An inverter power supply network configuration type control method based on artificial neural network

Abstract

The application discloses an inverter power supply network construction type control method based on an artificial neural network and relates to the technical field of inverter power supply network construction type control. An electrical parameter sequence set of an inverter power supply grid-connected point is collected; a power supply network construction control analyzer is constructed based on a deep feedforward neural network, and a first network construction control parameter is output; a power distribution network simulation topology model is built based on PSCAD to perform parameter optimization, and a second network construction control parameter is generated; an operation state fluctuation degree is analyzed and determined, a dynamic fitting strategy is set according to the operation state fluctuation degree, two types of control parameters are weighted and fused, an adaptive network construction control strategy is obtained, and regulation and control are implemented. Through the double-path parallel mechanism of neural network fast response and simulation model accurate verification, combined with the adaptive fitting strategy of operation state sensing, the online intelligent optimization of the control parameter is realized, and the stability, adaptability and control precision of the inverter power supply under complex working conditions are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the network configuration type control technology field of the inverter power supply, and in particular relates to an inverter power supply network configuration type control method based on an artificial neural network. BACKGROUND

[0002] With the increasing proportion of renewable energy grid connection, the proportion of inverter power supply in the power system is increasing. The traditional follow-network control strategy is prone to stability problems in the weak grid or high proportion of new energy access scenarios. The network configuration type control technology can enhance the inertia and damping support capacity of the grid by simulating the operating characteristics of synchronous generators.

[0003] At present, the network configuration type control method is mostly based on fixed control structure or classical proportional integral derivative control, and the parameters thereof usually depend on artificial setting or offline optimization, and it is difficult to adapt to the dynamic changes of the grid operating state. In the complex working conditions of time-varying grid parameters, load fluctuations or fault disturbances, the adaptive ability and robustness of the traditional control method are limited, which can easily lead to control instability or performance degradation. At the same time, the existing method is sensitive to operating state fluctuations, and over-regulation or oscillation phenomenon is prone to occur when the electrical parameters change. SUMMARY

[0004] The present application provides an inverter power supply network configuration type control method based on an artificial neural network, which solves the technical problems of the prior art that the network configuration type control parameters are fixed, the adaptive ability is insufficient, and the stability is limited in complex working conditions.

[0005] The technical solution of the present application to solve the above technical problems is as follows:

[0006] The present application provides an inverter power supply network configuration type control method based on an artificial neural network, which includes:

[0007] Collecting a set of electrical parameter sequences of the inverter power supply grid connection point in a historical time zone;

[0008] Building a power supply network control analyzer based on a deep feedforward neural network, performing control analysis in a preset time zone according to the set of electrical parameter sequences, and outputting first network configuration control parameters;

[0009] Building a power distribution network simulation topology model based on PSCAD, performing network configuration control parameter optimization according to the set of electrical parameter sequences, and generating second network configuration control parameters;

[0010] Determining the operating state fluctuation degree based on the set of electrical parameter sequences, setting a dynamic fitting strategy according to the operating state fluctuation degree to control fit the first network configuration control parameters and the second network configuration control parameters, obtaining an adaptive network configuration control strategy, and regulating and controlling the inverter power supply in the preset time zone.

[0011] The beneficial effects of the present application are:

[0012] Compared with the prior art, the present application first realizes the effective fusion of data-driven and physical model through the double-path parallel optimization mechanism of deep feedforward neural network and PSCAD simulation model, significantly improving the accuracy and reliability of the control strategy. Secondly, the running state fluctuation degree evaluation and dynamic fitting strategy are introduced, so that the control parameters can be adaptively adjusted according to the real-time running state of the power grid, enhancing the adaptability under complex working conditions. Thirdly, the intelligent parameter optimization and fitting method is adopted, replacing the traditional control parameter setting method which relies on artificial experience, greatly improving the generation efficiency and optimization accuracy of the control strategy. Finally, through the double-path parallel mechanism of neural network rapid response and simulation model accurate verification, the control framework close to the engineering practice is constructed, which has good engineering applicability and can provide effective support for the stable operation of the power grid under high proportion of new energy access. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 A flowchart of a grid-type control method for an inverter-type power supply based on an artificial neural network is provided.

[0014] Figure 2 A circuit control principle diagram of a grid-type control method for an inverter-type power supply based on an artificial neural network is provided. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0016] In the description of the present application, the terms "first", "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0017] In the description of the present application, the term "for example" is used to indicate "as an example, instance, or illustration." Any embodiment described as "for example" in the present application is not necessarily to be construed as preferred or advantageous over other embodiments. The following description is presented to enable any person skilled in the art to make and use the present application. In the following description, details are set forth for the purpose of explanation. It will be appreciated that one of ordinary skill in the art will realize that the application can be practiced without the use of these specific details. In other instances, well-known structures and processes have not been elaborated in detail in order not to obscure the description of the present application with unnecessary details. Therefore, the present application is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed.

[0018] Embodiment one, as shown in Figure 1 , Figure 2 The present application provides a control method for inverter power supply network based on artificial neural network, comprising:

[0019] S10: Collecting a set of electrical parameter sequences of the inverter power supply grid-connected point in a historical time zone;

[0020] Specifically, the electrical parameters of the inverter power supply grid-connected point at a plurality of continuous time nodes in the historical time zone are collected, and a plurality of electrical parameter sequences are obtained as the set of electrical parameter sequences, wherein the electrical parameters include three-phase voltage instantaneous value, three-phase current instantaneous value, d-axis voltage, q-axis voltage, d-axis current and q-axis current.

