Flexible control method for new energy grid connection

Abstract

The invention discloses a flexible control method for new energy grid connection, and relates to the technical field of new energy grid connection control. The method comprises the steps of obtaining grid-connected point electrical parameters through a self-measurement mode or a communication interface mode, performing harmonic feature extraction and compensation based on a deep learning network, and adopting a state estimation method to fuse multi-source data to estimate power grid parameters; and constructing a multi-objective optimization function including voltage quality, a power factor and harmonic suppression, dynamically adjusting a weight coefficient of the optimization function by using a swarm intelligence algorithm, and realizing inverter switching sequence optimization through multi-objective model prediction control. Through a dual-mode communication redundancy mechanism and local intelligent control, while a communication function of a traditional main link is reserved, a standby wireless communication link is introduced to realize real-time backup transmission of a control instruction, and through a local autonomous decision-making capability, the problem of control failure caused by a communication problem in a traditional scheme is fundamentally solved.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy grid-connected control, and in particular to a flexible control method for new energy grid-connected control. Background Art

[0002] Currently, grid-connected renewable energy sources (such as photovoltaics) typically achieve rigid control through grid-connected switches. Protocol converters receive commands from remote devices (such as concentrators) and forward them to power converters (such as photovoltaic grid-connected inverters), controlling their output to achieve flexible regulation. At the same time, some new devices combine grid-connected switches and protocol converters into a single device, integrating the functions of both.

[0003] However, this solution has a core technical bottleneck: the command transmission architecture based on a single power line carrier communication (HPLC) makes the system highly dependent on remote communication, limiting its real-time control capabilities and reliability.

[0004] Due to the inherent characteristics of HPLC communication and its master-multiple-slave network architecture, control commands are susceptible to grid noise, equipment load fluctuations, and other factors during transmission, making it difficult to meet the dynamic regulation requirements of new energy grid-connected systems for parameters such as voltage, power factor, and harmonics. More critically, in the event of abnormalities such as long-distance communication link failures or HPLC module failures, control strategies that rely on remote commands risk complete failure, leading to uncontrolled grid connection point operations and even grid security issues. Summary of the Invention

[0005] The purpose of the present invention is to provide a flexible control method for new energy grid connection to solve the problems raised in the above background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a flexible control method for new energy grid connection, comprising the following steps: Obtain grid connection point electrical parameters through self-measurement mode or communication interface mode, extract and compensate harmonic features based on deep learning networks, and estimate grid parameters using state estimation methods to fuse multi-source data; Construct a multi-objective optimization function that includes voltage quality, power factor, and harmonic suppression. Use a swarm intelligence algorithm to dynamically adjust the weight coefficients of the optimization function. Optimize the inverter switching sequence through multi-objective model predictive control. A multidimensional feature vector is constructed by integrating frequency deviation, voltage deviation, and harmonic impedance change rate. The islanding risk index is calculated based on robust control theory. When the index exceeds the threshold, the energy storage system is activated to maintain local load stability. Among them, the main link uses power line carrier communication to transmit control instructions, and the backup link uses wireless wide area communication to ensure the real-time transmission of instructions when the main link is interrupted.

[0007] Preferably, the method of obtaining the grid connection point electrical parameters through the self-measurement mode or the communication interface mode, extracting and compensating harmonic features based on a deep learning network, and estimating the grid parameters by fusing multi-source data using a state estimation method includes: Self-measurement mode directly measures the voltage, current and harmonic parameters of the grid connection point through the built-in analog-to-digital conversion unit; The communication interface mode obtains the power data of the grid-connected energy meter through a wired or wireless communication interface; The deep learning network outputs harmonic compensation based on the trained prediction model to achieve active suppression of total harmonic distortion.

[0008] Preferably, the deep learning network outputs harmonic compensation based on a trained prediction model, including: The input layer receives the time domain or frequency domain characteristics of the grid connection point electrical parameters; The middle layer contains multiple nonlinear transformation layers to extract harmonic features; The output layer outputs harmonic compensation instructions through regression analysis.

[0009] Preferably, the dynamically adjusting the weight coefficient of the optimization function using a swarm intelligence algorithm includes: Initialize the position and velocity vectors of the agent group; Iteratively update the agent's position to converge to the global optimal solution; The iteration termination condition is that the optimization function converges or the maximum number of iterations is reached.

[0010] Preferably, the inverter switching sequence optimization is realized by multi-objective model predictive control, comprising: The control problem is transformed into a finite-time optimization problem, and the optimal control sequence is calculated through the prediction function; Adopt fixed switching frequency control strategy to reduce switching losses of power electronic equipment; Maintain voltage regulation accuracy within the preset range.

[0011] Preferably, the multi-objective optimization function is: ; in, For the The actual value of the control target, is the reference value, is the weight matrix, is the dynamically adjusted weight coefficient, satisfying ; The swarm intelligence algorithm optimizes the weight coefficient vector by iterative updating .

