Method for determining current protection constant value of low-voltage power distribution network

Abstract

The invention relates to a low-voltage distribution network current protection constant value determination method, which comprises the steps of constructing a low-voltage side current prediction model based on adaptive feature selection, and adaptively selecting different features through the low-voltage side current prediction model, obtaining a prediction model of to-be-maintained / defect-eliminated interval low-voltage side current and a prediction model of standby power supply interval low-voltage side current; respectively predicting the interval low-voltage side current to be overhauled / defect-eliminated and the interval low-voltage side current of the standby power supply through the prediction model of the interval low-voltage side current to be overhauled / defect-eliminated and the prediction model of the interval low-voltage side current of the standby power supply; taking the sum of the interval low-voltage side current to be overhauled / defect-eliminated and the interval low-voltage side current of the standby power supply as the total load current after loop closing; setting a high-voltage side protection constant value of the standby power supply interval based on the total load current after loop closing; an associated current protection setpoint is set based on the predicted interval low side current. The method is beneficial to flexibly and accurately determining the current protection constant value of the low-voltage power distribution network, so that the reliability and the safety of operation of related power equipment are improved.

Description

Technical Field

[0001] The invention relates to the technical field of relay protection, and in particular to a method for determining a current protection setting value of a low-voltage distribution network. Background Art

[0002] With the rapid development of new power systems, the scale of regional power grids is getting larger and larger, the number of power grid equipment is increasing, and the possibility of equipment failure is increasing. However, the requirements for power supply reliability are getting higher and higher. The traditional information management and setting calculation methods of relay protection are gradually unable to adapt to the requirements of power grid development. Among them, the setting value of low-voltage distribution network protection is of great significance to improving the management level of relay protection.

[0003] At present, the setting of low-voltage power grid protection settings is mainly manual or simple local software calculations. Manual calculations have a high error rate, and local software calculations need to be deployed separately on each computer. In addition, the power grid parameters required for calculation are electronic or paper data reported separately by each equipment management unit. The data reliability is poor, resulting in inaccurate current protection settings for the low-voltage distribution network, and the inability to flexibly adjust them according to actual conditions, which reduces the reliability of the operation of related power equipment such as backup power supply devices.

[0004] Patent CN103227452A discloses a method for calculating the constant value of relay protection in a power plant, which includes the following steps: determining that the basic side is the low-voltage side, and the secondary current is close to the nominal current; obtaining the secondary rated current on the high-voltage side; obtaining the balance coefficient; calculating the minimum operating current; calculating the ratio braking coefficient, connecting the three-side braking; calculating the harmonic braking coefficient; obtaining the inflection point current; setting the instantaneous element current value; releasing the current transformer disconnection judgment step; determining the rated current; and verifying the sensitivity. This method can ensure the accurate and reliable operation of relay protection. Patent CN111398864A discloses a method, device and storage medium for dynamic setting of protection setting value, the method includes: establishing a heat distribution model inside the transformer, and studying the temperature rise characteristics of the transformer according to the heat distribution model and the thermal circuit differential equation; simulating the temperature field of the transformer, calculating the temperature rise and average temperature rise of the hottest point of the transformer distribution winding, and comparing it with the temperature rise and average temperature rise of the hottest point of the distribution winding of the mineral oil distribution winding; the patent solves the technical problem that the existing low-voltage main switch protection technology fails to effectively play the role of the high overload capacity distribution transformer when set to the rated current value, and if the set current value is too high, the high overload capacity distribution transformer cannot withstand the impact of long-term overload and cause damage. However, the protection setting value setting method disclosed in the above patent is not suitable for the setting value setting of the backup power supply input device, and it is impossible to establish a low-voltage side current prediction model according to the actual working conditions of the backup power supply input device, so as to flexibly and accurately set the current type setting value of the backup power supply input device and other related power equipment. Summary of the invention

[0005] The object of the present invention is to provide a method for determining a current protection setting value of a low-voltage distribution network, which method is conducive to flexibly and accurately determining the current protection setting value of the low-voltage distribution network, thereby improving the reliability and safety of the operation of related power equipment.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is: a method for determining a current protection setting value of a low-voltage distribution network, comprising:

[0007] Construct a low-voltage side current prediction model based on adaptive feature selection, and adaptively select different features through the low-voltage side current prediction model to obtain the prediction model of the low-voltage side current of the interval to be repaired / eliminated and the prediction model of the low-voltage side current of the backup power supply interval;

[0008] The low-voltage side current of the interval to be repaired / fault eliminated and the low-voltage side current of the interval of the backup power supply are predicted respectively by using the prediction model of the low-voltage side current of the interval to be repaired / fault eliminated and the prediction model of the low-voltage side current of the interval of the backup power supply;

[0009] The total load current after the loop is closed is the sum of the low-voltage side current of the interval to be repaired / fault eliminated and the low-voltage side current of the standby power supply interval;

[0010] The high-voltage side protection setting value of the backup power supply interval is set based on the total load current after the loop is closed;

[0011] Set relevant current protection settings based on the predicted low-voltage side current of the interval.

[0012] Furthermore, a low-voltage side current prediction model based on adaptive feature selection is constructed based on K-means clustering and ProbLassoNet algorithm, and different features are adaptively selected through the low-voltage side current prediction model to obtain the prediction model of the low-voltage side current of the interval to be repaired / eliminated and the prediction model of the low-voltage side current of the backup power supply interval, which specifically includes:

[0013] Acquire current data and corresponding time and scene data, then preprocess the acquired data to construct a training data set;

[0014] The data is clustered by K-means clustering to find out several change patterns of current data at different times and in different scenarios;

[0015] Constructing a ProbLassoNet model, and training the ProbLassoNet model through a training data set; the ProbLassoNet model adaptively determines the effective features of each change mode according to different change modes of current data, thereby determining a current prediction model corresponding to each change mode;

[0016] Based on the obtained ProbLassoNet model, prediction models for the low-voltage side current of the interval to be repaired / eliminated and the low-voltage side current of the backup power supply interval are constructed respectively.

[0017] Furthermore, the time and scene data include weather data, date time period data and new energy power generation situation data.

[0018] Furthermore, the acquired data is preprocessed, specifically including:

[0019] Find outliers and invalid values ​​in the data and replace them by interpolation;

[0020] Normalize the data to eliminate the mutual influence of different data levels.

[0021] Furthermore, the ProbLassoNet model is mainly composed of an input layer, a hidden layer, an output layer, and a residual layer between the input layer and the output layer; wherein the input layer inputs historical current data and other feature data, and the other feature data are time and scene data; the hidden layer has a set number of layers; and the output layer is a multi-quantile output.

[0022] Furthermore, the mapping function of the ProbLassoNet model is:

[0023] f(X)=Xθ+f ω (X) (1)

[0024] Among them, X is the input data; f ω (·) is the mapping function from the input layer to the hidden layer; θ and ω represent the weight parameters of the feedforward neural network and the residual layer;

[0025] The objective function of the ProbLassoNet model is:

[0026]

[0027] Where Y represents the predicted data; L(X, Y; θ, ω) represents the loss value of the model's multi-quantile output; λ||θ||1 represents the l1 regularization of the residual layer parameters, and λ represents the degree of control regularization; is the element of the jth row of the first hidden layer; θ j is the element of the j-th row of the residual layer.

[0028] Furthermore, the training of the ProbLassoNet model includes the following two steps:

[0029] First, the initial model is trained by updating θ and ω with the goal of making L(X, Y; θ, ω) as small as possible.

[0030] Then, θ and ω are updated by the proximal gradient method (1) ; The soft threshold operator is set as:

[0031] S λ (x)=sign(x)·max{x|-λ,0} (3)

[0032] Here, sign(x) means that it is 1 when x>0 and -1 when x<0.

[0033] Furthermore, the total load current after closing the loop is matched with the high-voltage side protection setting of the backup power supply interval, and the high-voltage side protection setting of the backup power supply interval is set greater than the total load current after closing the loop to avoid accident tripping on the low and medium voltage sides.

[0034] Furthermore, the related current protection constants include: a first-stage overcurrent constant, a second / third-stage overcurrent constant, an acceleration value after the switch is closed, and a leakage current constant.