[0021] The inverter power supply grid-connected point is the physical interface point for power exchange and connection between the inverter power supply and the public grid. This interface point is the key position for monitoring and controlling the operation state of the inverter power supply, and its electrical parameters directly reflect the interaction characteristics between the inverter power supply and the grid.

[0022] Collecting the electrical parameters of the inverter power supply grid-connected point in the historical time zone means that a historical time period is set according to the control requirements, such as 3 minutes, and a plurality of time nodes are selected in a continuous and equal interval manner in the historical time zone. The selection frequency of the time nodes is set according to the required control accuracy and data volume, such as collecting 1 point per second or 1 point per 10 milliseconds. Then, a plurality of sets of electrical parameter instantaneous values of the inverter power supply grid-connected point are synchronously collected at each selected time node. The collected electrical parameters specifically include three-phase voltage instantaneous value, three-phase current instantaneous value, d-axis voltage, q-axis voltage, d-axis current and q-axis current. Together, they constitute a set of electrical parameter sequences reflecting the operation state of the inverter power supply grid-connected point, providing a data basis for subsequent analysis and control.

[0023] Further, the running state fluctuation degree is analyzed and determined based on the set of electrical parameter sequences, comprising:

[0024] obtaining a set of electrical parameter sequences, wherein the set of electrical parameter sequences comprises a three-phase voltage instantaneous value sequence, a three-phase current instantaneous value sequence, a d-axis voltage sequence, a q-axis voltage sequence, a d-axis current sequence and a q-axis current sequence;

[0025] performing parameter fluctuation calculation on the three-phase voltage instantaneous value sequence, the three-phase current instantaneous value sequence, the d-axis voltage sequence, the q-axis voltage sequence, the d-axis current sequence and the q-axis current sequence respectively to obtain a plurality of electrical parameter fluctuation coefficients, wherein the electrical parameter fluctuation coefficient is a ratio of the standard deviation to the mean value of the parameter in the electrical parameter sequence;

[0026] performing weighted summation on the plurality of electrical parameter fluctuation coefficients based on the weight set according to the correlation degree between the parameter and the control stability of the inverter power supply to obtain the operation state fluctuation degree, wherein the weight is positively correlated with the correlation degree.

[0027] Specifically, the operation state fluctuation degree is determined based on the set of electrical parameter sequences obtained by collection. The operation state fluctuation degree is used to represent the stability degree and the change severity of the electrical operation state of the inverter power supply grid-connected point in a specific time period. This index is obtained by calculating the statistical fluctuation characteristics of each key electrical parameter sequence and comprehensively considering the influence weight of different parameters on the stability. The fundamental purpose of determining the operation state fluctuation degree is to realize the quantitative perception and accurate evaluation of the power grid operation condition, so as to provide a scientific basis for the intelligent adjustment of the control strategy.

[0028] Firstly, a complete set of electrical parameter sequences is obtained. The set of electrical parameter sequences comprises a three-phase voltage instantaneous value sequence, a three-phase current instantaneous value sequence, a d-axis voltage sequence, a q-axis voltage sequence, a d-axis current sequence and a q-axis current sequence. Each sequence is composed of electrical parameter instantaneous values arranged in time sequence in a historical time zone.

[0029] Secondly, parameter fluctuation calculation is performed on each electrical parameter sequence respectively. The fluctuation coefficient of each independent electrical parameter sequence is calculated. The specific calculation method of the electrical parameter fluctuation coefficient is as follows: the standard deviation of all values of the electrical parameter sequence is taken, and then divided by the arithmetic mean value, that is, the ratio of the standard deviation to the mean value is obtained. This ratio as a dimensionless index can effectively eliminate the influence of the dimension difference of different electrical parameters and accurately reflect the relative fluctuation degree of each electrical parameter in the time dimension. Through this step, a plurality of electrical parameter fluctuation coefficients corresponding to the three-phase voltage, the three-phase current, the d-axis voltage, the q-axis voltage, the d-axis current and the q-axis current can be obtained respectively.

[0030] Finally, the operating state fluctuation degree is calculated by weighted summation. According to the correlation degree of each electrical parameter with the control stability of the inverter power supply, an appropriate weight is assigned to the fluctuation coefficient of each electrical parameter. The higher the correlation degree of the electrical parameter, the greater the weight assigned to its fluctuation coefficient. Specifically, each weight is pre-set according to the influence degree of the parameter on the control stability of the inverter power supply. For example, the d-axis voltage and the q-axis voltage, which have the most direct influence on voltage stability, can be assigned a higher weight, such as 0.25; the three-phase voltage and the three-phase current parameters can be assigned a medium weight, such as 0.15; and the d-axis current and the q-axis current can be assigned a relatively lower weight, such as 0.05. The sum of all weight values is 1 to ensure the normalization of the calculation result.

[0031] Further, the fluctuation coefficients of all electrical parameters are multiplied by their corresponding weights, and the product results are summed to finally obtain the operating state fluctuation degree, which comprehensively represents the stability degree of the operating state and provides an important basis for the dynamic adjustment of the subsequent control strategy.

