in English, and the power regulation system is called "Power Management System" in English, abbreviated as PMS.

[0019] The method provided in this application includes Figure 2 Steps shown.

[0020] In step S201, when data exchange regarding power regulation of distributed photovoltaic power stations is to be carried out between the multi-MPPT transformer and the power regulation system, the adaptive data distribution module is used to input the multi-channel transmission status data collected by the communication node into the deep Q network to obtain the encoding method of each channel.

[0021] The reward function used to train the deep Q network is related to coding efficiency and transmission reliability, so that the deep Q network outputs a coding method with high coding efficiency when the transmission state of the channel is good, and outputs a coding method with high transmission reliability when the transmission state of the channel is poor.

[0022] In distributed photovoltaic power plant power regulation scenarios, the MPPT transformer can transmit PV panel measurement data and device status data to the power regulation system. PV panel measurement data includes current, voltage, power, and temperature; device status data includes the operating status (e.g., fault codes and operating mode) of devices such as the PV inverter and MPPT transformer. Based on the PV panel measurement data and device status data, the power regulation system generates control commands for power regulation in the distributed photovoltaic power plant and feeds them back to the MPPT transformer. The MPPT transformer then adjusts the PV panel power accordingly.

[0023] In the above scenario, the process of the MPPT transformer sending the photovoltaic panel measurement data and equipment status data to the power regulation system belongs to the uplink data transmission process; the process of the power regulation system sending control instructions to the MPPT transformer belongs to the downlink data transmission process.

[0024] In the scenario of multiple MPPT transformers, the solution provided in this application can be used for uplink data transmission and downlink data transmission to take into account both data transmission efficiency and transmission reliability.

[0025] For example, when the multi-MPPT transformer wants to feed back the measurement data and device status data of the photovoltaic panel to the power regulation system, the communication node can collect the latest multi-channel transmission status data and feed it back to the adaptive data distribution module.

[0026] The communication node is used to collect the packet loss rate, signal-to-noise ratio, average delay and bandwidth utilization of each channel to obtain the transmission status data of each channel to form multi-channel transmission status data.

[0027] The average delay can be specifically characterized by the queue length (q), and the bandwidth utilization can be specifically characterized by the packet size (size).

[0028] The packet loss rate (p), signal-to-noise ratio (snr), queue length (q) and packet size (size) of the i-th channel can form the state vector Assuming there are N channels, the multi-channel transmission status data can be expressed as .

[0029] The adaptive data distribution module inputs the multi-channel transmission status data into the deep Q network, and obtains the encoding method of each channel based on the output of the deep Q network.

[0030] This application provides an adaptive data distribution module based on deep reinforcement learning. This module is primarily responsible for dynamically adjusting the encoding method and data distribution strategy based on the multi-channel transmission status (which can be referred to as the transmission network status) to maximize data transmission efficiency and reliability. The adaptive data distribution module includes an adaptive encoding method that uses a deep Q network to dynamically adjust the encoding method. Specifically, it includes:

[0031] (1.1.1) Define the state space of the deep Q network: The state space is composed of the channel state vector, the channel packet loss rate (p), signal-to-noise ratio (snr), queue length (q) and packet size (size), etc., forming the state vector s = [p, snr, q, size].

[0032] State vector Describe the packet loss rate, signal-to-noise ratio, queue length, and packet size of the i-th channel.

[0033] This embodiment may involve four channels, which can be of the following types: one Ethernet channel, two 4G channels, and one Wi-Fi channel. The transmission status of each channel is fed back to the communication node at intervals of one second via the Real-time Transport Control Protocol (RTCP).

[0034] Deep Q network based on multi-channel transmission state data , determine the encoding method of each channel.

[0035] Training a deep Q-network requires training data. This training data can be generated using a transmission network simulation environment (the transmission network topology can be a star network with 4 to 8 channels), simulating packet loss rates p∈[0,0.2], signal-to-noise ratios snr∈[10,30]dB, queue lengths q∈[0,100], and packet sizes size∈[100,1000] bytes.

[0036] (1.1.2) Define the action space of a deep Q-network: The action space is composed of encoding schemes, which include encoding type (e.g., LDPC, Turbo), coding rate, and redundancy. The encoding type can be denoted as encoding_type, the coding rate as rate, and the redundancy as redundancy. Each encoding scheme can be considered an action, and the set of actions can be denoted as A = {encoding_type, rate, redundancy}.

[0037] (1.1.3) Define the reward function of the deep Q network: The reward function takes into account both coding efficiency and transmission reliability. The reward function can be expressed as: .

[0038] in, For coding efficiency, is the data transmission difference corresponding to the j-th MPPT transformer, and is the weight coefficient, which can be adjusted according to actual needs.

