A method for predicting the trip-out rate of a high-voltage transmission line

Abstract

The application provides a high-voltage transmission line trip-out rate prediction method, comprising the following steps: collecting characteristic input parameters of a to-be-predicted area, such as tower parameters, conductor lightning withstand level, impulse grounding resistance and protection angle; inputting the characteristic input parameters of the to-be-predicted area into a BP neural network of a combination optimization algorithm to respectively predict the shielding failure trip-out rate and the back strike trip-out rate of the plain area and the mountainous area; and adjusting according to the neural network to reduce the trip-out rate. The method can be applied to build a high-voltage transmission line lightning protection prediction model based on a 'blocking type' lightning protection method, and can predict the transmission line trip-out rate in the case of line back strike and line shielding failure.

Description

Technical Field

[0001] This invention belongs to the field of power lightning protection, and in particular relates to a method for predicting the tripping rate of high-voltage transmission lines. Background Technology

[0002] With the increasing demand for high-capacity, long-distance power transmission in my country, the protection requirements for high-voltage transmission lines are becoming increasingly stringent. Lightning strikes are one of the main causes of transmission line failures. Therefore, improving the lightning protection performance of transmission lines is of great significance to the safe and stable operation of high-voltage transmission systems.

[0003] Stochastic gradient descent (SGD) and Momentum algorithms, as early neural network training methods, require pre-setting hyperparameters such as the learning rate and momentum coefficient. For different neural networks, the learning rate needs to be manually adjusted to achieve the best training results. Adaptive learning rate algorithms such as Adagrad, RMSprop, and Adam can adjust the learning rate to achieve the best optimization effect. The Adam algorithm, as the most widely used optimization algorithm, has a fast training speed and stable gradient descent. However, in some test sets, the SGD algorithm outperforms the Adam algorithm. Under different geological and topographical influencing factors, the general applicability of a single algorithm is poor, and conducting performance tests on each algorithm individually can significantly reduce prediction efficiency. Summary of the Invention

[0004] In view of this, the present invention aims to propose a method for predicting the tripping rate of high-voltage transmission lines, which can dynamically switch the optimization algorithm according to the decrease of the loss function during the calculation process, and can significantly improve the prediction efficiency.

[0005] To achieve the above objectives, the technical solution of the present invention is implemented as follows:

[0006] A method for predicting the tripping rate of high-voltage transmission lines, characterized by comprising the following steps:

[0007] S1, Statistical tower data and tripping rate are used as the dataset;

[0008] S2. Initialize parameters, calculate the gradient at iteration step t of the Momentum algorithm, update neural network parameters, calculate the loss function, and switch algorithms when k < 0.5;

[0009] S3. Calculate the gradient at the Adam algorithm iteration step t, update the neural network parameters, and calculate the loss function;

[0010] S4. Use the tower data information of the area to be predicted as feature input parameters;

[0011] S5. Input the feature input parameters of the area to be predicted into the BP neural network of the combined optimization algorithm to predict the tripping rate of the plain area and the mountain area respectively. The combined optimization algorithm includes the Momentum algorithm in step S2 and the Adam algorithm in step S3.

[0012] S6. Adjust according to the BP neural network in step S5, such as by adjusting the grounding resistance and ground wire protection angle to reduce the tripping rate.

[0013] Furthermore, in step S4, the tower data information includes tower parameters, conductor lightning withstand level, impulse grounding resistance, and protection angle.

[0014] Furthermore, in step S5, the tripping rate includes the bypass tripping rate and the backflash tripping rate.

[0015] Furthermore, in step S5, the combinatorial optimization algorithm uses the switching coefficient k to determine the inflection point of gradient descent, where k is:

[0016]

[0017] in,

[0018] loss t , representing the error after t iterations;

[0019] loss t+100 , representing the error after t+100 iterations;

[0020] If k < 0.5, switch to the Adam algorithm; otherwise, continue using the Momentum algorithm.

[0021] Furthermore, in step S2, the expression for the Momentum algorithm is as follows:

[0022] m t =βm t-1 +(1-β)g t

[0023] θ t+1 =θ t -ηm t

[0024] in,

[0025] m t , where represents the gradient update direction in iteration t;

[0026] η represents the learning rate;

[0027] θ t , representing the set of weight values ​​and bias terms of the neural network parameters;

[0028] β represents the momentum coefficient;

[0029] g t , where represents the gradient at iteration t.

[0030] After t iterations, with the updated parameters θ t The neural network obtained the predicted value Y t The error after t iterations is calculated using the mean square error function:

[0031]

[0032] in,

[0033] Y represents the actual value;

[0034] Y t , representing the predicted value.