[0032] S20: constructing a power grid control analyzer based on a deep feedforward neural network, performing control analysis in a preset time zone according to the set of electrical parameter sequences, and outputting a first grid control parameter;

[0033] Specifically, the power grid control analyzer is constructed based on a deep feedforward neural network, including:

[0034] Based on the historical operation log of the inverter power supply, a plurality of sample electrical parameter sequence sets and a plurality of sample grid control parameters are collected as a sample training data set, wherein the grid control parameters include a d-axis reference voltage, a q-axis reference voltage, and a reference phase;

[0035] The sample training data set is equally divided into P parts, and P times of first training sets are randomly selected with replacement from the P sample training data sets. P training sets are obtained by iterative selection P times, wherein P is an integer greater than 20;

[0036] The P training sets are used to supervise the training of the deep feedforward neural network respectively until convergence, and P power grid control analysis units are obtained, which are combined to obtain the power grid control analyzer.

[0037] The power network configuration control analyzer is constructed based on a deep feedforward neural network. The deep feedforward neural network is an artificial neural network model comprising an input layer, multiple hidden layers, and an output layer, wherein information flows from the input layer to the output layer in one direction, the layers are fully connected through a weight matrix, and a nonlinear activation function is used to realize complex function mapping. The power network configuration control analyzer is an intelligent analyzer trained by the deep feedforward neural network, which is used to automatically analyze and generate the first network configuration control parameters that adapt to the current operating state according to the real-time collected electrical parameter sequence set.

[0038] Firstly, the sample training dataset is constructed. Based on the historical operation log of the inverter-type power supply, a large number of historical operation data samples are collected, each of which contains a sample electrical parameter sequence set and its corresponding sample network configuration control parameters. The sample electrical parameter sequence set contains time series data of three-phase voltage and current, d / q-axis voltage and current, etc. The sample network configuration control parameters are the control instructions verified as excellent at the corresponding historical time, including d-axis reference voltage, q-axis reference voltage, and reference phase. The paired input and output data together constitute the sample training dataset for training the power network configuration control analyzer.

[0039] Secondly, multiple differentiated training sets are constructed. The complete sample training dataset is evenly divided into P subsets, where P is an integer greater than 20. Then, P times of random sampling with replacement are performed, each time a subset is randomly selected from the P subsets to form a training set, and finally P training sets with slight differences in data distribution are obtained. By introducing random differences at the data level, the model units obtained based on different training sets are prompted to have prediction diversity.

[0040] Specifically, each power network configuration control analysis unit mainly consists of an input layer, a feature abstraction layer, and a control parameter output layer. The input layer receives the normalized electrical parameter sequence vector, which contains the sequence data of three-phase voltage instantaneous value, three-phase current instantaneous value, d-axis voltage, q-axis voltage, d-axis current, and q-axis current in the historical time zone. The feature abstraction layer adopts a deep fully connected neural network structure containing multiple hidden layers, the number of hidden layer neurons is configured according to the input feature dimension, each layer of neural network uses a ReLU activation function to introduce nonlinear transformation capability, and a Dropout layer is embedded between network layers with a dropout rate of 0.3 to 0.5 to effectively suppress model overfitting and improve its generalization performance. The output layer uses a linear activation function to map the final abstract features to three continuous control parameter values, which are used as the d-axis reference voltage, q-axis reference voltage, and reference phase, respectively.

[0041] During the training process, the key hyperparameters include the learning rate set to 0.0005, the number of training rounds set to 200, and the batch size set to 32. The learning rate is set based on the balance between training stability and convergence accuracy, the number of training rounds is set to ensure that the model learns the control law in the data sufficiently, and the batch size is selected to balance training efficiency and gradient stability. Specifically, the training method of supervised learning is adopted, and the obtained P training sets are divided into training sets, validation sets, and test sets in the ratio of 7:2:1.

[0042] Further, P deep feedforward neural networks with the same structure are independently supervised trained, and the network weight parameters are iteratively optimized through the backpropagation algorithm and the Adam optimizer. The mean square error loss function is used to measure the deviation between the predicted control parameters and the actual optimal control parameters, and the validation set is used to monitor the training process. When the validation set loss function value does not decrease continuously for multiple rounds and the model prediction root mean square error reaches a predetermined threshold, such as less than 0.05, the training is terminated, and the converged power network control analysis unit is obtained. The power network control analysis unit can effectively capture the complex nonlinear relationship between the electrical operating state and the optimal control instruction, and realize accurate network control parameter prediction. Finally, the P trained power network control analysis units are combined to form a complete power network control analyzer. In actual application, the power network control analyzer can integrate the output results of multiple units to improve the accuracy and reliability of control.

[0043] Further, the control analysis in a preset time zone is performed according to the electrical parameter sequence set, and a first network control parameter is output, including:

[0044] The ratio of the operating state fluctuation degree to the preset standard state fluctuation degree is set as a unit selection compensation coefficient;

[0045] The product of the unit selection compensation coefficient and the initial unit selection number is rounded to obtain an adaptive unit selection number K, wherein the initial unit selection number is 5, K is greater than or equal to 2 and less than or equal to P;

[0046] K units are randomly selected from the P power network control analysis units of the power network control analyzer, control analysis in a preset time zone is performed according to the electrical parameter sequence set, and the first network control parameter is obtained by mean fitting of the K output results.

[0047] First, the unit selection compensation coefficient is calculated. The unit selection compensation coefficient is the ratio of the operating state fluctuation degree to the preset standard state fluctuation degree. The preset standard state fluctuation degree is a benchmark value obtained based on historical stable operation data, which is used to represent the fluctuation level of the power network control under ideal stable working conditions. By calculating the unit selection compensation coefficient, the deviation of the current operating state from the standard state can be quantified.