[0039] Coding efficiency and data transmission difference The calculation formula is as follows:

[0040] ;

[0041] .

[0042] in, is the binary length of the encoded measurement data and control data of the j-th MPPT transformer, is the size of the transmitted data packet, The length of data successfully decoded by the receiving end.

[0043] (1.1.4) Deep Q-Network: The Deep Q-Network learns the optimal strategy by interacting with the environment, that is, selecting the appropriate encoding method under different transmission network conditions.

[0044] (1.1.5) Deep Q-Network Training Process:

[0045] To achieve the optimal encoding method for the adaptive data distribution module, the deep Q network is trained by interacting with the environment, optimizing the neural network parameters to approximate the optimal Q function Q*(s,a). The training process includes loss function definition, parameter update, and training process, as follows:

[0046] (1.1.5.1) Loss function and parameter update:

[0047] The deep Q network uses the temporal difference (TD) error as the loss function to update the neural network parameters. The loss function is defined as:

[0048] .

[0049] in, is the current reward, i.e. ; is the discount factor, which is greater than or equal to 0 and less than 1, indicating the weight of future rewards; For the deep Q network in state and actions The predicted Q value under is the network parameter; For the deep Q network in the next state and actions The Q value under are the network target parameters (which can be copied from the current network periodically); To select the next state The action with the maximum Q value.

[0050] The loss function optimizes the deep Q network to approximate the true Q value by minimizing the TD error. Parameters are updated using gradient descent combined with the Adam optimizer (adaptive moment estimation optimizer), with a learning rate of 0.001.

[0051] (1.1.5.2) Training process and hyperparameters:

[0052] The deep Q network training process is as follows:

[0053] ① Initialize the deep Q network parameters and network target parameters , set up the experience replay pool (recommended capacity is 10,000).

[0054] ② In state s = [p, snr, q, size], select action a∈{encoding_type, rate, redundancy} through the ε-greedy strategy.

[0055] ③Execute action a, obtain reward r and next state s', and store the experience (s, a, r, s') in the experience replay pool.

[0056] ④ Randomly sample a batch of experience from the replay pool, calculate the TD error, and update the network parameters .

[0057] ⑤ Update the network target parameters every fixed number of steps (recommended 1000 steps), that is, equal .

[0058] ⑥ Repeat steps ② to ⑤ until the deep Q network converges (that is, the reward value stabilizes).

[0059] In some scenarios, the Deep Q network can be fine-tuned online, with the model fine-tuned every hour based on the latest transmission network data, and the number of fine-tuning steps can be 1000.

[0060] Regarding the convergence of the deep Q network, you can set: the deep Q network converges within a specified number of steps and the reward value is stable.

[0061] The key hyperparameters used in training can include:

[0062] The learning rate can be set to 0.001; the discount factor γ can be set to 0.9; the ε-greedy strategy (initial ε equals 1.0, linearly decays to 0.1, and the number of decay steps is 10,000); the experience replay pool capacity can be set to 10,000; the target network update frequency can be set to every 1,000 steps; and the batch size can be set to 64.

[0063] The training data is generated based on the transmission network simulation environment, and offline pre-training is performed based on the historical transmission data of the transmission network, and online fine-tuning is performed during actual operation.

[0064] Step S202: The adaptive data distribution module is further configured to obtain an allocation ratio between channels according to the relative quality of transmission between channels.

[0065] The adaptive data distribution module uses the aforementioned deep Q network to determine the encoding method for each channel. The adaptive data distribution module also uses the data distribution strategy in step S202 to determine the channel allocation ratio, that is, the relative amount of data each channel is responsible for. To further improve data transmission reliability, this embodiment can determine the channel allocation ratio based on the relative quality of transmission between channels. For example, the higher the channel's signal-to-noise ratio, the better the channel's transmission quality, and the greater the amount of data each channel is responsible for.

[0066] Therefore, the present application also provides an embodiment, in which the adaptive data distribution module, when obtaining the distribution ratio between channels based on the relative quality of transmission between channels, is specifically used to: obtain the distribution ratio between channels based on the relative signal-to-noise ratio between channels; the higher the signal-to-noise ratio, the greater the distribution ratio.

[0067] This embodiment is based on the adaptive data distribution module of deep reinforcement learning:

[0068] (1) Core function: This module is the "brain" of data transmission, responsible for intelligently adjusting the data encoding method and data distribution strategy according to the current transmission network environment (that is, the channel transmission status) to maximize transmission efficiency and reliability.