[0012] Preferably, the method comprises: fusing frequency deviation, voltage deviation, and harmonic impedance change rate to construct a multidimensional feature vector; calculating an islanding risk index based on robust control theory; and starting the energy storage system to maintain local load stability when the index exceeds a threshold. The method includes: Constructing multidimensional feature vectors ,in, is the frequency deviation, is the voltage deviation, is the harmonic impedance change rate; Then, the island risk index is calculated based on robust control theory ,when When the preset threshold is exceeded, it is determined to be an island state; The energy storage system is then started, and the local load voltage and frequency are maintained within the allowable range through the power regulation unit.

[0013] Preferably, the main link uses power line carrier communication to transmit control instructions, and the backup link uses wireless wide area communication to ensure real-time instruction transmission when the main link is interrupted, including: The main link uses power line carrier communication technology to transmit control instructions; The backup link uses low-power wide-area IoT communication technology to ensure that the command transmission delay meets real-time control requirements when the main link is interrupted; When the primary link loses a preset number of heartbeat packets continuously, it automatically switches to the backup link and synchronizes data to the remote monitoring center.

[0014] The present invention also provides a flexible automatic control device for connecting new energy to the grid, comprising: Multimodal data acquisition module, used to obtain grid connection point electrical parameters through self-measurement mode or communication interface mode, extract and compensate harmonic characteristics based on deep learning network, and estimate grid parameters by fusing multi-source data using state estimation method; A hybrid intelligent control module is used to construct a multi-objective optimization function that includes voltage quality, power factor, and harmonic suppression. It uses a swarm intelligence algorithm to dynamically adjust the weight coefficients of the optimization function and optimizes the inverter switching sequence through multi-objective model predictive control. The islanding detection and fault-tolerant control module is used to integrate frequency deviation, voltage deviation, and harmonic impedance change rate to construct a multidimensional feature vector. Based on robust control theory, it calculates the islanding risk index and activates the energy storage system to maintain local load stability when the index exceeds the threshold. The dual-mode communication module is used to transmit control instructions using power line carrier communication on the main link and wireless wide area communication on the backup link to ensure real-time instruction transmission when the main link is interrupted.

[0015] Compared with the prior art, the present invention has the following beneficial effects: Through multimodal data acquisition and deep learning networks, combined with deep learning and state estimation, the device can be flexibly adjusted independently based on the grid-connected point's own parameters, without relying on large-scale remote data acquisition, greatly simplifying the control strategy; by constructing a multi-objective optimization function, the particle swarm optimization algorithm is used to dynamically adjust the weight coefficients of the multi-objective model predictive control, effectively improving the multi-objective collaborative adjustment capability, so that the operating status of each grid-connected point in the grid-connected system is within a reasonable range, thereby ensuring the stability of the overall system; by constructing a multi-dimensional feature vector to detect the island state, fast fault-tolerant control is achieved and the reliability of island detection is enhanced; through dual-mode communication redundancy design, real-time transmission of control instructions is guaranteed when communication is interrupted, and the reliability and flexibility of new energy grid-connected control are comprehensively improved. In addition to being suitable for new energy grid-connected scenarios, it is also applicable to the control of charging and discharging systems of charging piles and similar scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A main flow chart of a flexible control method for new energy grid connection provided by an embodiment of the present invention; Figure 2 A schematic structural diagram of a flexible automatic control device for connecting new energy to the grid provided by an embodiment of the present invention; Figure 3 A diagram showing the implementation steps of a flexible automatic control device for connecting new energy to the grid provided by an embodiment of the present invention; Figure 4 This is a diagram of the implementation steps of another flexible automatic control device for new energy grid connection provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0018] See also Figure 1 The present invention provides a flexible control method for new energy grid connection, which is applied to the flexible control of distributed energy grid connection scenarios such as photovoltaic power generation and wind power generation, and includes: Step 100: Obtain grid connection point electrical parameters through self-measurement mode or communication interface mode, perform harmonic feature extraction and compensation based on a deep learning network, and estimate grid parameters by fusing multi-source data using a state estimation method.

[0019] As can be understood, step 100 utilizes a combination of self-measurement mode and communication interface mode to achieve real-time and accurate acquisition of electrical parameters such as voltage, current, and frequency at the grid connection point. On this basis, a deep learning network is introduced to construct a harmonic feature extraction model, which can efficiently identify complex harmonic components in the power grid and perform targeted compensation to improve power quality. Simultaneously, state estimation methods are used to deeply fuse multi-source data from different sensors and communication channels to accurately estimate grid parameters, providing a reliable basis for the formulation of subsequent control strategies.