[0048] Secondly, the number K of the adaptive unit selection is determined. The calculated unit selection compensation coefficient is multiplied by the initial unit selection number, and the product result is rounded up to obtain the final K value. Specifically, the initial unit selection number is preset to 5. At the same time, the K value is constrained in the range of greater than or equal to 2 and less than or equal to the total number P of power network control analysis units, so that the number of units participating in the analysis can adapt to the stability of the power network control state: when the operating state fluctuation degree is large, the unit selection compensation coefficient increases, and the K value increases accordingly, so that the results of more analysis units are integrated to improve the robustness of the decision; when the operating state is stable, fewer units are used to improve the calculation efficiency.

[0049] Finally, integrated analysis is performed and the results are output. From the P power network control analysis units included in the power network control analyzer, K power network control analysis units are randomly selected. The K power network control analysis units are used to perform control analysis on the electrical parameter sequence set in the current preset time zone in parallel, and each power network control analysis unit independently outputs a set of network control parameters. Then, the control parameter results output by the K power network control analysis units are respectively averaged, that is, all d-axis reference voltage values are averaged, all q-axis reference voltage values are averaged, and all reference phase values are averaged, and the three average values obtained are used as the integrated and final first network control parameters.

[0050] In summary, through the mechanism of random selection and mean fitting, multiple differentiated power network control analysis units can be comprehensively utilized to balance the individual bias, thereby outputting more stable and reliable control instructions.

[0051] S30: building a power distribution network simulation topology model based on PSCAD, performing network control parameter optimization according to the electrical parameter sequence set, and generating second network control parameters;

[0052] Firstly, a power distribution network simulation topology model is built based on PSCAD, including:

[0053] In the PSCAD simulation environment, according to the topology structure of the physical power distribution network, an inverter type power element, a line impedance element, a load element and a switch element are selected and configured;

[0054] According to the actual electrical connection relationship of the physical power distribution network, the inverter type power element, the line impedance element, the load element and the switch element are electrically connected to form a closed-loop simulation network;

[0055] The inverter type power element in the simulation network is configured with a controllable network control interface, wherein the control interface allows external input of control parameters and drives the operation of the inverter type power element.

[0056] PSCAD is a professional electromagnetic transient simulation software widely used in power system simulation analysis. Its full name is Power Systems Computer Aided Design. The software can accurately simulate the electromagnetic characteristics of various elements in the power system under transient and steady state conditions, and provide a high-precision digital simulation environment for power analysis and control strategy verification. Based on PSCAD, a distribution network simulation topology model is built. The distribution network simulation topology model is a digital twin that can accurately reflect the electrical structure and dynamic characteristics of the actual distribution network. It not only contains the static topological connection relationship of the power grid, but also embeds the dynamic mathematical model of various elements, which is used to reproduce and predict the behavior of the distribution network under different operating conditions and disturbances in the virtual environment, providing a reliable simulation platform for offline testing and optimization of network control parameters. Specifically, the specific steps to build the distribution network simulation topology model are as follows:

[0057] First, select the elements and configure the parameters. In the PSCAD simulation software environment, according to the actual topological structure of the target physical distribution network, select and configure the corresponding elements from the software element library, including inverter-type power supply elements for simulating distributed power sources, line impedance elements for simulating transmission line characteristics, load elements for simulating power consumption loads, and switch elements for simulating grid topology changes or fault operations. After selection, according to the nameplate parameters or field measurement data of the corresponding equipment in the actual physical distribution network, the electrical parameters of each element are configured in detail, including setting the rated capacity and DC voltage of the inverter-type power supply element, setting the resistance and reactance values of the line impedance element, setting the active and reactive power of the load element, etc. Through detailed parameter configuration, the distribution network simulation topology model built can accurately reflect the static operating characteristics and dynamic response characteristics of the actual physical distribution network.

[0058] Second, electrical connection and network construction. Follow the actual electrical connection relationship of the physical distribution network, connect the configured elements, and the connection process should comply with the circuit connection rules described by Kirchhoff's law, such as connecting the output of the inverter-type power supply element to the load element through the line impedance element, and connecting the switch element at appropriate positions to simulate normal operation or fault isolation scenarios. Finally, a closed-loop simulation network that can realize power transmission and distribution and has complete topological structure is formed.

[0059] Finally, the control interface configuration needs to be performed. In order to realize the real-time regulation and control of the external control strategy on the inverter-type power supply in the power distribution network simulation topology model, a controllable network-forming type control interface needs to be configured for the inverter-type power supply element in the simulation network. The network-forming type control interface is used as a functional module in the power distribution network simulation topology model and has the ability to receive external input control parameters. After receiving the control parameters, the network-forming type control interface can directly drive the operation of the inverter-type power supply element, change the amplitude and phase of the output voltage, and thus simulate the real operation behavior of the inverter-type power supply under a specific control strategy.

[0060] Further, the network-forming control parameter optimization is performed according to the electrical parameter sequence set to generate a second network-forming control parameter, including:

[0061] The network-forming control parameter adjustment threshold of the inverter-type power supply is obtained, and a plurality of initial network-forming control parameters are randomly selected within the network-forming control parameter adjustment threshold;

[0062] A plurality of network-forming control schemes are respectively combined according to the electrical parameter sequence set and the plurality of initial network-forming control parameters;

[0063] In the power distribution network simulation topology model, network-forming control simulation is performed according to the plurality of network-forming control schemes to output a plurality of running stability index sets, wherein the running stability index includes a transient stability index, a voltage stability index, and a frequency stability index;

[0064] Based on the plurality of initial network-forming control parameters and the plurality of running stability index sets, network-forming control parameter optimization is performed to generate a second network-forming control parameter.