[0069] (2) The output of this module is an encoded and distributed data stream, which is directly input into the "Data Transmission Module Based on the Improved RTCP Protocol". At the same time, this module also receives feedback from the "Multi-objective Intelligent Power Regulation Module Considering Grid Stability" (a module located in the power regulation system) to continuously optimize its own strategy and form a closed-loop control system.

[0070] (3) State perception: Perceive the current state of the transmission network, including the channel’s packet loss rate, signal-to-noise ratio, queue length, packet size, etc. These parameters constitute the state vector, which is a quantitative description of the current transmission network environment.

[0071] (4) Strategy decision: Based on the perceived transmission network status, the pre-trained deep Q network is used to select the optimal action from the action space A, that is, to select the appropriate coding method (including coding type, coding rate, and redundancy). The training goal of the deep Q network is to maximize the aforementioned reward function, which comprehensively considers coding efficiency and transmission reliability. This means that the deep Q network will learn to select a coding method with high coding efficiency when the transmission network condition (i.e., the channel transmission condition) is good, and to select a coding method with high transmission reliability when the transmission network condition (i.e., the channel transmission condition) is poor.

[0072] Step S203, the communication node is used to distribute the data to be exchanged to the corresponding channels according to the distribution ratio between the channels, encode the corresponding data to be exchanged according to the encoding method of the corresponding channel, and transmit it between the multi-MPPT transformer and the power regulation system through the corresponding channel.

[0073] The communication node distributes the data to be exchanged to the corresponding channels so that each channel is responsible for transmitting a certain amount of data to be exchanged, and the ratio of the amount of data each channel is responsible for is consistent with the distribution ratio between channels.

[0074] If a part of the data to be exchanged is allocated to the i-th channel, the communication node uses the encoding method of the i-th channel to encode this part of the data to be exchanged, and transmits the encoded data to the i-th channel.

[0075] To further ensure data transmission reliability, the present application also provides an embodiment in which a data distribution strategy is employed: allocating highly important data to channels with good transmission quality. This data distribution strategy can be executed by communication nodes. When allocating data to be interacted with to corresponding channels according to the inter-channel allocation ratio, it is specifically used to: distribute the data to be interacted with to multiple channels according to the inter-channel allocation ratio and the relative importance of the data, so that the amount of data allocated to the multiple channels satisfies the inter-channel allocation ratio and the highly important data to be interacted with is allocated to channels with a high signal-to-noise ratio.

[0076] In the power regulation scenario of a distributed photovoltaic power plant, the data exchanged between the multi-MPPT transformer and the power regulation system includes control instructions, measurement data, and device status data. If control instructions are of the highest importance, measurement data is of the second highest importance, and device status data is of the lowest importance, then according to the aforementioned data distribution strategy, the following distribution method is used: control instructions are preferentially distributed to channels with a signal-to-noise ratio greater than 20dB, with a proportion greater than or equal to 80%; measurement data is distributed to channels with a signal-to-noise ratio greater than 15dB; and device status data is distributed to the remaining channels. If the delay of the control data exceeds the specified delay value, that is, the control data delay exceeds the standard, the Deep Q network increases redundancy or reallocates channels. The increase in redundancy in the Deep Q network can be understood as increasing the redundancy level in the encoding method of the corresponding channel.

[0077] The feedback-driven approach to data distribution strategies specifically includes an adaptive data distribution module that increases the redundancy in the encoding method of the corresponding channel or reallocates the channel if the delay or packet loss rate of highly important data exceeds the standard.

[0078] In addition, the power regulation system is used to feed back the power regulation result to the adaptive data distribution module; the adaptive data distribution module is used to adjust the encoding method according to the information fed back by the power regulation system.

[0079] The power regulation system can feed back power regulation results (power regulation error) to communication nodes. The communication nodes then feed back data transmission status (such as control data delay and packet loss rate) and power regulation results to the adaptive data distribution module. This data transmission status and power regulation results are used by the adaptive data distribution module to dynamically adjust the encoding method and data distribution strategy. If control data delay or packet loss rate exceeds the specified limit, the Deep Q network increases redundancy or reallocates channels. The Deep Q network can be updated through online fine-tuning.

[0080] In this embodiment, data encoding and distribution are distributed to different channels based on the data distribution strategy. Based on the actions (i.e., encoding methods) selected for each channel by the Deep Q Network, the data for each channel is encoded and then outputted by the corresponding channel. The data distribution strategy is also dynamically adjusted based on the transmission quality of each channel and the importance of the data. For example, more important data is preferentially allocated to channels with better transmission quality.