[0035] Furthermore, the first overcurrent constant is set to 15 times the predicted current of the interval to be repaired / eliminated, the second / third overcurrent constant is set to 1.5 times the predicted current of the interval to be repaired / eliminated, the acceleration value after the switch is closed is set to 1.5 times the predicted current of the interval to be repaired / eliminated, and the leakage current constant is set to 100mA.

[0036] Compared with the prior art, the present invention has the following beneficial effects: the present invention provides a method for determining a current protection setting of a low-voltage distribution network, which establishes a prediction model for the low-voltage side current of an interval to be repaired / eliminated and a prediction model for the low-voltage side current of a backup power supply interval by constructing a low-voltage side current prediction model based on adaptive feature selection, and then determines the high-voltage side protection setting of the backup power supply interval and other related current protection setting based on the predicted current of the model, so that the current protection setting of the low-voltage distribution network can be set and adjusted according to actual power consumption conditions, thereby improving the flexibility and accuracy of the setting of the current protection setting of the low-voltage distribution network, thereby improving the reliability and safety of the operation of related power equipment such as a backup power supply input device. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a flow chart of a method implementation of an embodiment of the present invention;

[0038] Figure 2 It is a block diagram of the implementation principle of the low-voltage side current prediction model based on adaptive feature selection in an embodiment of the present invention;

[0039] Figure 3 Schematic diagram of the ProbLassoNet model in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0041] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0042] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0043] like Figure 1 As shown, this embodiment provides a method for determining a current protection setting value of a low-voltage distribution network, comprising the following steps:

[0044] S1. Construct a low-voltage side current prediction model based on adaptive feature selection, and adaptively select different features through the low-voltage side current prediction model to obtain the prediction model of the low-voltage side current of the interval to be repaired / eliminated and the prediction model of the low-voltage side current of the backup power supply interval.

[0045] S2. Predict the low-voltage side current of the interval to be repaired / fault eliminated and the low-voltage side current of the backup power supply interval respectively through the prediction model of the low-voltage side current of the interval to be repaired / fault eliminated and the prediction model of the low-voltage side current of the backup power supply interval.

[0046] S3. The sum of the low-voltage side current of the interval to be repaired / fault eliminated and the low-voltage side current of the standby power supply interval is the total load current after the low-voltage section is closed.

[0047] S4. Set the high-voltage side protection setting of the backup power supply interval based on the total load current after the loop is closed.

[0048] S5. Set relevant current protection constants based on the predicted interval low-voltage side current.

[0049] like Figure 2As shown, in this embodiment, a low-voltage side current prediction model based on adaptive feature selection is constructed based on K-means clustering and ProbLassoNet algorithm, and different features are adaptively selected through the low-voltage side current prediction model to obtain the prediction model of the low-voltage side current of the interval to be repaired / eliminated and the prediction model of the low-voltage side current of the backup power supply interval. Finally, the low-voltage side current of the interval to be repaired / eliminated and the low-voltage side current of the backup power supply interval are predicted respectively through the prediction model of the low-voltage side current of the interval to be repaired / eliminated and the prediction model of the low-voltage side current of the backup power supply interval. Its implementation process can be divided into four stages.

[0050] (1) Data preprocessing stage

[0051] First, current data and corresponding time and scene data are obtained. The time and scene data include weather data, date period data and new energy power generation data. Then, the obtained data is preprocessed to construct a training data set.

[0052] In this embodiment, the sampling frequency of all data is 1 hour.

[0053] Since the acquired data may contain many abnormal values ​​and invalid values, it is necessary to preprocess the acquired data; specifically, the following steps are required:

[0054] 1) Find outliers and invalid values ​​in the data and replace them by interpolation;

[0055] 2) Normalize the data to eliminate the mutual influence of different data magnitudes.

[0056] (2) Classification stage

[0057] Since the current data has different operating modes at different times and in different scenarios, this method uses the K-means clustering method to cluster / classify the electricity consumption data, and combines it with weather, time, renewable energy power generation and other data to obtain several change modes of current data at different times and in different scenarios.

[0058] (3) Prediction model construction stage

[0059] A ProbLassoNet model is constructed and trained using a training data set. The ProbLassoNet model adaptively determines effective features of each change mode according to different change modes of current data, thereby determining a current prediction model corresponding to each change mode.

