AC / DC power distribution network protection method based on dynamic threshold value and related equipment

Abstract

The invention discloses an AC / DC power distribution network protection method based on a dynamic threshold value and related equipment. The method comprises the following steps: acquiring real-time operation data and historical operation data of an AC / DC power distribution network in real time; when detecting that the AC / DC power distribution network has a state change or receives a fault signal, performing probabilistic load flow calculation according to the real-time operation data, and determining fault current probability distribution information; a preset neural network is called to be combined with the historical operation data, and electrical parameter change trend information is predicted; constructing a dynamic threshold value optimization model by taking protection expectation loss minimization as a target and combining fault current probability distribution information and electrical parameter variation trend information; and iteratively solving the dynamic threshold optimization model to obtain a dynamic protection constant value of the AC / DC power distribution network, and issuing the dynamic protection constant value to a protection device in the AC / DC power distribution network. The method combines load flow calculation with change trend prediction of a neural network, quickly and accurately calculates a protection constant value, and effectively realizes fault isolation under different operation conditions.

Description

Technical Field

[0001] This invention relates to the field of power distribution network protection technology, and in particular to an AC / DC power distribution network protection method and related equipment based on dynamic threshold. Background Technology

[0002] As the global energy transition progresses, the proportion of renewable energy in the power system continues to increase, and hybrid AC / DC distribution networks are becoming increasingly widely used due to their ability to accommodate high proportions of renewable energy. However, while this new architecture improves system flexibility and energy absorption capacity, the intermittent, volatile, and random characteristics of renewable energy also lead to constantly changing grid operating conditions, making traditional protection methods based on fixed thresholds difficult to adapt.

[0003] Traditional AC distribution networks are radial, with unidirectional power flow and stable short-circuit currents. Protection settings can be set according to the most severe fault conditions for stable operation. However, in AC / DC distribution networks with a high proportion of renewable energy, the grid connection of distributed power sources changes the fault characteristics, and bidirectional power flow makes protection coordination more complex.

[0004] In the event of a fault, the fault current supplied by distributed generation sources can easily affect the operating characteristics of the original protection systems, potentially leading to maloperation or failure to operate. Furthermore, while interconnecting AC and DC distribution networks via VSCs (Voltage Source Converters) improves operational flexibility, it makes it difficult to quickly and accurately isolate faults under different operating conditions. Summary of the Invention

[0005] This invention provides a dynamic threshold-based AC / DC distribution network protection method and related equipment, which solves the technical problem that existing AC / DC distribution networks, when connected to a high proportion of renewable energy, are usually still protected based on fixed thresholds, making it difficult to quickly and accurately achieve fault isolation under different operating conditions.

[0006] The first aspect of this invention provides a protection method for AC / DC distribution networks based on dynamic thresholds, comprising:

[0007] Real-time acquisition of real-time and historical operating data of AC / DC distribution networks;

[0008] When a state change is detected in the AC / DC distribution network or a fault signal is received, probabilistic power flow calculation is performed according to the real-time operating data to determine the fault current probability distribution information.

[0009] The system invokes a preset neural network and combines it with the historical operating data to predict the trend of electrical parameter changes.

[0010] With the goal of minimizing expected losses, a dynamic threshold optimization model is constructed by combining the fault current probability distribution information and the electrical parameter change trend information.

[0011] The dynamic threshold optimization model is iteratively solved to obtain the dynamic protection settings of the AC / DC distribution network, and then distributed to the protection devices within the AC / DC distribution network.

[0012] Optionally, the step of determining the fault current probability distribution information by performing probabilistic power flow calculations based on the real-time operating data when a state change is detected in the AC / DC distribution network or a fault signal is received includes:

[0013] When a change in the state of the AC / DC distribution network is detected or a fault signal is received, the real-time operating data is preprocessed to obtain preprocessed data.

[0014] A Gaussian mixture model is used to perform probability fitting on the preprocessed data to obtain the probability distribution of random variables;

[0015] A probabilistic power flow model for the AC / DC distribution network is constructed using a linearized model.

[0016] The probability power flow model is solved based on the probability distribution of the random variables to determine the probability distribution information of the fault current.

[0017] Optionally, the neural network is a three-layer BP neural network; the step of calling a preset neural network in conjunction with the historical operating data to predict the trend information of electrical parameter changes includes:

[0018] The historical operating data is used to train a preset three-level BP neural network to obtain a target prediction model;

[0019] The target prediction model is used to predict the electrical parameter change trend information in the future acquisition cycle based on the real-time operating data.

[0020] Optionally, the three-layer BP neural network includes an input layer with 5 nodes, a hidden layer with 10 nodes, and an output layer with 1 node; the output layer is:

[0021]

[0022] in, It is the output of the j-th node in the hidden layer. The weights connecting the j-th node in the hidden layer to the nodes in the output layer are... This refers to the bias term for the output layer nodes. The activation function of the output layer. This provides information on the changing trends of electrical parameters.