[0020] Specifically, step 100 includes: Step 110 , in self-measurement mode, the voltage, current and harmonic parameters of the grid connection point are directly measured through the built-in analog-to-digital conversion unit; The self-measurement mode uses a built-in high-precision analog-to-digital converter (ADC) and oversampling and digital filtering technology to perform real-time synchronous sampling of the three-phase voltage and current at the grid connection point. The sampling frequency can reach above 5kHz, ensuring accurate capture of the fundamental wave and each harmonic component. At the same time, the fast Fourier transform (FFT) algorithm is used to perform spectral analysis on the sampled data to calculate key parameters such as the harmonic content rate of voltage and current and the total harmonic distortion (THD), providing basic data for system operation status assessment. Step 120, the communication interface mode obtains the power data of the grid-connected electric energy meter through the wired or wireless communication interface; The communication interface mode supports wired communication protocols such as RS485 and Ethernet, as well as wireless communication technologies such as LoRa and 5G, to achieve two-way data interaction with grid-connected energy meters. Through standard communication protocols such as Modbus and DL / T645, data such as active power, reactive power, apparent power, and power factor recorded by the energy meter can be obtained in real time. To ensure the reliability and real-time performance of data transmission, a breakpoint resume mechanism and data verification technology are used to ensure the integrity and accuracy of the collected data. In step 130 , the deep learning network outputs harmonic compensation based on the trained prediction model to achieve active suppression of total harmonic distortion.

[0021] The harmonic prediction model, trained using a deep learning network based on historical operating data, utilizes a long short-term memory (LSTM) network combined with a convolutional neural network (CNN) architecture to effectively capture the time series and frequency domain characteristics of harmonic variations. The model inputs include real-time collected voltage and current harmonic parameters, power data, and external influencing factors such as ambient temperature and light intensity. The model infers the optimal harmonic compensation value. Based on the prediction results, the system proactively adjusts the inverter control strategy to proactively suppress total harmonic distortion (THD), ensuring that grid-connected power quality meets the requirements of national standards such as GB / T14549. Furthermore, step 130 includes: Step 131, the input layer receives the time domain or frequency domain characteristics of the grid connection point electrical parameters; Step 132: the middle layer includes multiple nonlinear transformation layers to extract harmonic features; Step 133: The output layer outputs harmonic compensation instructions through regression analysis.

[0022] Optionally, the input layer serves as the starting point for data processing and uses an adaptive normalization algorithm to extract time-domain or frequency-domain features of electrical parameters such as voltage amplitude, phase, and frequency at the grid connection point. This layer is designed with a dynamic window mechanism that automatically adjusts the sampling window size based on the grid operation status to ensure that the most representative feature data is captured. The middle layer consists of eight cascaded deep convolutional neural network (CNN) layers and four long short-term memory (LSTM) layers. The CNN layer uses multi-layer convolution kernel scanning to extract multi-scale features from the input data, effectively capturing the frequency domain distribution characteristics of high-order harmonics. The LSTM layer is responsible for analyzing the temporal variation of harmonics, solving nonlinear and non-stationary problems that are difficult to handle with traditional methods. The residual connection structure is used between the layers to improve network training efficiency. The output layer is based on a deep reinforcement learning framework and combines the real-time operation constraints of the power grid to generate optimal harmonic compensation instructions through regression analysis. This layer introduces a multi-objective optimization algorithm to meet the harmonic suppression requirements while taking into account indicators such as minimizing active power loss and reactive power compensation accuracy to ensure the feasibility and economy of the compensation instructions.

[0023] Step 200: construct a multi-objective optimization function including voltage quality, power factor and harmonic suppression, dynamically adjust the weight coefficient of the optimization function using a swarm intelligence algorithm, and optimize the inverter switching sequence through multi-objective model predictive control.

[0024] As can be understood, step 200 aims to improve the performance of renewable energy grid integration by constructing a multi-objective optimization function encompassing voltage quality, power factor, and harmonic suppression, comprehensively considering multiple key grid operation indicators. To achieve a dynamic balance between these objectives under different operating conditions, swarm intelligence algorithms (such as particle swarm optimization and ant colony algorithms) are utilized to dynamically adjust the weight coefficients of the optimization function based on real-time grid conditions. Model predictive control technology is used to predict and optimize the inverter switching sequence in advance, effectively reducing switching losses and improving system efficiency.

[0025] The multi-objective optimization function is constructed using the weighted sum method, which converts multiple conflicting optimization objectives into a single objective function through weight distribution, thereby achieving collaborative optimization of the multi-dimensional performance of the new energy grid-connected system. The multi-objective optimization function is: ; in, For the The actual values ​​of the control targets are of great engineering significance in the scenario of new energy grid connection: Active power: directly affects the power supply capacity of the power grid, ensuring that the output power of renewable energy power generation equipment matches the load demand of the power grid; Reactive power: It is crucial to maintain grid voltage levels and improve power factor, effectively reducing transmission line losses; Voltage stability: It is a key indicator to ensure the safe and reliable operation of the power grid, and to avoid equipment damage or system collapse due to excessive voltage fluctuations; It is a reference value, usually determined by system operation requirements or strict safety standards to ensure that the power grid operates in a stable and efficient range. is a weight matrix whose element values ​​can be finely adjusted according to the dynamic characteristics of the system. It can perform scale normalization processing based on the dimensional differences and fluctuation characteristics of different control objectives to ensure the fairness of each objective in the optimization process. is a dynamically adjusted weight coefficient that reflects the relative importance of each control target under different working conditions and satisfies the normalization condition. For example, during peak load periods, the weight of the active power target can be appropriately increased; while in a weak grid environment, the weight of the voltage stability target needs to be strengthened. This function achieves coordinated optimization of multiple control objectives by minimizing the sum of squares of the weighted errors between the actual value and the reference value, balancing the complex performance requirements of the renewable energy grid connection process. Furthermore, the swarm intelligence algorithm (such as particle swarm optimization algorithm, genetic algorithm, etc.) simulates the collaborative behavior of biological groups to optimize the weight coefficient vector , and perform iterative updates. Taking the particle swarm optimization algorithm (PSO) as an example, each particle represents a set of weight coefficient combinations, and searches in the solution space through speed and position updates; the genetic algorithm (GA) evolves the weight coefficient population by simulating the selection, crossover and mutation operations in the biological evolution process; In each iteration, the algorithm is based on the preset fitness function (that is, the multi-objective optimization function The algorithm evaluates the performance of each individual (weight coefficient combination). Lower fitness values ​​indicate closer alignment between the actual control target and the reference value for that weight combination, leading to better control performance. The algorithm gradually searches for the global optimal solution through information sharing and evolutionary operations (such as crossover, mutation, and position updates).