[0060] like Figure 3As shown in the figure, the ProbLassoNet model is mainly composed of an input layer, a hidden layer, an output layer, and a residual layer between the input layer and the output layer. The input layer inputs historical current data and other feature data, including time and scene data such as weather, time, and new energy power generation. The hidden layer has an arbitrary number of layers. The output layer is a multi-quantile output.

[0061] The mapping function of the ProbLassoNet model is:

[0062] f(X)=Xθ+f ω (X) (1)

[0063] Among them, X is the input data; f ω (·) is the mapping function from the input layer to the hidden layer; θ and ω represent the weight parameters of the feedforward neural network and the residual layer.

[0064] The objective function of the ProbLassoNet model is:

[0065]

[0066] Where Y represents the predicted data; L(X, Y; θ, ω) represents the loss value of the model's multi-quantile output; λ||θ||1 represents the l1 regularization of the residual layer parameters, and λ represents the degree of control regularization; is the element of the jth row of the first hidden layer; θ j is the element of the j-th row of the residual layer.

[0067] The training of the ProbLassoNet model includes the following two steps:

[0068] First, the initial model is trained by updating θ and ω with the goal of making L(X, Y; θ, ω) as small as possible.

[0069] Then, θ and ω are updated by the proximal gradient method (1) ; The soft threshold operator is set as:

[0070] S λ (x)=sign(x)·max{x|-λ,0} (3)

[0071] Here, sign(x) means that it is 1 when x>0 and -1 when x<0.

[0072] Finally, based on the obtained ProbLassoNet model, prediction models for the low-voltage side current of the interval to be repaired / fault eliminated and the low-voltage side current of the backup power supply interval are constructed respectively.

[0073] (4) Current prediction stage

[0074] After obtaining the prediction model of the low-voltage side current of the interval to be repaired / fault eliminated and the prediction model of the low-voltage side current of the backup power supply interval, the obtained prediction models are used to predict the low-voltage side current of the interval to be repaired / fault eliminated and the low-voltage side current of the backup power supply interval respectively.

[0075] In this method, the total load current after the loop is closed is set to be the sum of the low-voltage side current of the interval to be repaired / eliminated and the low-voltage side current of the standby power supply interval.

[0076] Then, the total load current after the loop is closed is matched with the high-voltage side protection setting of the backup power supply interval, and the high-voltage side protection setting of the backup power supply interval is set to be greater than the total load current after the loop is closed to avoid accident tripping on the low and medium voltage sides.

[0077] On this basis, other relevant current protection constants are set, including: overcurrent first stage constant, overcurrent second / third stage constant, acceleration value after switch closing, and leakage current constant.

[0078] In this embodiment, the first overcurrent setting is set to 15 times the predicted current of the interval to be repaired / eliminated, the second / third overcurrent setting is set to 1.5 times the predicted current of the interval to be repaired / eliminated, the acceleration value after the switch is closed is set to 1.5 times the predicted current of the interval to be repaired / eliminated, and the leakage current setting is set to 100mA. It should be noted that the above setting values ​​can be modified according to actual conditions.

[0079] In addition, the voltage setting value can be further set according to the actual situation. Usually, the voltage setting value is 70% of the rated voltage, and the non-voltage setting value is 30% of the rated voltage.

[0080] This completes the setting of current and voltage settings for related power equipment such as the backup power supply input device.

[0081] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0082] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0083] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0084] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0085] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A method for determining a current protection setting value of a low-voltage distribution network, characterized in that: include: Construct a low-voltage side current prediction model based on adaptive feature selection, and adaptively select different features through the low-voltage side current prediction model to obtain the prediction model of the low-voltage side current of the interval to be repaired / eliminated and the prediction model of the low-voltage side current of the backup power supply interval; The low-voltage side current of the interval to be repaired / fault eliminated and the low-voltage side current of the interval of the backup power supply are predicted respectively by using the prediction model of the low-voltage side current of the interval to be repaired / fault eliminated and the prediction model of the low-voltage side current of the interval of the backup power supply; The total load current after the loop is closed is the sum of the low-voltage side current of the interval to be repaired / fault eliminated and the low-voltage side current of the standby power supply interval; The high-voltage side protection setting value of the backup power supply interval is set based on the total load current after the loop is closed; Set relevant current protection settings based on the predicted low-voltage side current of the interval.