[0065] Specifically, the network-forming control parameter optimization according to the electrical parameter sequence set is a process of searching for a control parameter combination that can make the power distribution network simulation topology model achieve the best comprehensive stability based on an optimization algorithm. Finally, a second network-forming control parameter can be generated. The second network-forming control parameter is a set of control instructions obtained by testing and evaluating a large number of candidate parameter combinations in the power distribution network simulation topology model built in PSCAD and according to clear stability index optimization. Its characteristic is that it has been verified by the power distribution network simulation topology model, has high reliability and scene adaptability, and can be an important supplement and check for the first network-forming control parameter based on data driving.

[0066] Firstly, the grid-forming control parameter adjustment threshold of the inverter power supply is obtained, which defines the safe upper and lower limits of the adjustment of the three control parameters of the d-axis reference voltage, the q-axis reference voltage and the reference phase, and the setting is based on the inverter hardware tolerance, the grid operation regulation and the stability requirement, such as the d-axis reference voltage threshold can be set to 0.8 to 1.2 times of the rated value, the q-axis reference voltage threshold is set to-0.3 to +0.3 times of the rated voltage, and the reference phase threshold is set to-π / 6 to +π / 6 radians. Within the feasible region specified by the grid-forming control parameter adjustment threshold, a random sampling method is used to select a plurality of initial grid-forming control parameters, and the initial grid-forming control parameters are used as the starting point of the optimization algorithm to achieve extensive exploration of the parameter space.

[0067] Secondly, the real-time collected electrical parameter sequence set is combined with each initial grid-forming control parameter to form a plurality of complete grid-forming control schemes. Each grid-forming control scheme contains a specific set of control parameters and the corresponding real-time operation state of the power grid, which can be used as an independent test case and input into the distribution network simulation topology model to evaluate the effect of applying this set of control parameters under the specific operating state.

[0068] Further, the plurality of grid-forming control schemes are sequentially input into the pre-built distribution network simulation topology model for grid-forming control simulation. After each simulation run, a set of operation stability indicators is recorded and output. The set of operation stability indicators includes multiple dimensions of stability evaluation criteria, including transient stability indicators reflecting power and angle swing, voltage stability indicators representing voltage deviation and recovery capability, and frequency stability indicators measuring frequency deviation. Through simulation, the control effect of each control scheme in the simulation environment can be quantitatively evaluated.

[0069] Finally, based on the plurality of initial grid-forming control parameters and the corresponding set of operation stability indicators, the grid-forming control parameter optimization is performed. Specifically, the optimization process aims to improve the grid-forming control stability, and by analyzing the mapping relationship between different parameter combinations and stability indicators, the grid-forming control parameter combination that can achieve the optimal comprehensive stability of the grid-forming control is found from all candidate schemes, and is determined as the final second grid-forming control parameter.

[0070] Specifically, based on the plurality of initial grid-forming control parameters and the plurality of sets of operation stability indicators, the grid-forming control parameter optimization is performed to generate a second grid-forming control parameter, including:

[0071] Based on the plurality of sets of operation stability indicators, a plurality of system stability coefficients are evaluated and obtained;

[0072] set the initial network configuration control parameters as initial solutions, arrange the initial network configuration control parameters according to the system stability coefficients from large to small to generate an initial solution sequence;

[0073] select the first N solutions of the initial solution sequence as good solutions and the last M solutions as bad solutions, wherein the sum of M and N is the number of initial solutions, M is L times of N, and L is greater than or equal to 10;

[0074] randomly perform equivalent clustering on the M bad solutions with the N good solutions as the center to obtain N solution sets, adjust the bad solutions in the solution sets according to a preset optimization step length with the good solutions as the adjustment direction to obtain N updated solution sets;

[0075] identify the N updated solution sets, and if there is a bad solution whose system stability coefficient is greater than or equal to that of a good solution in the same updated solution set, replace the good solution with the bad solution;

[0076] continue the iterative optimization until a preset optimization convergence number is reached, output N current updated solution sets, and select a good solution with the largest system stability coefficient in the N current updated solution sets as the second network configuration control parameter.

[0077] wherein a ratio of the preset standard state fluctuation degree to the operating state fluctuation degree is set as a convergence number compensation coefficient, and a product of the convergence number compensation coefficient and a preset standard optimization convergence number is rounded to obtain the preset optimization convergence number.

[0078] First, based on a plurality of operating stability index sets, each group of operating stability indexes is comprehensively evaluated, and a system stability coefficient is calculated. System stability coefficient = α × transient stability index normalized value + β × voltage stability index normalized value + γ × frequency stability index normalized value, wherein α, β, and γ are weight coefficients, and α + β + γ = 1. Each weight is pre-set according to the relative importance of each stability type in network configuration control. For example, in a scenario where voltage support is the primary goal, the voltage stability weight β can be set to 0.5, the transient stability weight α to 0.3, and the frequency stability weight γ to 0.2. The normalized value of each index is obtained by the extreme value normalization method. Specifically, the original value of each stability index is mapped to the [0, 1] interval, and the calculation formula is: index normalized value = (original value - historical minimum value of the index) / (historical maximum value of the index - historical minimum value of the index). The system stability coefficient is used to quantitatively represent the overall stability of the distribution network simulation topology model under the corresponding network configuration control scheme. The higher the value, the better the stability.

