Power transmission line icing galloping prediction method based on Bi-TCN-LSTM network model

Abstract

The invention discloses a power transmission line icing galloping prediction method based on a Bi-TCN-LSTM network model, and the method comprises the steps: generating a meteorological data set according to historical weather forecast data, the historical weather forecast data including but not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, and the weather conditions including sunny days, rainy days and snowy days; matching and marking the historical icing galloping information of the power transmission line with the meteorological data set to obtain an icing galloping data set; performing feature extraction on the icing galloping data set according to a time sequence convolutional network model to obtain an icing galloping feature data set; cascading the bidirectional long short-term memory network model and the time sequence convolutional network model, constructing a Bi-TCN-LSTM network model, and training the Bi-TCN-LSTM network model according to the icing galloping feature data set; and inputting the target weather forecast data into the trained Bi-TCN-LSTM network model, and predicting icing galloping information of the power transmission line. According to the method, the TCN model and the Bi-LSTM model are combined, the long-term and short-term dependency relationship in the time series data is effectively captured, and the prediction precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid security risk assessment, and in particular to a method and device for predicting ice dancing of transmission lines based on a Bi-TCN-LSTM network model. Background Art

[0002] my country has a vast territory and changeable weather. Every year from winter to early spring, the north-south cold and warm air currents converge, causing severe icing on power transmission lines in the northwest, southwest and north China. Transmission line icing is a natural phenomenon caused by factors such as climatic conditions and geographical environment. The main reason is that when the temperature drops, the water vapor in the air condenses into ice and adheres to the conductors, lightning rods, insulator strings and other components of the transmission line to form ice layers or ice chips. Icing has an adverse effect on the mechanical and electrical properties of the transmission line. In windy weather, iced transmission lines are likely to dance, leading to frequent accidents. Therefore, it is of great significance to predict the dancing of icing on transmission lines.

[0003] At present, domestic and foreign scholars have conducted many studies on the prediction of icing conditions of power transmission lines under rime and rime weather, mainly including two categories of methods: statistical analysis and intelligent evaluation. However, there are relatively few studies on the dancing of power transmission lines, which mainly use sensor monitoring and image recognition methods, considering a single factor. The icing phenomenon is closely related to the body coefficient of the transmission line and various meteorological factors such as temperature, humidity, wind speed, and rainfall. It is a complex meteorological physical process. Accurately predicting the risk of ice dancing is an urgent problem to be solved. Summary of the invention

[0004] In view of the above problems, the present invention provides a transmission line ice dancing prediction method based on the Bi-TCN-LSTM network model to accurately predict the ice dancing risk.

[0005] According to one aspect of the present invention, a method for predicting ice dancing of transmission lines based on a Bi-TCN-LSTM network model is provided, comprising:

[0006] Generate a meteorological data set according to historical meteorological forecast data, wherein the historical meteorological forecast data includes but is not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, and the weather conditions include but are not limited to sunny days, rainy days and snowy days;

[0007] Matching and labeling the historical ice dancing information of the transmission line with the meteorological data set to obtain an ice dancing data set;

[0008] Extracting features of the ice-covered dancing data set according to a temporal convolutional network model to obtain an ice-covered dancing feature data set;

[0009] Cascading a bidirectional long short-term memory network model and the temporal convolutional network model to construct a Bi-TCN-LSTM network model, and training the Bi-TCN-LSTM network model according to the ice-covered dancing feature dataset;

[0010] The target weather forecast data is input into the trained Bi-TCN-LSTM network model to predict the ice dancing information of the transmission lines.

[0011] In an optional manner, before generating the meteorological data set according to the historical meteorological forecast data, the method further comprises: performing outlier processing and normalization processing on the historical meteorological forecast data;

[0012] Wherein, the outlier processing further includes:

[0013] Step S11, performing long time series segmentation on the historical weather forecast data to obtain segmented time series data;

[0014] Step S12, performing anomaly detection on the segmented time series data according to a preset time series, if the meteorological data is always zero, it is determined to be abnormal and the meteorological data is deleted; if the meteorological data is missing only at a certain sampling point, the value of the sampling point is replaced according to the average sampling value of the preset range of the sampling point;

[0015] The normalization process further includes:

[0016] Step S13, normalizing the meteorological data without abnormality in step S12, wherein the calculation formula for the normalization processing is:

[0017]

[0018] Among them, X n represents the normalized meteorological data, X represents the initial data of the meteorological dataset, and X min Represents the smallest data in the meteorological data set, X max Represents the largest data in the meteorological dataset.