[0081] In the above-mentioned distributed photovoltaic power station power regulation method based on the multi-channel data distribution framework, when data interaction regarding regulating the power of the distributed photovoltaic power station is to be carried out between the multi-MPPT transformer and the power regulation system, the adaptive data distribution module is used to input the multi-channel transmission status data collected by the communication node into the deep Q network to obtain the encoding method of each channel; the reward function used to train the deep Q network is related to the encoding efficiency and transmission reliability, so that the deep Q network outputs an encoding method with high encoding efficiency when the transmission state of the channel is good, and outputs an encoding method with high transmission reliability when the transmission state of the channel is poor; the adaptive data distribution module is also used to obtain the channel allocation ratio based on the relative quality of transmission between the channels; the communication node is used to distribute the data to be interacted to the corresponding channel according to the channel allocation ratio, encode the corresponding data to be interacted according to the encoding method of the corresponding channel, and then transmit it between the multi-MPPT transformer and the power regulation system through the corresponding channel. This application adopts a deep Q-network (DQN) model to dynamically determine the encoding method of the channel according to the channel transmission state, and obtain the channel allocation ratio based on the relative quality of transmission between channels. The channel allocation ratio is used as one of the data distribution strategies to allocate the data to be interacted to different channels and encode them using the corresponding encoding methods to achieve adaptive optimization of data transmission; and the reward function used to train the deep Q-network is related to the coding efficiency and transmission reliability, so that the deep Q-network outputs a coding method with high coding efficiency when the channel transmission state is good, and outputs a coding method with high transmission reliability when the channel transmission state is poor, thereby achieving data distribution that can take into account both data transmission efficiency and transmission reliability.

[0082] In one embodiment, the data to be exchanged is transmitted based on a target real-time transport control protocol; the target real-time transport control protocol includes a forward error correction coding mechanism and a selective retransmission mechanism.

[0083] The target real-time transport control protocol of this embodiment is an improved RTCP (the full name of RTCP in English is "Real-time Transport Control Protocol", and its Chinese name is Real-time Transport Control Protocol). The target real-time transport control protocol combines a forward error correction coding mechanism and a selective retransmission mechanism to ensure the real-time and integrity of data transmission.

[0084] The target real-time transport control protocol of this embodiment mainly includes the following aspects:

[0085] (1) Core function: It specifies the data transmission method and improves on the traditional RTCP protocol to better meet the high requirements of power system for real-time performance and reliability.

[0086] (2) This protocol can act on measurement data and device status data in uplink data transmission scenarios, and can also act on control instructions in downlink data transmission scenarios. It serves as a communication bridge between the data sender and the data receiver.

[0087] (3) Applying this protocol to data includes the following aspects:

[0088] (3.1) Process the encoded data stream using the target real-time transport control protocol.

[0089] (3.2) Timestamp Addition and Synchronization: Timestamps are added to each data packet, and an improved timestamp synchronization algorithm is used to ensure the sequential nature of data transmission and time synchronization between units. Time synchronization is crucial for subsequent power regulation algorithms, as these algorithms require precise time information for coordinated control.

[0090] (3.3) Forward Error Correction Coding: Forward error correction coding (such as RS code) is performed on the data to increase data redundancy, improve anti-interference capabilities, and reduce data loss caused by channel noise and other factors.

[0091] (3.4) Data transmission: Data packets are sent to the receiving end through the channel.

[0092] (3.5) Selective retransmission: At the receiving end, if packet loss is detected, a retransmission request is sent through the target real-time transport control protocol. This can request only the retransmission of the lost data packet instead of retransmitting all data, thereby improving transmission efficiency.

[0093] For time synchronization, the target real-time transport control protocol uses a master-slave time synchronization mechanism based on the RTCP protocol and introduces a timestamp synchronization algorithm for optimization to reduce synchronization errors. Timestamps are not only used to ensure the timing of data packets, but also for accurate processing of time information in subsequent power regulation algorithms.

[0094] In terms of forward error correction coding mechanism, the target real-time transport control protocol can use Reed-Solomon (RS) code for forward error correction coding to improve the reliability of data transmission and reduce the number of retransmissions.

[0095] In the target real-time transport control protocol, three types of data packets can be designed: control instruction data packets, measurement data packets, and device status data packets. The format and content of the data packets are defined. Specifically:

[0096] ①Control instruction data packet: contains power adjustment instructions, control parameters, etc.

[0097] ②Measurement data packet: contains measurement data such as voltage, current, power, and temperature.

[0098] ③ Status data packet: Contains the operating mode (normal, standby, fault) and fault code (such as overvoltage, overcurrent) of the photovoltaic inverter; also includes the tracking status of the MPPT voltage regulator; and also includes the transmission network status (that is, channel transmission status).