[0079] Secondly, the initial network construction control parameters are regarded as the initial solution of the optimization problem. According to the calculated system stability coefficient, the initial network construction control parameters are arranged in descending order to form an initial solution sequence. Then, the top N solutions in the sequence are marked as superior solutions, and the last M solutions are marked as inferior solutions. The sum of the parameters M and N is equal to the total number of initial solutions, and the value of M is set to be L times of N, L is an integer greater than or equal to 10. This distribution method ensures that the number of inferior solutions is much larger than that of superior solutions, providing sufficient adjustment space for subsequent optimization operations.

[0080] Further, the M inferior solutions are distributed around the N superior solutions by random equivalent clustering method to form N solution sets. Within each solution set, the superior solution is used as the adjustment direction, and the parameter values of all inferior solutions in the solution set are updated according to a preset optimization step, thereby obtaining N updated solution sets. The optimization step control parameter is a key parameter for the adjustment amplitude of each iteration, which is set according to the range of the network construction control parameter adjustment threshold, such as 1% to 5% of the d-axis reference voltage adjustment threshold, 1% to 5% of the q-axis reference voltage adjustment threshold, and 1% to 5% of the reference phase adjustment threshold. This setting method can ensure that the parameter adjustment process has both sufficient precision for local search and fast convergence speed.

[0081] Further, the N updated solution sets are identified. Within the same solution set, if it is found that the system stability coefficient of a certain adjusted inferior solution is greater than or equal to the system stability coefficient of the current superior solution, the inferior solution is used to replace the original superior solution, ensuring that the position of the superior solution can move in a better direction.

[0082] Finally, the above clustering, adjusting and replacing steps are repeatedly performed until a preset optimization convergence number is reached. After the iteration process ends, N current updated solution sets are output, and the optimal solution with the largest system stability coefficient is selected from the N solution sets and determined as the final second network configuration control parameter. The ratio of the preset standard state fluctuation degree to the operating state fluctuation degree is set as a convergence number compensation coefficient, and the product of the convergence number compensation coefficient and the preset standard optimization convergence number is rounded to obtain the preset optimization convergence number. Specifically, the preset standard state fluctuation degree is a benchmark value obtained based on historical stable operation data and represents the fluctuation level of the network configuration control in an ideal stable working condition. The operating state fluctuation degree is a quantitative index reflecting the stability degree of the current network configuration control and is calculated through real-time electrical parameter sequences. The ratio of the two values is defined as the convergence number compensation coefficient. The convergence number compensation coefficient directly reflects the deviation degree of the current operating state from the standard state. When the operating state fluctuation degree is greater than the preset standard state fluctuation degree, it indicates that the network configuration control is in a large disturbance or unstable state, and the convergence number compensation coefficient is greater than 1. By multiplying the convergence number compensation coefficient by the preset standard optimization convergence number, such as 50, and rounding the product, the preset optimization convergence number suitable for the current working condition is obtained. When the operating state is stable, the iteration number can be automatically reduced to improve the calculation efficiency; when the operating state has a large fluctuation, the iteration number is automatically increased to provide more search time for the optimization algorithm, thereby ensuring that the second network configuration control parameter with high precision can be obtained under complex working conditions.

[0083] Therefore, the obtained second network configuration control parameter is a control instruction that can make the system stability reach a local or global optimum after multiple rounds of iteration optimization.

[0084] S40: determining an operating state fluctuation degree based on the electrical parameter sequence set analysis, setting a dynamic fitting strategy for control fitting of the first network configuration control parameter and the second network configuration control parameter according to the operating state fluctuation degree, obtaining an adaptive network configuration control strategy, and regulating the inverter-type power supply in the preset time zone.

[0085] Specifically, the operating state fluctuation degree is determined based on the electrical parameter sequence set analysis, as described in the calculation method in step S10.

[0086] Further, the dynamic fitting strategy is set for control fitting of the first network configuration control parameter and the second network configuration control parameter according to the operating state fluctuation degree, and an adaptive network configuration control strategy is obtained, including:

[0087] The prediction optimization accuracy is obtained according to the preset optimization convergence number, wherein the optimization accuracy is positively correlated with the optimization convergence number.

[0088] optimizing and adjusting the second initial weight based on the predicted optimization accuracy, to obtain a second adaptive weight, wherein the second adaptive weight is a confidence weight of the second network construction control parameter, is positively correlated with the predicted optimization accuracy, the second initial weight is 0.4, and the second adaptive weight is greater than or equal to 0.3 and less than or equal to 0.6;

[0089] obtaining a first adaptive weight by subtracting the second adaptive weight from 1, wherein the first adaptive weight is a confidence weight of the first network construction control parameter;

[0090] controlling and fitting the first network construction control parameter and the second network construction control parameter based on the first adaptive weight and the second adaptive weight, to obtain an adaptive network construction control strategy.

[0091] First, based on historical data, a convergence number-optimization accuracy comparison table is constructed for matching analysis according to a preset final actual use optimization convergence number, and a predicted optimization accuracy is obtained by matching. The more the number of simulation iterations, the more sufficient the search, and the more reliable the result is, so the predicted optimization accuracy is positively correlated with the preset optimization convergence number.