[0023] Optionally, the step of iteratively solving the dynamic threshold optimization model to obtain the dynamic protection settings of the AC / DC distribution network and distributing them to the protection devices within the AC / DC distribution network includes:

[0024] The dynamic threshold optimization model is iteratively solved using the gradient descent method under probabilistic sensitivity constraints to obtain the dynamic protection settings of the AC / DC distribution network.

[0025] The dynamic protection settings are sent to the protection devices in the AC / DC distribution network to update the local settings of the protection devices online.

[0026] The dynamic protection setting is:

[0027]

[0028] in, The optimized dynamic protection settings, It is the inverse function of the cumulative distribution function. Is when The unit loss coefficient at that time Is when The unit loss coefficient at that time.

[0029] Optionally, the objective function of the dynamic threshold optimization model is:

[0030]

[0031] in, To protect expected losses, The dynamic protection settings to be optimized. It is a random current variable; The fault current probability distribution information is represented by the probability density function of the current. Is when The unit loss coefficient at that time Is when The unit loss coefficient at that time.

[0032] Optionally, after performing the step of calculating the probability distribution information of the fault current based on the real-time operating data when a state change is detected in the AC / DC distribution network or a fault signal is received, the method further includes:

[0033] Obtain existing protection settings;

[0034] Calculate the overcurrent probability based on the existing protection settings and the fault current probability distribution information;

[0035] If the overcurrent probability is greater than a preset probability threshold, then proceed to the step of calling a preset neural network to combine the historical operating data and predict the trend information of electrical parameter changes.

[0036] If the overcurrent probability is not greater than the preset probability threshold, then proceed to the step of collecting real-time operating data and historical operating data of the AC / DC distribution network.

[0037] A second aspect of the present invention provides an AC / DC distribution network protection device based on a dynamic threshold, comprising:

[0038] The data acquisition module is used to collect real-time and historical operating data of the AC / DC distribution network.

[0039] The power flow calculation module is used to perform probabilistic power flow calculations based on the real-time operating data when a state change is detected in the AC / DC distribution network or a fault signal is received, and to determine the fault current probability distribution information.

[0040] The parameter change trend prediction module is used to call a preset neural network and combine it with the historical operating data to predict the change trend information of electrical parameters.

[0041] The model building module is used to construct a dynamic threshold optimization model with the goal of minimizing the expected protection loss, by combining the fault current probability distribution information and the electrical parameter change trend information.

[0042] The setting value determination and distribution module is used to iteratively solve the dynamic threshold optimization model to obtain the dynamic protection setting value of the AC / DC distribution network, and distribute it to the protection devices in the AC / DC distribution network.

[0043] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to perform the steps of the AC / DC distribution network protection method based on dynamic threshold as described in any one of the first aspects of the present invention.

[0044] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the AC / DC distribution network protection method based on dynamic threshold as described in any of the first aspects of the present invention.

[0045] As can be seen from the above technical solutions, the present invention has the following advantages:

[0046] This invention collects real-time and historical operating data of AC / DC distribution networks. When a state change is detected in the AC / DC distribution network or a fault signal is received, probabilistic power flow calculations are performed based on the real-time operating data to determine the probability distribution of fault current. A pre-set neural network is invoked, combined with historical operating data, to predict the trend of electrical parameter changes. With the goal of minimizing expected protection losses, a dynamic threshold optimization model is constructed by combining the fault current probability distribution information and the trend information of electrical parameter changes. The dynamic threshold optimization model is iteratively solved to obtain the dynamic protection settings of the AC / DC distribution network, which are then distributed to the protection devices within the AC / DC distribution network. By combining power flow calculations and neural network trend predictions, protection settings can be calculated quickly and accurately, effectively achieving fault isolation under different operating conditions. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A flowchart illustrating the steps of an AC / DC distribution network protection method based on dynamic thresholds, provided in an embodiment of the present invention;

[0049] Figure 2 A simulation structure diagram of an adaptive protection system for AC / DC distribution networks provided in an embodiment of the present invention;

[0050] Figure 3 This is a structural block diagram of an AC / DC distribution network protection device based on dynamic threshold provided in an embodiment of the present invention. Detailed Implementation

[0051] This invention provides a dynamic threshold-based AC / DC distribution network protection method and related equipment to address the technical problem that existing AC / DC distribution networks, when connected to a high proportion of renewable energy, are typically still protected based on fixed thresholds, making it difficult to quickly and accurately achieve fault isolation under different operating conditions.

[0052] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0053] Please see Figure 1 , Figure 1 The present invention provides a flowchart of a dynamic threshold-based AC / DC distribution network protection method.