[0035] Furthermore, in step S3, the Adam algorithm expression is as follows:

[0036] m t =β1m t-1 +(1-β1)g t

[0037] v t =β2v t-1 +(1-β2)g t 2

[0038]

[0039]

[0040]

[0041] in,

[0042] β1 and β2 represent the attenuation coefficients;

[0043] m t , which represents the first moment estimate of the gradient;

[0044] v t , which represents the second moment estimate of the gradient;

[0045] Indicates the relationship between m t Correction;

[0046] Indicates to v t Correction;

[0047] ε represents a constant added to maintain numerical stability.

[0048] An electronic device includes a processor and a memory communicatively connected to the processor and used to store processor-executable instructions, the processor being used to execute a method for predicting the tripping rate of a high-voltage transmission line.

[0049] A server includes at least one processor and a memory communicatively connected to the processor, the memory storing instructions executable by the at least one processor, the instructions being executed by the processor to cause the at least one processor to perform a high-voltage transmission line tripping rate prediction method.

[0050] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements a method for predicting the tripping rate of high-voltage transmission lines.

[0051] Compared with existing technologies, the high-voltage transmission line tripping rate prediction method of the present invention has the following advantages:

[0052] (1) The high-voltage transmission line tripping rate prediction method described in this invention uses the Momentum algorithm to quickly converge, and then switches to the Adam algorithm to quickly and stably find the optimal solution.

[0053] (2) The high-voltage transmission line tripping rate prediction method described in this invention has a lower prediction error rate and higher accuracy due to the adaptive combination optimization algorithm of Momentum algorithm and Adam algorithm. Attached Figure Description

[0054] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0055] Figure 1 Flowchart of combinatorial optimization algorithm;

[0056] Figure 2 Accuracy of back-to-back trip rate prediction;

[0057] Figure 3 Accuracy of predicting tripping rate due to circuit breaker failure. Detailed Implementation

[0058] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0059] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0060] The purpose of this invention is to provide a method for predicting the tripping rate of high-voltage transmission lines, which can predict the backflash tripping rate and the backflash tripping rate in mountainous and plain areas.

[0061] The technical solution of this invention is as follows:

[0062] S1: Statistical tower data and tripping rate are used as the dataset;

[0063] S2: Initialize parameters, calculate the gradient at iteration step t of the Momentum algorithm, update neural network parameters, calculate the loss function, and switch algorithms when k < 0.5;

[0064] S3: Calculate the gradient at the Adam algorithm iteration step t, update the neural network parameters, and calculate the loss function;

[0065] S4: Input parameters such as tower parameters, conductor lightning withstand level, impulse grounding resistance and protection angle of the area to be predicted;

[0066] S5: Input the feature input parameters of the area to be predicted into the BP neural network of the combined optimization algorithm to predict the bypass trip rate and back-attack trip rate of the plain area and the mountain area respectively.

[0067] S6: Adjustments are made based on the neural network to reduce the tripping rate.

[0068] Preferably, the expression for the Momentum algorithm is as follows:

[0069] m t =βm t-1 +(1-β)g t

[0070] θ t+1 =θ t -ηm t

[0071] After t iterations, with the updated parameters θ t The neural network obtained the predicted value Y t The error after t iterations is calculated using the mean square error function:

[0072]

[0073] In the formula, Y is the actual value.

[0074] Preferably, a switching coefficient k is set to determine the inflection point of gradient descent:

[0075]

[0076] If k < 0.5, switch to the Adam algorithm; otherwise, continue using the Momentum algorithm.

[0077] Preferably, the Adam algorithm expression is as follows:

[0078] m t =β1m t-1 +(1-β1)g t

[0079] v t =β2v t-1 +(1-β2)g t 2

[0080]

[0081]

[0082]

[0083] This implementation case provides a method for predicting the tripping rate of high-voltage transmission lines, which includes at least step 1 (value 6), and the specific steps are as follows:

[0084] Step 1: Compile tower data and tripping rate as a dataset;

[0085] Specifically, tower data includes tower parameters, conductor lightning withstand level, impulse grounding resistance and protection angle, etc.; tripping rate includes backflashover tripping rate and backflashover tripping rate.

[0086] Step 2: Initialize parameters, calculate the gradient at iteration step t of the Momentum algorithm, update neural network parameters, calculate the loss function, and switch algorithms when k < 0.5;

[0087] Step 3: Calculate the gradient at the Adam algorithm iteration step t, update the neural network parameters, and calculate the loss function;

[0088] Specifically, combinatorial optimization algorithms such as Figure 1 As shown, the optimization algorithm is switched during the iteration process based on the characteristics of gradient descent. The Momentum algorithm is used first, with the following expression:

[0089] m t =βm t-1 +(1-β)g t

[0090] θ t+1 =θ t -ηm t

[0091] After t iterations, with the updated parameters θ t The neural network obtained the predicted value Y t The error after t iterations is calculated using the mean square error function:

[0092]