[0019] In an optional manner, matching and labeling the historical ice dancing information of the transmission line with the meteorological data set to obtain the ice dancing data set further includes:

[0020] Step S21: extracting corresponding meteorological parameters from the meteorological data set according to the time and location of each ice covering and dancing event; wherein the meteorological parameters include temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions;

[0021] Step S22: Match and annotate each historical ice cover and dancing event with the corresponding meteorological data to obtain a feature label data set; wherein the annotated content includes time, location, ice cover thickness and dancing amplitude.

[0022] In an optional manner, the temporal convolutional network model includes a one-dimensional full convolution, a causal convolution, a residual connection module and a hole convolution module.

[0023] In an optional manner, the bidirectional long short-term memory network model includes an input gate, a forget gate, and an output gate;

[0024] The input gate determines the extent to which new information will be written into the cell state, and is used to selectively accept input data at the current moment and hidden state information at the previous moment;

[0025] The forget gate determines the information in the cell state that needs to be retained or forgotten, and is used to dynamically adjust the length of memory, discard unimportant historical information, and retain key information related to prediction;

[0026] The output gate determines the part of the cell state that affects the hidden state, thereby participating in the calculation of the next moment, and is used to selectively output information that is helpful for predicting subsequent moments based on the input data at the current moment and the updated cell state.

[0027] In an optional manner, the structure of the input gate is:

[0028] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0029]

[0030] Among them, i t is the activation value of the input gate, which controls the degree of passing of meteorological information at the current moment; represents a new candidate cell state; [h t-1 ,x t ] represents the hidden state of the previous moment and the meteorological data input at the current moment; W i , W C 、b C 、b i represents weight and bias parameters;

[0031] The structure of the forget gate is:

[0032] f t =σ(W f ·[h t-1,x t ]+b f )

[0033] Among them, f t Represents the activation value of the forget gate. The forget gate is opened only when the activation value is met. f and b f are weight and bias parameters;

[0034] The structure of the output gate is:

[0035] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0036] h t =o t tanh(C t )

[0037] Among them, t Represents the activation value of the output gate, controlling the output degree, h t represents the hidden state at the current moment, W o , b o are weight and bias parameters.

[0038] In an optional manner, the method further includes:

[0039] Threshold conditions are set according to the precipitation, temperature, humidity and wind field when the transmission line galloping event occurs, so as to screen the transmission line galloping events that meet the threshold conditions, wherein the threshold conditions are:

[0040]

[0041] Among them, P is precipitation, T is temperature, S is humidity, and W is wind field.

[0042] In an optional manner, after predicting the ice dancing information of the transmission line, the method further includes:

[0043] Establishing wind excitation model judgment conditions to perform risk rating on predicted ice dancing information; the wind excitation model judgment conditions include first-level judgment conditions, second-level judgment conditions, third-level judgment conditions and fourth-level judgment conditions;

[0044] Wherein, the first-level determination condition is:

[0045] The accumulated time of temperature between -5 and 0°C and humidity > 90% in the past 24 hours and the next 24 hours is ≥ 5 hours, and the wind speed is within the range of 8 ≤ V ≤ 20 m / s for a duration of more than 10 hours;

[0046] The secondary determination conditions are:

[0047] The accumulated time of temperature between -7 and 1°C and humidity > 80% in the past 24 hours and the next 24 hours is ≥ 5 hours, and the wind speed is within the range of 8 ≤ V ≤ 20 m / s for a duration of more than 4 hours;

[0048] The three-level determination conditions are:

[0049] The accumulated time of temperature between -9 and 1°C and humidity > 70% in the past 24 hours and the next 24 hours is ≥ 4 hours, and the wind speed is within the range of 4 ≤ V ≤ 20 m / s for a duration of more than 2 hours;

[0050] The four-level determination conditions are:

[0051] The cumulative time in the past 24 hours and the next 24 hours that the temperature is between -9 and 1°C and the humidity is greater than 70% is ≥ 4 hours, and the wind speed is within the range of 4≤V≤20m / s for a duration of less than or equal to 2 hours.