[0092] Second, based on the second initial weight 0.4, the predicted optimization accuracy obtained in the previous step is optimized and adjusted to obtain a final second adaptive weight. The specific adjustment method is as follows: the ratio of the predicted optimization accuracy to the preset standard optimization accuracy is calculated, the ratio is set as an adjustment coefficient, and the second adaptive weight is obtained by multiplying the adjustment coefficient by the second initial weight. The preset standard optimization accuracy is a benchmark value, which is set according to the corresponding relationship between the optimization convergence number and the obtained parameter accuracy in the historical simulation data, for example, the accuracy corresponding to the minimum convergence number required to achieve a satisfactory optimization effect is set to 0.8 by statistical analysis method. The second adaptive weight represents the confidence of the second network construction control parameter in the final decision, and is positively correlated with the predicted optimization accuracy and is constrained in the interval of 0.3 to 0.6. When the simulation optimization process is considered reliable, the weight of the second network construction control parameter can be as high as 60%; when the reliability is low, the weight can be as low as 30% at most.

[0093] For example, the standard optimization accuracy is set to 0.8, and the second initial weight is 0.4; if the current predicted optimization accuracy is 0.7, the adjustment coefficient is 0.7 / 0.8=0.875, and the second adaptive weight is 0.875x0.4=0.35. The second adaptive weight is constrained in the range of 0.3 to 0.6, which ensures that the confidence of the second network construction control parameter can be dynamically and reasonably floated according to the estimated reliability of the simulation optimization process.

[0094] Further, the first adaptation weight is obtained by subtracting the second adaptation weight from 1, representing the confidence weight of the first network configuration control parameter, ensuring that the sum of the weights of the two control parameters is always 1, constituting a complete decision basis.

[0095] Finally, the first adaptation weight and the second adaptation weight are used to weight and fuse the first network configuration control parameter and the second network configuration control parameter respectively, so as to obtain a final network configuration control strategy which integrates the advantages of data-driven rapid response and physical model verification and adapts to the current operating state. Specifically, the final network configuration control strategy = first adaptation weight x first network configuration control parameter + second adaptation weight x second network configuration control parameter.

[0096] In summary, the embodiments of the present application have at least the following technical effects:

[0097] Compared with the prior art, the present application first realizes the deep fusion of data-driven and physical model through the construction of a double-path parallel optimization mechanism of deep feedforward neural network and PSCAD simulation model, significantly improving the accuracy and reliability of the control strategy; secondly, by introducing the operating state fluctuation degree evaluation and dynamic fitting strategy, the control parameters can be adaptively adjusted according to the real-time operating state of the power grid, effectively enhancing the adaptability under complex working conditions; finally, the intelligent parameter optimization and fitting method is adopted to replace the traditional control parameter setting method which relies on manual experience, greatly improving the generation efficiency and optimization precision of the control strategy.

[0098] In summary, through the double-path parallel mechanism of neural network rapid response and simulation model accurate verification, combined with the adaptive fitting strategy of operating state perception, the online intelligent optimization of control parameters is realized, effectively improving the stability, adaptability and control precision of the inverter power supply under complex working conditions, and having good engineering applicability, providing reliable technical support for the stable operation of the power grid under the condition of high proportion of new energy access.

[0099] It should be noted that the above sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. Moreover, the above describes specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or can be advantageous.

[0100] The above only describes the preferred embodiments of the present application and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0101] The specification and drawings are, of course, subject to various interpretations and should not be viewed in any limiting sense. It will be understood that various modifications and changes can be made to the application disclosed without departing from the scope thereof. Accordingly, you are to understand that there is no intention, either express or implied, that any of the described embodiments of the application is more efficient or effective than any of the other possible embodiments. It is therefore intended to cover in the appended claims all such changes and modifications that fall within the scope of the application.

Claims

1. A method for networked control of an inverter-based power supply based on artificial neural networks, characterized in that the method The method comprises the following steps: Collecting a set of electrical parameter sequences of an inverter power grid connection point in a historical time zone; Building a power grid control analyzer based on a deep feedforward neural network, performing control analysis in a preset time zone according to the set of electrical parameter sequences, and outputting first grid control parameters; Building a power distribution network simulation topology model based on PSCAD, performing grid control parameter optimization according to the set of electrical parameter sequences, and generating second grid control parameters; Based on the set of electrical parameter sequences, the running state fluctuation degree is determined, and the dynamic fitting strategy is set according to the running state fluctuation degree to control and fit the first grid control parameters and the second grid control parameters, thereby obtaining an adaptive grid control strategy, and the inverter power in the preset time zone is regulated.

2. The control method of claim 1, wherein, Collecting electrical parameters of an inverter power grid connection point at a plurality of time nodes in a historical time zone, obtaining a plurality of electrical parameter sequences as an electrical parameter sequence set, wherein the electrical parameters include three-phase voltage instantaneous value, three-phase current instantaneous value, d-axis voltage, q-axis voltage, d-axis current and q-axis current.

3. The control method of claim 2, wherein, Based on the set of electrical parameter sequences, the running state fluctuation degree is determined, comprising: Obtaining an electrical parameter sequence set, wherein the electrical parameter sequence set includes a three-phase voltage instantaneous value sequence, a three-phase current instantaneous value sequence, a d-axis voltage sequence, a q-axis voltage sequence, a d-axis current sequence and a q-axis current sequence; Respectively calculating the parameter fluctuation of the three-phase voltage instantaneous value sequence, the three-phase current instantaneous value sequence, the d-axis voltage sequence, the q-axis voltage sequence, the d-axis current sequence and the q-axis current sequence, and obtaining a plurality of electrical parameter fluctuation coefficients, wherein the electrical parameter fluctuation coefficient is the ratio of the standard deviation to the mean value of the parameter in the electrical parameter sequence; Based on the correlation degree of the parameters and the control stability of the inverter power, the weights are set, the plurality of electrical parameter fluctuation coefficients are weighted and summed, and the running state fluctuation degree is obtained, wherein the weight is positively correlated with the correlation degree.