[0054] This invention provides a protection method for AC / DC distribution networks based on dynamic thresholds, comprising:

[0055] Step 101: Collect real-time and historical operating data of the AC / DC distribution network.

[0056] Real-time operational data refers to the electrical quantities and / or system status quantities of each node at the current moment during the operation of the AC / DC distribution network, including but not limited to node voltage, branch current, real-time output of distributed power sources, load power, and real-time status of network topology.

[0057] Historical operating data refers to various operating records stored in the past operating cycles of AC / DC distribution networks, including time-series data of electrical parameters and system state change data under different operating conditions and fault scenarios.

[0058] In this embodiment of the invention, a data acquisition network is established by deploying intelligent electronic devices at key nodes of the AC / DC distribution network. This enables full-domain, high-frequency acquisition of real-time operational data across the entire network, while simultaneously retrieving historical operational data stored in local or cloud databases, providing comprehensive data support for subsequent analysis. (Refer to...) Figure 2 This is a simulation structure diagram of the AC / DC distribution network adaptive protection system of the present invention. As a simulation verification section, this diagram illustrates the test framework and component interactions of the system under experimental conditions. The simulation structure diagram includes the following elements: the topology of the simulated distribution network, distributed power sources, and load connection points. These components are implemented using the simulation software MATLAB to verify the dynamic performance of the adaptive protection system.

[0059] The collected real-time operational data must first undergo a preprocessing process. The min-max normalization method is used to eliminate dimensional differences between different electrical quantities. A data verification mechanism can also be introduced to identify and correct abnormal or missing data, ensuring the integrity and accuracy of the input data. Historical data needs to be categorized and archived according to operating conditions and time dimensions to improve the efficiency of subsequent retrieval and analysis. Intelligent electronic devices refer to terminal devices deployed at various monitoring nodes in the distribution network, possessing data acquisition, preliminary processing, and communication interaction capabilities.

[0060] Step 102: When a state change is detected in the AC / DC distribution network or a fault signal is received, probabilistic power flow calculation is performed based on real-time operating data to determine the fault current probability distribution information.

[0061] Fault signals refer to alarm information of abnormal operating conditions issued by the power distribution network fault monitoring device, covering trigger signals of fault types such as short circuit, overvoltage, and power over-limit.

[0062] Fault current probability distribution information refers to the range of values ​​and corresponding probabilities of the current in each branch of the distribution network under fault conditions.

[0063] In this embodiment, the state change of the distribution network system can be identified by comparing real-time operating data and baseline operating data in real time. Simultaneously, it monitors in real time whether a fault signal uploaded by a fault monitoring device is received. If the difference between the real-time operating data and the baseline operating data exceeds a preset threshold, or if a fault signal is received, an AC / DC system is constructed based on a Gaussian mixture model. A probabilistic power flow calculation model is then used, and a linear relationship between the AC system state variables and node injected power is established through a linearized VSC power equation, thereby deriving the probability distribution characteristics of the fault current.

[0064] In addition, it can be extended to adapt to probabilistic modeling of multiple types of uncertainty sources. In addition to distributed power sources and loads, factors such as fluctuations in converter station control parameters can also be included to improve the comprehensiveness and accuracy of fault current probability distribution calculation.

[0065] In one example of the present invention, step 102 may include the following sub-steps:

[0066] When a change in the state of the AC / DC distribution network is detected or a fault signal is received, the real-time operating data is preprocessed to obtain preprocessed data.

[0067] A Gaussian mixture model was used to perform probability fitting on the preprocessed data to obtain the probability distribution of random variables;

[0068] A probabilistic power flow model for AC / DC distribution networks is constructed using a linearized model.

[0069] The probability power flow model is solved based on the probability distribution of random variables to determine the probability distribution information of fault current.

[0070] In this embodiment, when a change in the state of the AC / DC distribution network is detected (such as distributed power source switching, sudden load capacity changes, or converter station operation mode switching) or a fault signal is received (such as short circuit or overvoltage alarm), the real-time operating data collected by the intelligent electronic devices at each node of the power grid is first preprocessed, and the min-max normalization method is used to eliminate the dimensional differences of different electrical quantities such as voltage, current, and power.

[0071]

[0072] in, To run data in real time, For the normalized preprocessed data, The minimum value of real-time running data. This represents the maximum value of the real-time running data.

[0073] Meanwhile, a data verification mechanism can be introduced to identify and correct abnormal and missing data by comparing the fluctuation range of data at adjacent times and the logical relationship of related electrical quantities, so as to obtain preprocessed data that meets the accuracy requirements of subsequent calculations. Furthermore, it can be extended to access data compression algorithms to reduce storage and transmission pressure while ensuring data validity.