[0026] Specifically, step 200 includes: Step 210, initializing the position and velocity vectors of the agent group; Step 220, iteratively updating the agent's position to converge to the global optimal solution; Step 230: The iteration termination condition is that the optimization function converges or reaches the maximum number of iterations.

[0027] As can be understood, in step 210, the initial positions and velocity vectors of the agent group are first randomly generated in the search space. The position of each agent represents a set of parameter combinations for the flexible control of renewable energy grid connection, covering key control parameters such as voltage regulation coefficient, frequency compensation factor, power allocation weight, etc. The velocity vector determines its movement direction and step size in the search space. Dynamic adjustment is used to achieve efficient exploration of the parameter space. To ensure efficient exploration of the algorithm, the number of agents, position range, and velocity boundary need to be reasonably set. The number of agents is set according to the complexity of the problem and the computing resources, usually between 20 and 100. Too few agents will easily fall into local optimality, while too many agents will increase the computing burden. The location range is based on the physical constraints and operating specifications of the new energy grid-connected system, setting a reasonable value range for each parameter. For example, the voltage regulation coefficient is controlled between 0.95 and 1.05 per unit. The speed boundary is limited through experiments and empirical formulas. The speed range is limited to prevent the agent from "flying out" of the effective search area. For example, the speed limit is set to 10% of the position range. Reasonable initial value setting can significantly reduce the risk of the algorithm falling into local optimality.

[0028] Furthermore, step 220 continuously iterates and updates the agent's position, allowing the group to gradually converge to the global optimal solution. In each iteration, the agent adjusts its speed and position according to a specific update formula based on its own historical optimal position and the global optimal position shared by the group. Taking the particle swarm optimization algorithm (PSO) as an example, its update formula is as follows: ; Where, For particles In the The speed of the particle's movement is directly determined by its value and direction. is the position coordinate of the particle in this dimension, As an inertia weight, it balances the particle's inheritance of historical velocity and exploration of new search directions. Helps with global search, smaller It tends to be more local and fine-tuned. 、 is the learning factor, which controls the degree to which the particle learns from its own historical optimal position and the global optimal position. It can be understood as the weight coefficient of individual experience and group wisdom. 、 For The random numbers uniformly distributed within the interval introduce randomness into the algorithm to avoid falling into the local optimal solution. For particles The historical optimal position reflects the best results of individual exploration. It is the global optimal position currently found by the entire particle swarm, and it embodies the wisdom of group collaboration.

[0029] Understandably, this process essentially simulates the collaborative behavior of biological groups in nature. For example, when a flock of birds is foraging, each bird will dynamically adjust its flight direction and speed based on its own past experience of finding food (historical optimal position) and the position of its companions who have found the most food in the entire flock (global optimal position). When migrating, schools of fish will also efficiently find a suitable living environment through information exchange between individuals and the group. In the flexible control scenario of renewable energy grid connection, this mechanism can optimize the distribution of renewable energy output power in real time based on the real-time load fluctuations of the power grid and the intermittent and random characteristics of renewable energy power generation. For example, when the power generation of renewable energy such as wind power and photovoltaic power fluctuates due to weather changes, the control strategy based on the particle swarm optimization algorithm can quickly adjust the output ratio of each renewable energy power station to ensure the stable operation of the power grid, improve the efficiency of renewable energy absorption, and achieve dynamic balance and optimized control of the power system.

[0030] Furthermore, step 230 is the termination condition determination, where the iteration termination condition is set to the convergence of the optimization function or the reaching of the maximum number of iterations. In specific implementation, a dual determination mechanism needs to be set, including: Optimization function convergence judgment: When the change range of the optimization function is less than the preset threshold, After iterations, the objective function value fluctuation is less than The algorithm can be considered to have converged to a satisfactory solution. For example, when optimizing the power loss target of renewable energy grid connection, if the loss change rate is less than 0.01% for 10 consecutive iterations, the calculation is stopped. Maximum iteration limit: If the number of iterations reaches a pre-set upper limit (e.g., 500) and convergence has not yet occurred, the algorithm is forced to terminate and the current optimal solution is output as the optimization parameter for flexible control of renewable energy grid connection. This mechanism effectively balances computational efficiency and solution quality, avoiding resource waste due to excessive computation while ensuring that feasible solutions can still be output for complex problems.