[0099] In one embodiment, the power conditioning system may perform Figure 3 The steps shown are:

[0100] In step S301, based on the control decision-related data required for power regulation of the distributed photovoltaic power station fed back by the multi-MPPT voltage regulator, the grid operation stability assessment result is obtained; in step S302, based on the grid operation stability assessment result, the multi-objective optimization algorithm and the fuzzy control strategy, the control instructions are generated; the control instructions belong to the data to be interacted.

[0101] The power regulation system can be equipped with a "multi-objective intelligent power regulation module considering grid stability", which mainly includes the following aspects:

[0102] (1) Core function: This module is the “decision center” of the power regulation system, responsible for performing intelligent power regulation based on the received data and grid status to ensure the stable operation of the grid.

[0103] (2) This module receives measurement data and equipment status data and feeds back control instructions to the MPPT voltage regulator. At the same time, the feedback information of this module serves as the input of the adaptive data distribution module, forming a complete closed-loop control system.

[0104] (3.1) Data reception and decoding: Receive data packets transmitted by the channel and decode them to restore the original measurement data and device status data.

[0105] (3.2) Grid stability assessment: Based on the decoded data, the operational stability of the current grid is assessed, such as by calculating indicators such as voltage deviation, frequency deviation, and power fluctuation rate.

[0106] (3.3) Power Regulation Strategy Development: Based on the grid stability assessment results, a power regulation strategy is developed and control instructions are generated by combining fuzzy control strategies and multi-objective optimization algorithms. The fuzzy control strategy is responsible for responding to rapid grid disturbances, while the multi-objective optimization algorithm is responsible for global optimization over longer timescales. The multi-objective optimization algorithm can utilize a Pareto-dominated multi-objective genetic algorithm (NSGA-II) to optimize the power regulation strategy.

[0107] (3.4) Control command transmission: The generated control command is fed back to the MPPT voltage controller according to the target real-time transmission control protocol to control the output power of the photovoltaic panel.

[0108] (3.5) Feedback: The power regulation results are fed back to the adaptive data distribution module through the communication node, so that the adaptive data distribution module can dynamically adjust the encoding method and data distribution strategy based on the data transmission status to form a closed-loop control.

[0109] The power regulation system of this embodiment determines the grid operation stability assessment result based on the received data, adopts a multi-objective optimization algorithm and fuzzy control strategy to realize intelligent power regulation of distributed photovoltaic power stations and ensure the stable operation of the grid.

[0110] In one embodiment, the multi-objective optimization function acted upon by the multi-objective optimization algorithm includes: maximizing the power generation function, minimizing the energy loss function, and minimizing the grid voltage deviation function.

[0111] The multi-objective optimization algorithm can use a multi-objective genetic algorithm based on Pareto dominance (NSGA-II) to optimize the power regulation strategy.

[0112] The objective functions of the multi-objective optimization algorithm include maximizing generated power, minimizing energy loss, and minimizing grid voltage deviation. This embodiment considers the three objectives of maximizing generated power, minimizing energy loss, and minimizing grid voltage deviation to construct the multi-objective optimization function.

[0113] The maximum power generation function can be specifically expressed as .

[0114] The minimization energy loss function can be specifically expressed as .

[0115] The function of minimizing the grid voltage deviation can be specifically expressed as .

[0116] in, Indicates the total power generation of distributed photovoltaic power stations; Represents the control variable vector; including the output power of each photovoltaic inverter, etc.; represents the current on line ij; represents the resistance of line ij; represents the voltage at node i; Indicates the reference voltage.

[0117] The constraints applied by the multi-objective optimization algorithm include, in addition to the constraints on voltage (|V|≤V_max) and current (|I|≤I_max), power factor constraint (0.95≤PF≤1), harmonic content constraint (THD≤5%), and grid frequency stability constraint (Δf≤±0.2Hz).

[0118] In one embodiment, the power regulation system generates control instructions by combining the fuzzy control strategy, specifically including: Figure 4 The steps shown are:

[0119] In step S401, the current grid voltage deviation and the current grid frequency deviation are obtained based on the grid operation stability assessment results. In step S402, the fuzzy linguistic variables that match the current grid voltage deviation and the current grid frequency deviation are determined, and the fuzzy linguistic variables that match the power adjustment amount output by the photovoltaic inverter are determined in combination with the preset fuzzy control rules. In step S403, a control instruction is obtained based on the fuzzy linguistic variables that match the power adjustment amount output by the photovoltaic inverter.

[0120] In response to fluctuations in grid voltage and frequency, this embodiment adopts a fuzzy control strategy for rapid response and regulation.