[0074] After obtaining the preprocessed data, a Gaussian Mixture Model (GMM) is then used to perform probability fitting on the preprocessed data. Based on historical operating data under similar conditions, the Gaussian component weights, mean vector, and covariance matrix of the GMM are estimated iteratively using the expectation-maximization algorithm. This constructs a probability distribution of random variables that can accurately describe uncertainties such as distributed power generation output and load power. The formula is as follows:

[0075]

[0076] Where M is the number of Gaussian components. As weight, It follows a Gaussian distribution. The derivation originates from probability theory: GMM approximates any distribution by weighting multiple Gaussian distributions, where the weights satisfy... The parameters are estimated from historical data using the expectation-maximization algorithm.

[0077] For AC / DC systems, the joint probability distribution of active power P and reactive power Q is:

[0078]

[0079] in, It is the mean vector. The covariance matrix describes the correlation between P and Q.

[0080] Next, combining the topology and equipment parameters of the AC / DC distribution network, and based on the practical engineering scenario where the voltage phase angle difference is small during normal operation of the distribution network, the power equation of the voltage source converter (VSC) is linearized. By inverting the network admittance matrix, the linear correlation coefficient matrix between state variables (such as node voltage and branch current) and node injected power is obtained, thus constructing a probabilistic power flow model that balances AC / DC coupling characteristics and computational efficiency.

[0081]

[0082] in, For state variables, α and β are coefficient matrices, and γ is a constant term. The coefficients of the linearized equations can also be dynamically adjusted for different power grid operating modes (such as rectification mode and inverter mode), enhancing the model's adaptability to system operating states.

[0083] In distribution networks, the voltage phase angle difference is small, and the power equation can be linearized. The coefficient matrix is ​​obtained by inverting the network admittance matrix, and the state variables... The probability distribution is as follows:

[0084]

[0085] in, , .

[0086] Based on the obtained probability distribution of random variables, and utilizing the linear invariance of the Gaussian mixture model, if X follows a Gaussian mixture model (GMM), then after the linear transformation Y = AX + b, Y also follows a GMM, with the mean and covariance updated according to the linear transformation. The random variables are substituted into the probabilistic power flow model for solution. By deriving the probabilistic characteristics of the state variables after the linear transformation (such as mean, variance, and probability density function), the range of fault current values ​​and corresponding probabilities for each branch are further calculated, thus determining the probability distribution information of the fault current.

[0087] In addition, a deviation verification mechanism between real-time data and model output can be introduced simultaneously during the solution process. If the deviation exceeds the preset threshold, online fine-tuning of GMM parameters or linearized model coefficients will be initiated to ensure that the fault current probability distribution information can accurately reflect the actual fault characteristics of the power grid.

[0088] In another example of the present invention, after performing step 102, the method further includes the following steps:

[0089] Obtain existing protection settings;

[0090] Calculate the overcurrent probability based on existing protection settings and fault current probability distribution information;

[0091] If the overcurrent probability is greater than the preset probability threshold, then jump to the step of calling the preset neural network to combine historical operating data to predict the trend of electrical parameter changes.

[0092] If the overcurrent probability is not greater than the preset probability threshold, then proceed to the step of collecting real-time and historical operating data of the AC / DC distribution network.

[0093] In this embodiment, a communication connection is established with the relay protection devices at each node in the AC / DC distribution network. Existing protection settings (including core parameters such as operating thresholds and operating time limits) for overcurrent protection and voltage protection of each branch are retrieved. Simultaneously, related information such as the setting time and applicable operating conditions of these settings are synchronously obtained from the system database. Then, based on the existing protection settings and the fault current probability distribution information obtained in the preceding steps, the overcurrent probability is calculated.

[0094]

[0095] The cumulative distribution function (CDF) is obtained through the integral probability density function.

[0096] The CDF expression for the cumulative distribution function GMM is:

[0097]

[0098] In addition, a coordination assessment index can be used for evaluation. This index includes a failure-to-operate risk index, a false-operation risk index, and a selective assessment index. Specifically, it includes: calculating the probability of the fault current exceeding the existing protection action threshold based on the fault current probability density function (failure-to-operate risk index), the probability of the normal operating current falsely triggering the existing protection threshold (false-operation risk index), and simultaneously calculating the coordination margin of the action thresholds of upper and lower level protection devices in conjunction with the power grid topology (selective assessment index). Finally, the calculated coordination assessment index is compared with the preset change requirement conditions (such as failure-to-operate risk index exceeding 5%, false-operation risk index exceeding 3%, and selective assessment index below 80%). If the index meets the change requirement conditions, it indicates that the current existing protection settings cannot adapt to the current fault characteristics of the power grid, and the process immediately jumps to step 103. If the index does not meet the change requirement conditions, it indicates that the existing protection settings can still ensure the reliable operation of the protection system. In this case, the process returns to step 101 to continue dynamically monitoring the power grid operation status to balance resource consumption and protection setting updates.