[0031] Optionally, step 200 further includes: Step 240 , converting the control problem into a finite-time optimization problem, and calculating the optimal control sequence through the prediction function; The control problem is transformed into a finite-horizon optimization problem. By establishing a mathematical model that includes the system state equation, control input constraints, and objective function, combined with a rolling-horizon optimization strategy, a prediction function is used to estimate the system state at multiple future sampling moments. On this basis, an optimization algorithm (such as a quadratic programming algorithm) is used to solve the optimal control sequence to achieve precise regulation of the dynamic performance of the new energy grid-connected system. Step 250 , adopting a fixed switching frequency control strategy to reduce switching losses of power electronic equipment; A fixed switching frequency control strategy is used to precisely set the switching cycles of power electronic devices, allowing them to operate at a constant frequency. This approach effectively avoids additional losses caused by switching frequency fluctuations, reduces electromagnetic interference, improves system stability and reliability, and significantly reduces switching losses in power electronic devices. Step 260, maintaining the voltage regulation accuracy within a preset range; A voltage regulation mechanism based on feedback compensation is constructed to collect system voltage data in real time and compare it with the preset reference value. When voltage deviation is detected, the control parameters are adaptively adjusted to adjust the power output to ensure that the voltage regulation accuracy is always maintained within the preset range, thereby ensuring the power quality and stable interaction between the new energy grid-connected system and the power grid.

[0032] Step 300: A multi-dimensional feature vector is constructed by integrating the frequency deviation, voltage deviation, and harmonic impedance change rate. An islanding risk index is calculated based on robust control theory. When the index exceeds a threshold, the energy storage system is activated to maintain local load stability.

[0033] It is understandable that step 300 integrates the frequency deviation , voltage deviation and harmonic impedance change rate A multidimensional feature vector is constructed to comprehensively describe the grid's operating status. Based on robust control theory, this multidimensional feature vector is thoroughly analyzed to calculate a quantitative indicator that accurately reflects islanding risk. Once this indicator exceeds a pre-set threshold, the energy storage system is immediately activated. By regulating the storage system's charge and discharge, the system maintains stable operation of the local load, preventing safety incidents and equipment damage caused by islanding.

[0034] Specifically, in step 300, the frequency deviation is first fused , voltage deviation and harmonic impedance change rate The three-dimensional characteristic matrix of , where the frequency deviation By collecting the grid frequency in real time With rated frequency (50Hz) is calculated, that is, , voltage deviation Based on the three-phase voltage RMS With rated voltage The relative error is determined by , harmonic impedance change rate Fast Fourier transform (FFT) is used to extract each harmonic component, and the dynamic change rate is calculated based on the harmonic impedance values ​​of the two previous and next sampling periods; Afterwards, the islanding risk index is calculated using the μ integrated controller , the controller is based on Robust control theory, by constructing a weighted function matrix for the three-dimensional characteristic matrix Comprehensive processing is performed to optimize detection performance while taking system uncertainty into account. (preset threshold) is determined to be an islanding state. After extensive simulation and field testing, the detection method has an accuracy rate of ≥ 99.8% and a false positive rate of ≤ 0.05%. It can effectively identify islanding phenomena under complex working conditions in new energy grid-connected systems. Furthermore, after detecting an island, the fast response mechanism of the high-speed digital signal processor (DSP) triggers the solid-state relay disconnection logic within 0.1s, and simultaneously activates the supercapacitor-lithium battery hybrid energy storage system. This system adopts a modular design. The supercapacitor module has high power density characteristics and can prioritize instantaneous power support within milliseconds to meet the dynamic response requirements of critical loads and maintain autonomous control for 2 hours. The lithium battery module, with its high energy density, serves as a stable energy backup to ensure long-term system operation. Finally, through The integrated controller regulates the energy storage output power, constructing a dual closed-loop control strategy consisting of a voltage outer loop and a power inner loop. The voltage outer loop monitors the local load voltage in real time and dynamically adjusts the power command based on the deviation. The power inner loop achieves rapid response of the energy storage system through current tracking control. Actual testing has shown that this control strategy can maintain local load voltage fluctuations ≤±3% and a stable frequency at 50Hz±0.5Hz, effectively ensuring power supply quality and system stability during island operation.

[0035] In step 400 , the main link uses power line carrier communication to transmit control instructions, and the backup link uses wireless wide area communication to ensure real-time instruction transmission when the main link is interrupted.

[0036] As can be understood, step 400 establishes a primary-backup dual-link communication mechanism. The primary link adopts power line carrier communication technology, using power lines as the transmission medium to achieve efficient transmission of control commands, with low cost and high reliability. The backup link adopts wireless wide-area communication methods such as 4G / 5G and NB-IoT. When the primary link is interrupted due to line failure, electromagnetic interference, etc., the backup link can quickly take over, ensuring the real-time and continuity of control command transmission and safeguarding the stable operation of the new energy grid-connected system.