[0121] The input variables of the fuzzy control strategy are the current grid voltage deviation ΔV and the current grid frequency deviation Δf. The output variables are the power adjustment values (active power adjustment ΔP and reactive power adjustment ΔQ) output by the PV inverter. The fuzzy linguistic variables involved in the fuzzy control strategy include negative small, zero, positive small, and positive large.

[0122] Exemplarily, the fuzzy control rules may include:

[0123] If the fuzzy linguistic variables matching the current grid voltage deviation and the current grid frequency deviation are both negative and large, then the fuzzy linguistic variable matching the active power adjustment amount is also negative and large;

[0124] If the fuzzy linguistic variable matching the current grid voltage deviation is large in negative, and the fuzzy linguistic variable matching the current grid frequency deviation is small in negative, then the fuzzy linguistic variable matching the active power adjustment amount is small in negative;

[0125] If the fuzzy linguistic variables matching the current grid voltage deviation and the current grid frequency deviation are both positive, then the fuzzy linguistic variable matching the active power adjustment amount is also positive.

[0126] According to the fuzzy language variables matched with the power adjustment amount output by the photovoltaic inverter, the control instructions can be obtained. Specifically:

[0127] The output active power of the photovoltaic inverter is: ;The output reactive power of the photovoltaic inverter is: .in, is the power obtained by the MPPT algorithm, is the reference reactive power (usually 0). and The calculation formula is: , .in, and Represents fuzzy reasoning.

[0128] In terms of grid operation stability assessment, the power regulation system can monitor grid voltage, grid frequency, grid power and other parameters in real time, and calculate indicators for grid operation stability assessment, such as grid voltage deviation, grid frequency deviation, grid power fluctuation rate, etc.

[0129] The decision-making process of the power regulation system mainly includes: receiving decoded data, including measurement data and equipment status data; obtaining the grid operation stability assessment results based on the grid operation stability assessment indicators; calculating the required active power adjustment and reactive power adjustment according to the fuzzy control strategy; and optimizing the power regulation strategy and generating control instructions through the NSGA-II algorithm, while satisfying the constraints.

[0130] The power regulation system sends control instructions to each MPPT transformer in the form of a control instruction data packet. Specifically: the power regulation system sends the control instructions to the communication node; the communication node obtains the latest multi-channel transmission status data; the adaptive data distribution module inputs the latest multi-channel transmission status data into the deep Q network, and the deep Q network outputs an action to optimize the encoding method and channel allocation ratio of each channel, ensuring that the control data transmission delay is less than the specified delay value and the packet loss rate is less than the specified packet loss rate value.

[0131] Among them, the control formulas of voltage and current are:

[0132] ;

[0133] .

[0134] in, is the control voltage on the grid side, and is the voltage controller gain, is the reference voltage, is the actual grid voltage, Is the integral term, in control theory, especially in PID controller, it represents the accumulation of error signal, is the feedforward compensation term; is the control current on the grid side, and is the current controller gain, is the reference current, is the actual grid current.

[0135] In order to better understand the above method, an application example of the distributed photovoltaic power station power regulation method based on the multi-channel data distribution framework of the present application is described in detail below.

[0136] Traditional power regulation methods present several challenges. First, they fail to consider how data transmission efficiency and reliability are affected by the underlying data distribution framework. Consequently, distributed PV power plant power regulation cannot fully leverage the advantages of data distribution technology. Second, while time synchronization-based methods can ensure data timeliness, they struggle to guarantee continuity and stability over long time series. Furthermore, PV output power is closely linked to the external environment, leading to uncertainty in the output power of distributed PV power plants. This necessitates a control strategy that comprehensively considers both transmission efficiency and stability to ensure stable output.

[0137] The shortcomings of traditional technologies in data transmission efficiency and reliability are specifically reflected in the following aspects:

[0138] (1) Fixed coding method: Traditional communication methods usually use a fixed coding method, which cannot be adaptively adjusted according to changes in the transmission network environment. This leads to low data transmission efficiency when the channel conditions are poor, and even a large number of packet losses, affecting control performance.

[0139] (2) Simple retransmission mechanism: Some systems use a simple retransmission mechanism to ensure data reliability. However, this method will introduce large delays when the transmission network is congested and is not suitable for power systems with high real-time requirements.

[0140] (3) Lack of effective multi-channel management: Traditional technologies often lack effective management of multi-channel data distribution and cannot fully utilize the advantages of multiple communication channels to improve data transmission efficiency and reliability.

[0141] (4) The grid stability control strategy is not intelligent enough:

[0142] (4.1) Single-objective optimization: Traditional power regulation methods usually only consider a single optimization objective, such as maximizing generated power or minimizing energy loss, ignoring the importance of grid stability.