[0099] In addition, the threshold for changing demand conditions can be adjusted based on changes in the quarterly operating conditions of the power grid (such as peak load season and peak season for renewable energy output), thereby improving the flexibility of the judgment logic.

[0100] Step 103: Invoke the preset neural network and combine it with historical operating data to predict the trend information of electrical parameter changes;

[0101] Electrical parameter change trend information refers to the evolution direction and magnitude characteristics of key electrical quantities (such as node voltage, branch current, fault current, etc.) in AC / DC distribution networks over a future period.

[0102] In this embodiment of the invention, a three-layer BP (Back Propagation) neural network preset in the system is invoked to organize historical operating data into model input samples according to time series. The input layer corresponds to historical operating data at multiple time points, the hidden layer uses the Sigmoid activation function to achieve nonlinear feature extraction, and the output layer uses a linear activation function to output the single-step electrical parameter change trend information.

[0103] Furthermore, an online update mechanism can be introduced during model operation to periodically fine-tune the neural network weights and biases using newly collected historical data, correcting prediction biases. At the same time, the latest fragments of real-time running data are incorporated into the prediction input to achieve rolling prediction of electrical parameter change trends, providing accurate trend references for subsequent threshold optimization.

[0104] In one example of the present invention, the neural network is a three-layer BP neural network; step 103 may include the following sub-steps:

[0105] A target prediction model is obtained by training a pre-defined three-level BP neural network using historical operational data.

[0106] The target prediction model is used to predict the trend of electrical parameters in future acquisition cycles based on real-time operating data.

[0107] Furthermore, the three-layer BP neural network includes an input layer with 5 nodes, a hidden layer with 10 nodes, and an output layer with 1 node; the output layer is:

[0108]

[0109] in, It is the output of the j-th node in the hidden layer. The weights connecting the j-th node in the hidden layer to the nodes in the output layer are... This refers to the bias term for the output layer nodes. The activation function of the output layer. This provides information on the changing trends of electrical parameters.

[0110] In this embodiment, the historical operation data is first processed into a time series, and the training set and validation set are divided according to the collection time order. For example, 85% of the historical data of nearly 6 months is divided into the training set for model learning and 15% is divided into the validation set for generalization ability verification. The min-max normalization method consistent with the preprocessing of real-time operation data is used to eliminate the difference in units and ensure that the data format matches the three-level BP neural network structure. The five nodes of the input layer correspond to the electrical parameter time series data of five consecutive historical collection cycles.

[0111] Initialize network parameters, including connection weights between the input layer and hidden layers (10 nodes), hidden layer bias terms, and connection weights from the hidden layers to the output layer (1 node), and output layer bias terms. The hidden layers use the Sigmoid activation function to extract the nonlinear variation characteristics of electrical parameters, while the output layer uses a linear activation function to accommodate the prediction requirements of continuous electrical parameter values. During training, the network output is calculated via forward propagation: first, the output of the j-th node of each hidden layer is calculated based on the historical input data and hidden layer parameters; then, the model prediction value is obtained by combining the calculation logic of the output layer weights, bias terms, and activation functions in the formula. The mean squared error is used as the loss function to quantify the deviation between the predicted value and the actual historical data. The gradient of the loss function with respect to each weight and bias term is calculated using the backpropagation algorithm. The parameters are iteratively adjusted using gradient descent until the loss function converges to a preset threshold or reaches the maximum number of iterations, thus obtaining the target prediction model.

[0112] During the prediction phase, the real-time operating data undergoes the same preprocessing operations as during the training phase. The real-time electrical parameter data from the latest five consecutive acquisition cycles are extracted and input into the target prediction model. After forward propagation, the output of each node in the hidden layer is calculated first, and then the output layer operation is completed according to the formula. Finally, the trend information of electrical parameter changes in the next acquisition cycle is output.

[0113] The hidden layer output is:

[0114]

[0115] in, The value of the i-th node in the input layer. The weights connecting the i-th node in the input layer to the j-th node in the hidden layer are... The bias term for the j-th node in the hidden layer provides an independent regulatory baseline for each neuron. This is the result of a linear combination of activation functions applied to the input data. This is the final output of the j-th node in the hidden layer.

[0116] The output layer outputs:

[0117]

[0118] in, It is the output from the j-th node of the hidden layer, which is used as the input to the output layer. Let J be the weight connecting the j-th node in the hidden layer to the node in the output layer. This refers to the bias term for the output layer nodes. The activation function of the output layer. This provides information on the changing trends of electrical parameters in the final predicted output of the entire neural network.

[0119] Step 104: With the goal of minimizing the expected protection loss, a dynamic threshold optimization model is constructed by combining the fault current probability distribution information and the electrical parameter change trend information.

[0120] The expected loss of protection refers to the comprehensive loss caused by the failure or malfunction of the protection device.