[0037] Specifically, step 400 includes: Step 410, the main link uses power line carrier communication technology to transmit control instructions; The main link uses power line carrier (PLC) communication technology as the main transmission link. This technology uses power lines as the signal transmission medium, modulating control commands and loading them onto the high-frequency carrier of the power lines for transmission. PLC has the inherent advantage of power network coverage and can efficiently transmit data within the power system. It is suitable for control command transmission scenarios with high real-time requirements, ensuring the rapid and stable delivery of commands under normal operating conditions. Step 420 : The backup link uses low-power wide-area IoT communication technology to ensure that the command transmission delay meets the real-time control requirements when the primary link is interrupted. The backup link deploys low-power wide-area IoT (LPWAN) communication technology as a backup link. Technologies such as NB-IoT or LoRa are selected, offering low power consumption, wide coverage, and strong penetration. When the primary link is interrupted, the backup link automatically takes over data transmission tasks. By optimizing the protocol stack design and data compression algorithm, the control command transmission delay is kept within 200ms, meeting the stringent real-time control requirements of renewable energy grid connection and preventing system loss of control due to communication interruptions. Step 430: When the primary link loses a preset number of heartbeat packets continuously, it automatically switches to the backup link and synchronizes data to the remote monitoring center; The system continuously monitors the status of the primary link. If the primary link loses three consecutive heartbeat packets (adjustable based on actual needs), intelligent switching logic is triggered, automatically and seamlessly switching to the backup link. Simultaneously, bidirectional data synchronization technology rapidly synchronizes key operating parameters, control command queues, and other data at the time of switchover to a remote monitoring center, ensuring data consistency between the cloud and local systems. This allows operations personnel to promptly monitor system status and facilitate remote diagnosis and emergency response.

[0038] In this embodiment, through multimodal data acquisition and deep learning networks, combined with deep learning and state estimation, the device can independently and flexibly adjust based on the grid-connected point's own parameters, without relying on large-scale remote data acquisition, greatly simplifying the control strategy; by constructing a multi-objective optimization function, the particle swarm optimization algorithm is used to dynamically adjust the weight coefficients of the multi-objective model predictive control, effectively improving the multi-objective collaborative adjustment capability, so that the operating status of each grid-connected point in the grid-connected system is within a reasonable range, thereby ensuring the stability of the overall system; by constructing a multi-dimensional feature vector to detect the island state, fast fault-tolerant control is achieved and the reliability of island detection is enhanced; through dual-mode communication redundancy design, real-time transmission of control instructions is guaranteed when communication is interrupted, and the reliability and flexibility of new energy grid-connected control are comprehensively improved. In addition to being suitable for new energy grid-connected scenarios, it is also suitable for the control of charging pile charging and discharging systems and similar scenarios.

[0039] Based on the above embodiments, Figure 2 As shown, the present invention further provides a flexible automatic control device for new energy grid connection, which is used to support the flexible control method for new energy grid connection of the above embodiment. The flexible automatic control device for new energy grid connection includes: The multimodal data acquisition module 11 is used to obtain the electrical parameters of the grid connection point through the self-measurement mode or the communication interface mode, extract and compensate the harmonic characteristics based on the deep learning network, and estimate the grid parameters by fusing multi-source data using the state estimation method; A hybrid intelligent control module 12 is used to construct a multi-objective optimization function including voltage quality, power factor, and harmonic suppression, dynamically adjust the weight coefficients of the optimization function using a swarm intelligence algorithm, and optimize the inverter switching sequence through multi-objective model predictive control; The islanding detection and fault-tolerance control module 13 is used to integrate frequency deviation, voltage deviation, and harmonic impedance change rate to construct a multi-dimensional feature vector, calculate the islanding risk index based on robust control theory, and activate the energy storage system to maintain local load stability when the index exceeds a threshold; The dual-mode communication module 14 is used to transmit control instructions using power line carrier communication in the main link and wireless wide area communication in the backup link to ensure real-time instruction transmission when the main link is interrupted.

[0040] Optionally, the flexible automatic control device retains the original functions of the protocol converter and the grid-connected switch, and when there is remote control, the remote control instruction execution can still be given priority.

[0041] Understandably, the flexible automatic control device ensures that control instructions can still be transmitted in real time when communication is interrupted, while retaining the remote control instruction priority mechanism. On the basis of improving local control efficiency and real-time performance, it is compatible with the existing remote management system and comprehensively enhances the reliability and flexibility of new energy grid-connected control.

[0042] In this embodiment, through multimodal data acquisition and deep learning networks, combined with deep learning and state estimation, the device can independently and flexibly adjust based on the grid-connected point's own parameters, without relying on large-scale remote data acquisition, greatly simplifying the control strategy; by constructing a multi-objective optimization function, the particle swarm optimization algorithm is used to dynamically adjust the weight coefficients of the multi-objective model predictive control, effectively improving the multi-objective collaborative adjustment capability, so that the operating status of each grid-connected point in the grid-connected system is within a reasonable range, thereby ensuring the stability of the overall system; by constructing a multi-dimensional feature vector to detect the island state, fast fault-tolerant control is achieved and the reliability of island detection is enhanced; through dual-mode communication redundancy design, real-time transmission of control instructions is guaranteed when communication is interrupted, and the reliability and flexibility of new energy grid-connected control are comprehensively improved. In addition to being suitable for new energy grid-connected scenarios, it is also suitable for the control of charging pile charging and discharging systems and similar scenarios.