2. The method for determining a current protection setting value of a low-voltage distribution network according to claim 1, characterized in that: Based on K-means clustering and ProbLassoNet algorithm, a low-voltage side current prediction model based on adaptive feature selection is constructed. Different features are adaptively selected through the low-voltage side current prediction model to obtain the prediction model of the low-voltage side current of the interval to be repaired / eliminated and the prediction model of the low-voltage side current of the backup power supply interval, which specifically includes: Acquire current data and corresponding time and scene data, then preprocess the acquired data to construct a training data set; The data is clustered by K-means clustering to find out several change patterns of current data at different times and in different scenarios; Constructing a ProbLassoNet model, and training the ProbLassoNet model through a training data set; the ProbLassoNet model adaptively determines the effective features of each change mode according to different change modes of current data, thereby determining a current prediction model corresponding to each change mode; Based on the obtained ProbLassoNet model, prediction models for the low-voltage side current of the interval to be repaired / eliminated and the low-voltage side current of the backup power supply interval are constructed respectively.

3. The method for determining a current protection setting value of a low-voltage distribution network according to claim 2, characterized in that: The time and scene data include weather data, date time period data and new energy power generation situation data.

4. The method for determining a current protection setting value of a low-voltage distribution network according to claim 2, characterized in that: Preprocess the acquired data, including: Find outliers and invalid values ​​in the data and replace them by interpolation; Normalize the data to eliminate the mutual influence of different data magnitudes.

5. The method for determining a current protection setting value of a low-voltage distribution network according to claim 2, characterized in that: The ProbLassoNet model is mainly composed of an input layer, a hidden layer, an output layer, and a residual layer between the input layer and the output layer; wherein the input layer inputs historical current data and other feature data, wherein the other feature data are time and scene data; the hidden layer has a set number of layers; and the output layer is a multi-quantile output.

6. The method for determining a current protection setting value of a low-voltage distribution network according to claim 5, characterized in that: The mapping function of the ProbLassoNet model is: f(X)=Xθ+f ω (X) (1) Where X is the input data; f ω (·) is the mapping function from the input layer to the hidden layer; θ and ω represent the weight parameters of the feedforward neural network and the residual layer; The objective function of the ProbLassoNet model is: Where Y represents the predicted data; L(X, Y; θ, ω) represents the loss value of the model's multi-quantile output; λ||θ||1 represents the l1 regularization of the residual layer parameters, and λ represents the degree of control regularization; is the element of the jth row of the first hidden layer; θ j is the element of the j-th row of the residual layer.

7. The method for determining a current protection setting value of a low-voltage distribution network according to claim 6, characterized in that: The training of the ProbLassoNet model includes the following two steps: First, the initial model is trained by updating θ and ω with the goal of making L(X, Y; θ, ω) as small as possible. Then, θ and ω are updated by the proximal gradient method (1) ; The soft threshold operator is set as: S λ (x)=sign(x)·max{x|-λ,0} (3) Here, sign(x) means that it is 1 when x>0 and -1 when x<0.

8. The method for determining a current protection setting value of a low-voltage distribution network according to claim 1, characterized in that: Match the total load current after closing the loop with the high-voltage side protection setting of the backup power supply interval, and set the high-voltage side protection setting of the backup power supply interval to be greater than the total load current after closing the loop to avoid accident tripping on the low and medium voltage sides.

9. The method for determining a current protection setting value of a low-voltage distribution network according to claim 1, characterized in that: The related current protection constants include: a first-stage overcurrent constant, a second / third-stage overcurrent constant, an acceleration value after the switch is closed, and a leakage current constant.

10. The method for determining a current protection setting value of a low-voltage distribution network according to claim 9, characterized in that: The first overcurrent constant is set to 15 times the predicted current of the interval to be repaired / eliminated, the second / third overcurrent constant is set to 1.5 times the predicted current of the interval to be repaired / eliminated, the acceleration value after the switch is closed is set to 1.5 times the predicted current of the interval to be repaired / eliminated, and the leakage current constant is set to 100mA.

Images