[0121] The dynamic threshold optimization model refers to the protection setting optimization model that integrates the real-time fault probability characteristics of the system with the trend of electrical parameters, and has the characteristic of adapting to the dynamic changes of system operating conditions.

[0122] In this embodiment, with the core objective of minimizing the expected protection loss, the fault current probability distribution information is used as the probabilistic constraint basis of the model. At the same time, the electrical parameter change trend information is introduced to correct the time-domain adaptability of the model. Combined with probabilistic sensitivity analysis, the impact of system parameter changes on protection risk is quantified, and a dynamic threshold optimization model is constructed.

[0123] The objective function of the dynamic threshold optimization model is:

[0124]

[0125] in, To protect expected losses, The dynamic protection settings to be optimized. It is a random current variable; The fault current probability distribution information is represented by the probability density function of the current. Is when The unit loss coefficient at that time Is when The unit loss coefficient at that time.

[0126] It should be noted that, in the process of constructing the threshold optimization model, in order to adapt to the protection risk control under the uncertainty fluctuations of the power grid, the failure to operate loss is defined as the overcurrent probability ( The weighted sum of the power parameter fluctuation (such as the output of distributed power sources) × the unit loss coefficient for failure to operate × the probability sensitivity factor is used to quantify the impact of power parameter fluctuations on the risk of failure to operate. Higher weights are given to the failure to operate losses corresponding to high sensitivity parameters, reflecting the amplification effect of such parameter fluctuations on protection risks.

[0127] The probability sensitivity factor is:

[0128]

[0129] By differentiating the CDF and applying the chain rule, we can consider the impact of power adjustment on the shift of the probability distribution.

[0130] Step 105: Iteratively solve the dynamic threshold optimization model to obtain the dynamic protection settings of the AC / DC distribution network, and send them to the protection devices in the AC / DC distribution network.

[0131] Dynamic protection settings refer to the action threshold of protection devices that are adapted to the current system operating conditions. Unlike traditional fixed settings, they have the characteristic of dynamically adjusting with the system status.

[0132] Protection devices refer to devices used for fault identification and isolation in AC / DC distribution networks, and play a protective role, such as relay protection devices.

[0133] In this embodiment, the gradient descent method is used to iteratively solve the dynamic threshold optimization model. The optimal solution of the model is quickly obtained by using the inverse function of the current accumulation distribution function, and a dynamic protection setting value adapted to the current operating condition is obtained. After the setting value is solved, the dynamic protection setting value is sent to the protection devices of each node of the distribution network through the communication link of the protection execution module.

[0134] In addition, the reception and activation status of the protection device can be verified after the setting is issued. At the same time, a setting version archiving system can be established to achieve full traceability of the protection setting adjustment process and ensure the stable implementation of the adaptive protection function.

[0135] In one example of the present invention, step 105 may include the following sub-steps:

[0136] The dynamic threshold optimization model is iteratively solved using the gradient descent method under probabilistic sensitivity constraints to obtain the dynamic protection settings of the AC / DC distribution network;

[0137] Dynamic protection settings are sent to protection devices within the AC / DC distribution network to update the local settings of the protection devices online;

[0138] Dynamic protection settings are:

[0139]

[0140] in, The optimized dynamic protection settings, It is the inverse function of the cumulative distribution function. Is when The unit loss coefficient at that time Is when The unit loss coefficient at that time.

[0141] In this embodiment, the protection action threshold of each branch of the AC / DC distribution network is used as the optimization variable. The initial value of the threshold is initialized based on the existing protection settings. Then, the expected protection loss corresponding to the current threshold is calculated. At the same time, combined with the probabilistic sensitivity constraint (such as the overcurrent probability needing to be lower than the preset safety upper limit under high sensitivity power parameter fluctuation), the constraint is transformed into a penalty term and embedded into the objective function using the Lagrange multiplier method. Then, the gradient descent method is used to calculate the gradient of the objective function with respect to the threshold (the gradient needs to be related to the derivative of the overcurrent probability with respect to the threshold, the coupling relationship between the probabilistic sensitivity factor and the power parameter). The learning rate is dynamically adjusted to avoid iterative oscillation. After each iteration, the fault current probability distribution and the corresponding probabilistic sensitivity factor are recalculated, the constraint satisfaction is checked, and the objective function value is updated. This process is repeated until the objective function converges and all probabilistic sensitivity constraints are satisfied, thus obtaining a dynamic protection setting that adapts to the current power grid operating conditions.

[0142] Subsequently, an encrypted communication link is established with each protection device through the distribution network communication bus conforming to the IEC 61850 standard. First, the matching between the dynamic protection settings and the device model and corresponding branch parameters is verified. Then, the generated dynamic protection settings are distributed according to the unique address of the protection device, triggering the local setting update process of the protection device. The update confirmation message returned by the device is received synchronously, thereby realizing the online update of the settings.