[0052] In an optional manner, the risk rating of the predicted ice dancing information further includes:

[0053] Step S441, when the wind excitation judgment condition meets the first-level judgment condition, the galloping risk level is output as level one, otherwise, it goes to step S442;

[0054] Step S442, when the wind excitation judgment condition meets the second-level judgment condition, the galloping risk level is output as level 2, otherwise, the process goes to step S443;

[0055] Step S443, when the wind excitation judgment condition meets the third-level judgment condition, the galloping risk level is output as level three, otherwise, proceed to step S444;

[0056] Step S444: when the wind excitation judgment condition meets the fourth-level judgment condition, the dancing risk level is output as fourth-level, otherwise the icing risk level is output as no risk, and the step ends;

[0057] Step S445, judging in the past 24 hours and the future 24 hours, if the accumulated time of temperature between -5 and 0°C and humidity greater than 90% is ≥ 10 hours, the time of > 0°C is < 3 hours, and the wind excitation judgment condition meets the first-level judgment condition, the galloping risk level is output as level 1, otherwise, proceeding to step S336;

[0058] Step S446, if the accumulated time of the temperature between -7 and 1°C and the humidity greater than 80% in the past 24 hours and the future 24 hours is ≥10 hours, the time of >1°C during the period is <3 hours, and the wind excitation judgment condition meets the second-level judgment condition, the galloping risk level is output as the second level, otherwise, go to step S447;

[0059] Step S447, judging the past 24 hours and the future 24 hours, if the cumulative time of temperature between -9 and -5°C and humidity greater than 70% is ≥10 hours, or the cumulative time of temperature between -5 and 0°C and humidity greater than 70% is ≥5 hours, and the time of >1°C is <3 hours, and the wind excitation judgment condition meets the third-level judgment condition, the galloping risk level is output as the third level, otherwise, proceed to step S448;

[0060] Step S448, judge the past 24 hours and the future 24 hours, if the cumulative time of temperature being between -9 and -5°C and humidity being greater than 70% is ≥8 hours, or the cumulative time of temperature being between -5 and 0°C and humidity being greater than 70% is ≥5 hours, and the time of >1°C during the period is <5 hours, and the wind excitation judgment condition meets the four-level judgment condition, the dancing risk level output is level four, otherwise the dancing risk level output is no risk, and the step ends.

[0061] According to another aspect of the present invention, a transmission line ice dancing prediction device based on a Bi-TCN-LSTM network model is provided, comprising:

[0062] A data acquisition module, used to generate a meteorological data set based on historical meteorological forecast data, wherein the historical meteorological forecast data includes but is not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, wherein the weather conditions include but are not limited to sunny days, rainy days and snowy days;

[0063] A data annotation module, used for matching and annotating the historical ice dancing information of the transmission line with the meteorological data set to obtain an ice dancing data set;

[0064] A feature extraction module, used for extracting features from the ice-covered dancing data set according to a temporal convolutional network model to obtain an ice-covered dancing feature data set;

[0065] A model building module, used to cascade a bidirectional long short-term memory network model and the temporal convolutional network model to build a Bi-TCN-LSTM network model, and train the Bi-TCN-LSTM network model according to the ice-covered dancing feature data set;

[0066] The prediction module is used to input the target meteorological forecast data into the trained Bi-TCN-LSTM network model to predict the ice dancing information of the transmission line.