4. The control method of claim 1, wherein, Building a power grid control analyzer based on a deep feedforward neural network, comprising: Based on the historical operation log of the inverter power, a plurality of sample electrical parameter sequence sets and a plurality of sample grid control parameters are collected as sample training data sets, wherein the grid control parameters include d-axis reference voltage, q-axis reference voltage and reference phase; The sample training data set is equally divided into P parts, and P times of first training set are constructed by randomly selecting P times with replacement in the P sample training data, and P training sets are obtained by iterative selection, wherein P is an integer greater than 20; The P training sets are used to supervise the training of the deep feedforward neural network respectively until convergence, and P power grid control analyzer units are obtained, which are combined to obtain a power grid control analyzer.

5. The control method of claim 4, wherein, According to the set of electrical parameter sequences, control analysis is performed in a preset time zone, and first grid control parameters are output, comprising: The ratio of the running state fluctuation degree to the preset standard state fluctuation degree is set as a unit selection compensation coefficient; The product of the unit selection compensation coefficient and the initial unit selection number is rounded to obtain an adaptive unit selection number K, wherein the initial unit selection number is 5, K is greater than or equal to 2 and less than or equal to P; Randomly select K units in the P power grid control analysis units of the power grid control analyzer, perform control analysis in a preset time zone according to the electrical parameter sequence set, and obtain a first grid control parameter after mean fitting of K output results.

6. The control method of claim 1, wherein, A power distribution network simulation topology model is built based on PSCAD, including: In the PSCAD simulation environment, according to the topology structure of the physical power distribution network, an inverter type power supply element, a line impedance element, a load element and a switch element are selected and configured; According to the actual electrical connection relationship of the physical power distribution network, the inverter type power supply element, the line impedance element, the load element and the switch element are electrically connected to form a closed-loop simulation network; The inverter type power supply element in the simulation network is configured with a controllable grid type control interface, wherein the control interface allows external input of control parameters and drives the operation of the inverter type power supply element.

7. The control method of claim 6, wherein, According to the electrical parameter sequence set, the grid control parameter optimization is performed to generate a second grid control parameter, including: Obtain the grid control parameter adjustment threshold of the inverter type power supply, and randomly select a plurality of initial grid control parameters within the grid control parameter adjustment threshold; According to the electrical parameter sequence set and the plurality of initial grid control parameters, a plurality of grid control schemes are respectively combined to obtain a plurality of grid control schemes; In the power distribution network simulation topology model, according to the plurality of grid control schemes, a plurality of operation stability index sets are output, wherein the operation stability index includes transient stability index, voltage stability index and frequency stability index; Based on the plurality of initial grid control parameters and the plurality of operation stability index sets, the grid control parameter optimization is performed to generate a second grid control parameter.

8. The control method of claim 7, wherein, Based on the plurality of initial grid control parameters and the plurality of operation stability index sets, the grid control parameter optimization is performed to generate a second grid control parameter, including: Based on the plurality of operation stability index sets, a plurality of system stability coefficients are obtained; The initial grid control parameters are set as initial solutions, and the plurality of initial grid control parameters are arranged in descending order of system stability coefficients to generate an initial solution sequence; Select the first N solutions of the initial solution sequence as optimal solutions and the last M solutions as suboptimal solutions, wherein the sum of M and N is the number of initial solutions, M is L times of N, and L is greater than or equal to 10; Randomly perform equal value clustering on the M suboptimal solutions with the N optimal solutions as the center to obtain N solution sets, and adjust the suboptimal solutions in the solution sets in the adjustment direction of the optimal solutions according to the preset optimization step to obtain N updated solution sets; Identify the N updated solution sets, and if the system stability coefficient of a suboptimal solution is greater than or equal to that of an optimal solution in the same updated solution set, replace the optimal solution with the suboptimal solution; Continue the iterative optimization until the preset optimization convergence number is reached, output N current updated solution sets, and select the optimal solution with the largest system stability coefficient in the N current updated solution sets as the second grid control parameter.

9. The control method of claim 8, wherein, The ratio of the preset standard state fluctuation degree and the operation state fluctuation degree is set as a convergence number compensation coefficient, and the preset optimization convergence number is obtained by rounding the product of the convergence number compensation coefficient and a preset standard optimization convergence number.

10. The control method of claim 9, wherein, According to the operation state fluctuation degree, a dynamic fitting strategy is set to control and fit the first network control parameter and the second network control parameter, to obtain an adaptive network control strategy, including: According to the preset optimization convergence number, a predicted optimization accuracy rate is obtained, wherein the optimization accuracy rate and the optimization convergence number are positively correlated; Based on the predicted optimization accuracy rate, a second initial weight is optimized and adjusted to obtain a second adaptive weight, wherein the second adaptive weight is a confidence weight of the second network control parameter, is positively correlated with the predicted optimization accuracy rate, the second initial weight is 0.4, and the second adaptive weight is greater than or equal to 0.3 and less than or equal to 0.6; A first adaptive weight is obtained by subtracting the second adaptive weight from 1, wherein the first adaptive weight is a confidence weight of the first network control parameter; Based on the first adaptive weight and the second adaptive weight, the first network control parameter and the second network control parameter are controlled and fitted to obtain an adaptive network control strategy.

Images