[0143] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0144] The AC / DC distribution network protection device based on dynamic threshold provided in the embodiments of the present invention will be described below. The AC / DC distribution network protection device based on dynamic threshold described below can be referred to in correspondence with the AC / DC distribution network protection method based on dynamic threshold described above.

[0145] Please see Figure 3 , Figure 3 The diagram shows a structural block diagram of an AC / DC distribution network protection device based on a dynamic threshold according to an embodiment of the present invention.

[0146] This invention provides an AC / DC distribution network protection device based on dynamic threshold, comprising:

[0147] The data acquisition module 301 is used to collect real-time and historical operating data of the AC / DC distribution network.

[0148] The power flow calculation module 302 is used to perform probabilistic power flow calculation based on real-time operating data to determine the probability distribution information of fault current when a state change in the AC / DC distribution network is detected or a fault signal is received.

[0149] The parameter change trend prediction module 303 is used to call a preset neural network and combine it with historical operating data to predict the change trend information of electrical parameters.

[0150] The model building module 304 is used to build a dynamic threshold optimization model with the goal of minimizing the expected protection loss, by combining fault current probability distribution information and electrical parameter change trend information.

[0151] The setting value determination and distribution module 305 is used to iteratively solve the dynamic threshold optimization model to obtain the dynamic protection setting value of the AC / DC distribution network and distribute it to the protection devices in the AC / DC distribution network.

[0152] Optionally, the power flow calculation module 302 is specifically used for:

[0153] When a change in the state of the AC / DC distribution network is detected or a fault signal is received, the real-time operating data is preprocessed to obtain preprocessed data.

[0154] A Gaussian mixture model was used to perform probability fitting on the preprocessed data to obtain the probability distribution of random variables;

[0155] A probabilistic power flow model for AC / DC distribution networks is constructed using a linearized model.

[0156] The probability power flow model is solved based on the probability distribution of random variables to determine the probability distribution information of fault current.

[0157] Optionally, the neural network is a three-layer BP neural network; the parameter change trend prediction module 303 is specifically used for:

[0158] A target prediction model is obtained by training a pre-defined three-level BP neural network using historical operational data.

[0159] The target prediction model is used to predict the trend of electrical parameters in future acquisition cycles based on real-time operating data.

[0160] Optionally, the three-layer BP neural network includes an input layer with 5 nodes, a hidden layer with 10 nodes, and an output layer with 1 node; the output layer is:

[0161]

[0162] in, It is the output of the j-th node in the hidden layer. The weights connecting the j-th node in the hidden layer to the nodes in the output layer are... This refers to the bias term for the output layer nodes. The activation function of the output layer. This provides information on the changing trends of electrical parameters.

[0163] Optionally, the setting determination and distribution module 305 is specifically used for:

[0164] The dynamic threshold optimization model is iteratively solved using the gradient descent method under probabilistic sensitivity constraints to obtain the dynamic protection settings of the AC / DC distribution network;

[0165] Dynamic protection settings are sent to protection devices within the AC / DC distribution network to update the local settings of the protection devices online;

[0166] Dynamic protection settings are:

[0167]

[0168] in, The optimized dynamic protection settings, It is the inverse function of the cumulative distribution function. Is when The unit loss coefficient at that time Is when The unit loss coefficient at that time.

[0169] Optionally, the objective function of the dynamic threshold optimization model is:

[0170]

[0171] in, To protect expected losses, The dynamic protection settings to be optimized. It is a random current variable; The fault current probability distribution information is represented by the probability density function of the current. Is when The unit loss coefficient at that time Is when The unit loss coefficient at that time.

[0172] Optionally, the device also includes an overcurrent detection module, specifically used for:

[0173] Obtain existing protection settings;

[0174] Calculate the overcurrent probability based on existing protection settings and fault current probability distribution information;

[0175] If the overcurrent probability is greater than the preset probability threshold, then jump to the step of calling the preset neural network to combine historical operating data to predict the trend of electrical parameter changes.

[0176] If the overcurrent probability is not greater than the preset probability threshold, then proceed to the step of collecting real-time and historical operating data of the AC / DC distribution network.

[0177] This invention provides an electronic device, including a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the AC / DC distribution network protection method based on dynamic threshold as described in any embodiment of this invention.

[0178] This invention provides a computer-readable storage medium storing a computer program thereon, which, when executed, implements the AC / DC distribution network protection method based on dynamic threshold as described in any embodiment of this invention.

[0179] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device and module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0180] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0181] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0182] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0183] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0184] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A protection method for AC / DC distribution networks based on dynamic thresholds, characterized in that, include: Real-time acquisition of real-time and historical operating data of AC / DC distribution networks; When a state change is detected in the AC / DC distribution network or a fault signal is received, probabilistic power flow calculation is performed according to the real-time operating data to determine the fault current probability distribution information. The system invokes a preset neural network and combines it with the historical operating data to predict the trend of electrical parameter changes. With the goal of minimizing expected losses, a dynamic threshold optimization model is constructed by combining the fault current probability distribution information and the electrical parameter change trend information. The dynamic threshold optimization model is iteratively solved to obtain the dynamic protection settings of the AC / DC distribution network, and then distributed to the protection devices within the AC / DC distribution network.