[0143] (4.2) Lack of rapid response capability: Some control methods have a slow response speed and are unable to respond to rapid disturbances in the power grid in a timely manner, which can easily lead to excessive fluctuations in grid voltage and frequency, and even cause safety accidents.

[0144] (4.3) Poor robustness: Traditional control methods are highly dependent on grid parameters and operating conditions. When the grid environment changes, the control performance will be significantly reduced.

[0145] (5) Low level of intelligence:

[0146] (5.1) Lack of adaptability: Traditional systems lack adaptability and cannot automatically adjust to changes in the transmission network environment and grid status. They require manual intervention, increasing operation and maintenance costs.

[0147] (5.2) Lack of closed-loop control: Some systems lack effective feedback mechanisms, making it impossible to continuously optimize and improve system performance.

[0148] The solution provided in this application example uses a deep Q-network to dynamically adjust the coding method (coding type, coding rate, redundancy) and data distribution strategy based on the transmission network status (channel packet loss rate, signal-to-noise ratio, queue length, etc.), achieving adaptive optimization of data transmission. Furthermore, based on the traditional RTCP protocol, it combines forward error correction coding (RS code) and a selective retransmission mechanism to improve the reliability and efficiency of data transmission. Furthermore, a multi-objective genetic algorithm (NSGA-II) based on Pareto dominance is used to comprehensively consider multiple objectives such as maximizing power generation, minimizing energy loss, and minimizing grid voltage deviation. Combined with a fuzzy control strategy, it achieves rapid response to grid fluctuations and global optimization.

[0149] This application example provides a multi-channel data distribution framework that enables efficient and reliable data transmission between units in a distributed photovoltaic power plant and supports intelligent power regulation. This framework comprises three components: an adaptive data distribution module based on deep reinforcement learning, data transmission based on the Targeted Real-Time Transmission Control Protocol, and a multi-objective power regulation algorithm that considers grid stability.

[0150] In the uplink data transmission scenario, the MPPT voltage regulator can send measurement data and equipment status data (collectively referred to as relevant data required for control decisions) to the communication node; the communication node sends the latest multi-channel transmission status data (updated once per second) to the adaptive data distribution module; the adaptive data distribution module inputs the multi-channel transmission status data into the deep Q network, and obtains the encoding method of each channel based on the action output by the deep Q network; the adaptive data distribution module obtains the channel allocation ratio based on the relative quality of transmission between channels; the communication node allocates the relevant data required for control decisions to the corresponding channels according to the said channel allocation ratio, encodes the relevant data required for control decisions according to the encoding method of the corresponding channel, and then sends the relevant data required for control decisions to the power regulation system through the corresponding channel in combination with the target real-time transmission control protocol.

[0151] In the decision-making scenario, the power regulation system receives the transmitted data packets and obtains the relevant data (including measurement data and equipment status data) required for control decision-making after decoding based on the target real-time transmission control protocol and the corresponding decoding method. The power regulation system evaluates the grid operation stability based on the relevant data required for control decision-making and obtains the grid operation stability assessment result. The power regulation system generates control instructions based on the grid operation stability assessment result, the fuzzy control strategy and the NSGA-II algorithm.

[0152] In downlink data transmission scenarios, the power regulation system sends control instructions to communication nodes according to the packet format specified by the target real-time transmission control protocol. The communication nodes then send the latest multi-channel transmission status data (updated once per second) to the adaptive data distribution module. The adaptive data distribution module inputs the multi-channel transmission status data into the deep Q network and, based on the actions output by the deep Q network, determines the encoding scheme for each channel. The adaptive data distribution module determines the channel allocation ratio based on the relative transmission quality between channels. The communication nodes then allocate control instructions to the corresponding channels according to this channel allocation ratio. After encoding the control instructions according to the encoding scheme of the corresponding channel, they send the control instructions to the photovoltaic inverter (MPPT voltage regulator is a component of the photovoltaic inverter) through the corresponding channel in accordance with the target real-time transmission control protocol. The photovoltaic inverter then obtains the active power adjustment ΔP and reactive power adjustment ΔQ based on the control instructions, which the MPPT voltage regulator uses to adjust the operating point of the photovoltaic panel.

[0153] In terms of feedback and optimization, communication nodes can monitor transmission performance and generate data transmission status. They can also receive power regulation results from the power regulation system. The communication nodes encapsulate the data transmission status and power regulation results into status packets and send them to the adaptive data distribution module. The adaptive data distribution module adjusts the Deep Q network's strategy based on the data transmission status and power regulation results. If control data delay or packet loss rate exceeds the specified limit, the Deep Q network increases redundancy or reallocates channels. The Deep Q network is then updated through online fine-tuning. The power regulation system can use this feedback to indirectly optimize control command transmission (for example, prioritizing high-quality channels).