[0093] In the formula, Y is the actual value. A switching coefficient k is set to determine the inflection point of gradient descent:

[0094]

[0095] When k < 0.5, the Adam algorithm is switched to, as shown in the following expression:

[0096] m t =β1m t-1 +(1-β1)g t

[0097] v t =β2v t-1 +(1-β2)g t 2

[0098]

[0099]

[0100]

[0101] Step 4: Statistically analyze the characteristic input parameters of the tower parameters, conductor lightning withstand level, impulse grounding resistance and protection angle of the area to be predicted;

[0102] Step 5: Input the feature input parameters of the area to be predicted into the BP neural network of the combined optimization algorithm to predict the bypass trip rate and backflash trip rate of the plain area and the mountain area respectively;

[0103] Specifically, the topology of a backpropagation (BP) neural network model consists of an input layer, hidden layers, and an output layer. Feature values ​​are processed through linear operations to serve as input to the hidden layers, and the outputs of the hidden layers are processed through linear operations to serve as input to the output layers. The inputs to the output layers are then processed by an activation function to obtain predicted values. By continuously adjusting the weights and biases, the model approximates the training objective. Through multiple iterations, the error between the neural network and the objective function is continuously reduced until an acceptable error range is reached, or the predetermined number of iterations is completed.

[0104] Step 6: Adjust according to the neural network to reduce the tripping rate.

[0105] Specifically, the backflashover tripping rate or backflashover tripping rate of a certain transmission line is statistically analyzed. Based on the BP neural network, parameters such as grounding resistance and ground wire protection angle are adjusted to predict the backflashover tripping rate or backflashover tripping rate, thereby reducing the tripping rate.

[0106] Those skilled in the art will recognize that the units and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0107] In the several embodiments provided in this application, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the division of units described above is merely a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. The aforementioned units may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention according to actual needs.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. 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 or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

[0109] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting the tripping rate of high-voltage transmission lines, characterized in that, Includes the following steps: S1, Statistical tower data and tripping rate are used as the dataset; S2. Initialize parameters, calculate the gradient at iteration step t of the Momentum algorithm, update neural network parameters, calculate the loss function, and switch algorithms when k < 0.5; S3. Calculate the gradient at the Adam algorithm iteration step t, update the neural network parameters, and calculate the loss function; S4. Use the tower data information of the area to be predicted as feature input parameters; S5. Input the feature input parameters of the area to be predicted into the BP neural network of the combined optimization algorithm to predict the tripping rate of the plain area and the mountain area respectively. The combined optimization algorithm includes the Momentum algorithm in step S2 and the Adam algorithm in step S3. S6. Adjust according to the BP neural network in step S5 to reduce the tripping rate; The combinatorial optimization algorithm uses a switching coefficient k to determine the inflection point of gradient descent, where k is: ； in, , representing the error after t iterations; , representing the error after t+100 iterations; If k < 0.5, switch to the Adam algorithm; otherwise, continue using the Momentum algorithm.

2. The method for predicting the tripping rate of high-voltage transmission lines according to claim 1, characterized in that: In step S4, the tower data information includes tower parameters, conductor lightning withstand level, impulse grounding resistance, and protection angle.

3. The method for predicting the tripping rate of high-voltage transmission lines according to claim 1, characterized in that: In step S5, the tripping rate includes the bypass tripping rate and the backflash tripping rate.

4. The method for predicting the tripping rate of high-voltage transmission lines according to claim 1, characterized in that: In step S2, the expression for the Momentum algorithm is as follows: in, , indicating the gradient update direction in iteration t; , representing the learning rate; , representing the set of weight values ​​and bias terms of the neural network parameters; , represents the momentum coefficient; , representing the gradient at iteration t; After t iterations, with the updated parameters The predicted value obtained by the neural network is The error after t iterations is calculated using the mean square error function: in, Y represents the actual value; , representing the predicted value.

5. The method for predicting the tripping rate of high-voltage transmission lines according to claim 4, characterized in that: In step S3, the Adam algorithm expression is as follows: in, , , represents the attenuation coefficient; , which represents the first moment estimate of the gradient; , which represents the second moment estimate of the gradient; , indicating to Correction; , indicating to Correction; , which represents a constant added to maintain numerical stability.

6. An electronic device, comprising a processor and a memory communicatively connected to the processor and used for storing processor-executable instructions, characterized in that: The processor is used to execute the high-voltage transmission line tripping rate prediction method according to any one of claims 1-5.

7. A server, characterized in that: The device includes at least one processor and a memory communicatively connected to the processor, the memory storing instructions executable by the at least one processor, the instructions being executed by the processor to cause the at least one processor to perform a high-voltage transmission line tripping rate prediction method as described in any one of claims 1-5.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, it implements the tripping rate prediction method for high-voltage transmission lines as described in any one of claims 1-5.

Images