2. The AC / DC distribution network protection method based on dynamic threshold according to claim 1, characterized in that, The step of determining the fault current probability distribution information by performing probabilistic power flow calculations based on the real-time operating data when a state change in the AC / DC distribution network is detected or a fault signal is received includes: When a state change is detected in the AC / DC distribution network or a fault signal is received, the real-time operating data is preprocessed to obtain preprocessed data. A Gaussian mixture model is used to perform probability fitting on the preprocessed data to obtain the probability distribution of random variables; A probabilistic power flow model for the AC / DC distribution network is constructed using a linearized model. The probability power flow model is solved based on the probability distribution of the random variables to determine the probability distribution information of the fault current.

3. The AC / DC distribution network protection method based on dynamic threshold according to claim 1, characterized in that, The neural network is a three-layer BP neural network; the step of calling the preset neural network in conjunction with the historical operating data to predict the trend information of electrical parameter changes includes: The historical operating data is used to train a preset three-level BP neural network to obtain a target prediction model; The target prediction model is used to predict the electrical parameter change trend information in the future acquisition cycle based on the real-time operating data.

4. The AC / DC distribution network protection method based on dynamic threshold according to claim 3, characterized in that, The three-layer BP neural network includes an input layer with 5 nodes, a hidden layer with 10 nodes, and an output layer with 1 node; the output layer is: in, It is the output of the j-th node in the hidden layer. The weights connecting the j-th node in the hidden layer to the nodes in the output layer are... This refers to the bias term for the output layer nodes. The activation function of the output layer. This provides information on the changing trends of electrical parameters.

5. The AC / DC distribution network protection method based on dynamic threshold according to claim 1, characterized in that, The step of iteratively solving the dynamic threshold optimization model to obtain the dynamic protection settings of the AC / DC distribution network and distributing them to the protection devices within the AC / DC distribution network includes: The dynamic threshold optimization model is iteratively solved using the gradient descent method under probabilistic sensitivity constraints to obtain the dynamic protection settings of the AC / DC distribution network. The dynamic protection settings are sent to the protection devices in the AC / DC distribution network to update the local settings of the protection devices online. The dynamic protection setting is: in, The optimized dynamic protection settings, It is the inverse function of the cumulative distribution function. Is when The unit loss coefficient at that time Is when The unit loss coefficient at that time.

6. The AC / DC distribution network protection method based on dynamic threshold according to claim 1, characterized in that, The objective function of the dynamic threshold optimization model is: in, To protect expected losses, The dynamic protection settings to be optimized. It is a random current variable; The fault current probability distribution information is represented by the probability density function of the current. Is when The unit loss coefficient at that time Is when The unit loss coefficient at that time.

7. The AC / DC distribution network protection method based on dynamic threshold according to any one of claims 1 to 6, characterized in that, After performing the step of calculating the probability distribution information of the fault current based on the real-time operating data when a state change is detected in the AC / DC distribution network or a fault signal is received, the method further includes: Obtain existing protection settings; Calculate the overcurrent probability based on the existing protection settings and the fault current probability distribution information; If the overcurrent probability is greater than a preset probability threshold, then proceed to the step of calling a preset neural network to combine the historical operating data and predict the trend information of electrical parameter changes. If the overcurrent probability is not greater than the preset probability threshold, then proceed to the step of collecting real-time operating data and historical operating data of the AC / DC distribution network.

8. A protection device for AC / DC distribution networks based on dynamic threshold, characterized in that, include: The data acquisition module is used to collect real-time and historical operating data of the AC / DC distribution network. The power flow calculation module is used to perform probabilistic power flow calculations based on the real-time operating data when a state change is detected in the AC / DC distribution network or a fault signal is received, and to determine the fault current probability distribution information. The parameter change trend prediction module is used to call a preset neural network and combine it with the historical operating data to predict the change trend information of electrical parameters. The model building module is used to construct a dynamic threshold optimization model with the goal of minimizing the expected protection loss, by combining the fault current probability distribution information and the electrical parameter change trend information. The setting value determination and distribution module is used to iteratively solve the dynamic threshold optimization model to obtain the dynamic protection setting value of the AC / DC distribution network, and distribute it to the protection devices in the AC / DC distribution network.

9. An electronic device, characterized in that, The system includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the AC / DC distribution network protection method based on dynamic thresholds as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the AC / DC distribution network protection method based on dynamic thresholds as described in any one of claims 1-7.

Images