[0154] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0155] The above communication node or power regulation system can be realized by computer equipment. The internal structure diagram of the computer equipment can be as follows: Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store data involved in the above method. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a distributed photovoltaic power station power regulation method based on a multi-channel data distribution framework is implemented.

[0156] Those skilled in the art will understand that Figure 5The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0157] In one embodiment, a communication node is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps performed by the above-mentioned communication node when executing the computer program.

[0158] In one embodiment, a power regulation system is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps performed by the above-mentioned power regulation system when executing the computer program.

[0159] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above embodiments are implemented.

[0160] In one embodiment, a computer program product is provided, on which a computer program is stored. The computer program is used by a processor to execute the steps in the above embodiments.

[0161] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0162] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0163] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0164] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A distributed photovoltaic power station power regulation method based on a multi-channel data distribution framework, characterized in that: The method comprises: When data exchange regarding power regulation of a distributed photovoltaic power station is to be conducted between the multi-MPPT transformer and the power regulation system, an adaptive data distribution module is used to input multi-channel transmission status data collected by the communication node into the deep Q network to obtain a coding method for each channel; the reward function used to train the deep Q network is related to coding efficiency and transmission reliability, so that the deep Q network outputs a coding method with high coding efficiency when the channel transmission status is good, and outputs a coding method with high transmission reliability when the channel transmission status is poor; The adaptive data distribution module is further configured to obtain an allocation ratio between channels based on the relative quality of transmission between channels; The communication node is used to distribute the data to be interacted to the corresponding channels according to the distribution ratio between the channels, encode the corresponding data to be interacted according to the encoding method of the corresponding channel, and transmit it between the multi-MPPT transformer and the power regulation system through the corresponding channel.

2. The method according to claim 1, characterized in that The communication node is configured to: The packet loss rate, signal-to-noise ratio, average delay and bandwidth utilization of each channel are collected to obtain the transmission status data of each channel to form multi-channel transmission status data.

3. The method according to claim 1, characterized in that When the adaptive data distribution module obtains the distribution ratio between channels according to the relative quality of transmission between channels, it is specifically used to: The allocation ratio between channels is obtained according to the relative signal-to-noise ratio between channels; the higher the signal-to-noise ratio, the greater the allocation ratio.

4. The method according to claim 1, wherein When the communication node distributes the to-be-interacted data to the corresponding channels according to the distribution ratio between the channels, the communication node is specifically configured to: According to the distribution ratio between channels and the relative importance of data, the data to be interacted is distributed to multiple channels, so that the amount of data distributed to multiple channels meets the distribution ratio between channels and the data to be interacted with high importance is distributed to the channel with high signal-to-noise ratio.

5. The method according to claim 1, wherein The power regulation system is used to: Based on the control decision-related data required for power regulation of distributed photovoltaic power stations fed back by multiple MPPT voltage regulators, the grid operation stability assessment results are obtained, and control instructions are generated based on the grid operation stability assessment results, multi-objective optimization algorithm and fuzzy control strategy; the control instructions belong to the data to be interacted.

6. The method according to claim 5, characterized in that The multi-objective optimization functions acted upon by the multi-objective optimization algorithm include: maximizing the power generation function, minimizing the energy loss function, and minimizing the grid voltage deviation function.

7. The method according to claim 5, characterized in that The power regulation system is specifically used to generate control instructions in combination with the fuzzy control strategy: According to the grid operation stability evaluation result, a current grid voltage deviation and a current grid frequency deviation are obtained; Determine the fuzzy linguistic variables that match the current grid voltage deviation and the current grid frequency deviation, and determine the fuzzy linguistic variables that match the power adjustment amount output by the photovoltaic inverter in combination with the preset fuzzy control rules; The control instructions are obtained according to the fuzzy linguistic variables matched with the power adjustment amount output by the photovoltaic inverter.

8. The method according to claim 1, characterized in that The power regulation system is used to feed back the power regulation result to the adaptive data distribution module; The adaptive data distribution module is used to adjust the encoding method according to the information fed back by the power regulation system.

9. The method according to claim 8, characterized in that The adaptive data distribution module is used to increase the redundancy in the encoding method of the corresponding channel or reallocate the channel if the delay or packet loss rate of the highly important data exceeds the standard.

10. The method according to claim 1, characterized in that The data to be interacted is transmitted based on a target real-time transport control protocol; the target real-time transport control protocol includes a forward error correction coding mechanism and a selective retransmission mechanism.

Images