[0043] Furthermore, the flexible automatic control device for new energy grid connection can run the flexible control method for new energy grid connection described above. For specific implementation, please refer to the method embodiment, which will not be described in detail here.

[0044] Based on the above embodiments, Figure 3 As shown, the present invention also provides implementation steps of a flexible automatic control device for new energy grid connection, and the implementation steps are applied to the "full grid access" scenario. The specific scenario is: a distributed photovoltaic system with an installed capacity of 2MW, adopting a "full grid access" mode, the user load is mainly inductive loads such as motors, and the reactive power demand is significant. The flexible automatic control device is directly connected to the low-voltage bus on the user side.

[0045] The implementation steps include: S11, user load status perception: Real-time acquisition of user-side voltage through self-measurement mode (10kHz high-precision analog-to-digital converter ADC) , current , calculate the user load reactive power ; The communication interface mode interacts with the user's electricity meter (RS485 interface) to obtain real-time load power curves and identify load peak periods (14:00-17:00); S12, inverter reactive output optimization: The multi-objective model predictive control (MPC) objective function focuses on power factor optimization: ,After dynamic adjustment through the particle swarm optimization algorithm, the ,user load reactive loss is compensated first; Control inverter output reactive current Eliminate 50% of the reactive power demand on the user side, and the rest is provided by the grid to ensure the power factor on the main grid side ; S13, harmonic suppression and voltage stability: The harmonic prediction model detects that the initial value of the 5th harmonic distortion rate is 3.5%, and then outputs the compensation value. ,After injecting the reverse harmonic current, the harmonic distortion rate is effectively reduced to 1.2%; Extended Kalman filter (EKF) algorithm to estimate grid impedance , dynamically adjust the voltage reference value , maintain the voltage fluctuation of the grid connection point ≤±2%.

[0046] In this embodiment, the "full grid access" flexible automatic control device does not interact with the "grid-connected energy meter" but instead interacts with the "user energy meter" to read the operating status of the "user load." For example, it controls the reactive output of the "grid-connected inverter" to maximize the reactive power loss of the "user load," keeping the power factor of the main grid within a reasonable range.

[0047] Based on the above embodiments, Figure 4 As shown, the present invention also provides implementation steps of another flexible automatic control device for new energy grid connection, which is applied to the "self-generation and self-use + surplus power to the grid" scenario. The specific scenario is: a building photovoltaic system with an installed capacity of 500kW, adopting the "self-generation and self-use + surplus power to the grid" mode, equipped with a 100kWh lithium battery energy storage system, and the flexible automatic control device is connected to a 380V low-voltage bus.

[0048] The implementation steps include: S21, real-time monitoring of grid-connected status: Communication interface mode obtains real-time data from grid-connected energy meters (DL / T645 protocol): active power (PV output 400kW, load consumption 100kW), reactive power ; Self-measurement mode supplements the detection of grid connection point voltage ,frequency , construct a three-dimensional feature matrix ; S22, multi-objective coordinated control: Particle swarm optimization algorithm dynamically allocates weights , balanced voltage stability (±1%) and harmonic suppression (THD≤2%); Inverter output active power (Local consumption 100kW, surplus 250kW connected to the grid), while outputting capacitive reactive power , offset the inductive reactive power on the grid side and increase the power factor to 0.99; S23, island protection and energy storage coordination: Integrated controller calculates islanding risk index (Threshold 0.8), although no islanding is triggered, the lithium battery pre-charging is started in advance (SOC=30%); When the photovoltaic system is shut down at night, the energy storage system outputs power through dual closed-loop control (voltage outer loop + power inner loop) , maintain the load voltage , to prevent power grid fluctuations from affecting user equipment.

[0049] In this embodiment, the flexible automatic control device of "self-generation and self-use of surplus electricity and grid connection" has the function of communicating with the "grid-connected electricity meter", can read the operating parameters of the "grid-connected electricity meter" to obtain the electrical parameters of the grid connection point, and adjust the output of the "grid-connected inverter" according to the above electrical parameters, so that the grid connection point operates in a suitable state.

[0050] In addition, it should be noted that the combination of the various technical features in this case is not limited to the combination described in the claims of this case or the combination described in the specific embodiments. All technical features recorded in this case can be freely combined or combined in any way unless there is a contradiction between them.

[0051] It should be noted that the above examples are merely specific embodiments of the present invention. Obviously, the present invention is not limited to the above examples, and many similar variations are possible. All variations directly derived from or associating with the present invention by those skilled in the art are intended to fall within the scope of protection of the present invention.

[0052] The above are only preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A flexible control method for new energy grid connection, characterized in that: The following steps are involved: Obtain grid connection point electrical parameters through self-measurement mode or communication interface mode, extract and compensate harmonic features based on deep learning networks, and estimate grid parameters using state estimation methods to fuse multi-source data; Construct a multi-objective optimization function that includes voltage quality, power factor, and harmonic suppression. Use a swarm intelligence algorithm to dynamically adjust the weight coefficients of the optimization function. Optimize the inverter switching sequence through multi-objective model predictive control. A multidimensional feature vector is constructed by integrating frequency deviation, voltage deviation, and harmonic impedance change rate. The islanding risk index is calculated based on robust control theory. When the index exceeds the threshold, the energy storage system is activated to maintain local load stability. Among them, the main link uses power line carrier communication to transmit control instructions, and the backup link uses wireless wide area communication to ensure the real-time transmission of instructions when the main link is interrupted.

2. The flexible control method for new energy grid connection according to claim 1, characterized in that: The method of obtaining the grid connection point electrical parameters through the self-measurement mode or the communication interface mode, extracting and compensating harmonic features based on the deep learning network, and estimating the grid parameters by fusing multi-source data using the state estimation method includes: Self-measurement mode directly measures the voltage, current and harmonic parameters of the grid connection point through the built-in analog-to-digital conversion unit; The communication interface mode obtains the power data of the grid-connected energy meter through a wired or wireless communication interface; The deep learning network outputs harmonic compensation based on the trained prediction model to achieve active suppression of total harmonic distortion.

3. The flexible control method for new energy grid connection according to claim 2, characterized in that: The deep learning network outputs harmonic compensation based on the trained prediction model, including: The input layer receives the time domain or frequency domain characteristics of the grid connection point electrical parameters; The middle layer contains multiple nonlinear transformation layers to extract harmonic features; The output layer outputs harmonic compensation instructions through regression analysis.

4. The flexible control method for new energy grid connection according to claim 1, characterized in that: The method of dynamically adjusting the weight coefficient of the optimization function using a swarm intelligence algorithm includes: Initialize the position and velocity vectors of the agent group; Iteratively update the agent's position to converge to the global optimal solution; The iteration termination condition is that the optimization function converges or the maximum number of iterations is reached.

5. The flexible control method for new energy grid connection according to claim 1, characterized in that: The inverter switching sequence optimization is realized by multi-objective model predictive control, including: The control problem is transformed into a finite-time optimization problem, and the optimal control sequence is calculated through the prediction function; Adopt fixed switching frequency control strategy to reduce switching losses of power electronic equipment; Maintain voltage regulation accuracy within the preset range.

6. The flexible control method for new energy grid connection according to claim 1, characterized in that: The multi-objective optimization function is: ； in, For the The actual value of the control target, is the reference value, is the weight matrix, is the dynamically adjusted weight coefficient, satisfying ; The swarm intelligence algorithm optimizes the weight coefficient vector by iterative updating .

7. The flexible control method for new energy grid connection according to claim 1, characterized in that: The method integrates frequency deviation, voltage deviation, and harmonic impedance change rate to construct a multi-dimensional feature vector, calculates an islanding risk index based on robust control theory, and activates the energy storage system to maintain local load stability when the index exceeds a threshold, including: Constructing multidimensional feature vectors ,in, is the frequency deviation, is the voltage deviation, is the harmonic impedance change rate; Then, the island risk index is calculated based on robust control theory ,when When the preset threshold is exceeded, it is determined to be an island state; The energy storage system is then started, and the local load voltage and frequency are maintained within the allowable range through the power regulation unit.

8. The flexible control method for new energy grid connection according to claim 1, characterized in that: The main link uses power line carrier communication to transmit control instructions, and the backup link uses wireless wide area communication to ensure the real-time transmission of instructions when the main link is interrupted, including: The main link uses power line carrier communication technology to transmit control instructions; The backup link uses low-power wide-area IoT communication technology to ensure that the command transmission delay meets real-time control requirements when the main link is interrupted; When the primary link loses a preset number of heartbeat packets continuously, it automatically switches to the backup link and synchronizes data to the remote monitoring center.

9. A flexible automatic control device for new energy grid connection, characterized in that: include: Multimodal data acquisition module, used to obtain grid connection point electrical parameters through self-measurement mode or communication interface mode, extract and compensate harmonic characteristics based on deep learning network, and estimate grid parameters by fusing multi-source data using state estimation method; A hybrid intelligent control module is used to construct a multi-objective optimization function that includes voltage quality, power factor, and harmonic suppression. It uses a swarm intelligence algorithm to dynamically adjust the weight coefficients of the optimization function and optimizes the inverter switching sequence through multi-objective model predictive control. The islanding detection and fault-tolerant control module is used to integrate frequency deviation, voltage deviation, and harmonic impedance change rate to construct a multidimensional feature vector. Based on robust control theory, it calculates the islanding risk index and activates the energy storage system to maintain local load stability when the index exceeds the threshold. The dual-mode communication module is used to transmit control instructions using power line carrier communication on the main link and wireless wide area communication on the backup link to ensure real-time instruction transmission when the main link is interrupted.

Images