[0067] According to the solution provided by the present invention, a meteorological data set is generated according to historical meteorological forecast data, wherein the historical meteorological forecast data includes but is not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, and the weather conditions include but are not limited to sunny days, rainy days and snowy days; the historical ice-dancing information of the transmission line is matched and labeled with the meteorological data set to obtain an ice-dancing data set; the ice-dancing data set is feature extracted according to a temporal convolutional network model to obtain an ice-dancing feature data set; a bidirectional long short-term memory network model and the temporal convolutional network model are cascaded to construct a Bi-TCN-LSTM network model, and the Bi-TCN-LSTM network model is trained according to the ice-dancing feature data set; the target meteorological forecast data is input into the trained Bi-TCN-LSTM network model to predict the ice-dancing information of the transmission line. The present invention combines the weather forecast information of the meteorological department, fully considers the strong correlation factors of the ice dancing of transmission lines in severe weather, and has certain effectiveness and rationality. By combining the TCN model and the Bi-LSTM model, it effectively captures the long-term and short-term dependencies in the time series data, improves the prediction accuracy, ensures the authority and reliability of the data, and contributes to the maintenance and safety management of the power system.

[0068] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0070] Figure 1 A schematic diagram of a method for predicting ice dancing of a transmission line based on a Bi-TCN-LSTM network model according to an embodiment of the present invention is shown;

[0071] Figure 2 A schematic diagram of the structure of a Bi-TCN-LSTM network model according to an embodiment of the present invention is shown;

[0072] Figure 3 The framework structure diagram of the transmission line ice dancing prediction device based on the Bi-TCN-LSTM network model in an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0073] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present invention can be combined with each other without conflict.

[0074] The following detailed description is an exemplary description, which is intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present invention belongs. The terms used in the present invention are only for describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present invention.

[0075] Figure 1 A schematic diagram of a method for predicting ice dancing of a transmission line based on a Bi-TCN-LSTM network model according to an embodiment of the present invention is shown, comprising the following steps:

[0076] Step S1, generating a meteorological data set based on historical meteorological forecast data, wherein the historical meteorological forecast data includes but is not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, and the weather conditions include but are not limited to sunny days, rainy days and snowy days.

[0077] For example, we obtained hourly observation data provided by the National Meteorological Station in the past 10 years, and formed a meteorological data set of ice-dancing influencing factors based on the observation information such as hourly temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions of local meteorological stations during historical transmission line dancing events.

[0078] In this embodiment, before generating a meteorological data set based on historical meteorological forecast data, outlier processing and normalization processing are performed on the historical meteorological forecast data;

[0079] Wherein, the outlier processing further includes:

[0080] Step S11, performing long time series segmentation on the historical weather forecast data to obtain segmented time series data;

[0081] Step S12, performing anomaly detection on the segmented time series data according to a preset time series, if the meteorological data is always zero, it is determined to be abnormal and the meteorological data is deleted; if the meteorological data is missing only at a certain sampling point, the value of the sampling point is replaced according to the average sampling value of the preset range of the sampling point;

[0082] The normalization process further includes:

[0083] Step S13, normalizing the meteorological data without abnormality in step S12, wherein the calculation formula for the normalization processing is:

[0084]

[0085] Among them, X n represents the normalized meteorological data, X represents the initial data of the meteorological dataset, and X min Represents the smallest data in the meteorological data set, X max Represents the largest data in the meteorological dataset.

[0086] Step S2, matching and labeling the historical ice dancing information of the transmission line with the meteorological data set to obtain an ice dancing data set.

[0087] For example, by combining the historical ice dance statistics provided by the power sector, the meteorological data is matched and labeled with historical events to form a complete ice dance data set.

[0088] Specifically, the steps include:

[0089] Step S21: extracting corresponding meteorological parameters from the meteorological data set according to the time and location of each ice covering and dancing event; wherein the meteorological parameters include temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions;

[0090] Step S22: Match and label each historical ice and dancing event with the corresponding meteorological data to obtain a feature label data set (feature-label pair); wherein the labeled content includes time, location, ice thickness and dancing amplitude.

[0091] The integrated features and label data are combined into a structured data set to ensure that each record contains complete features and corresponding labels.

[0092] Step S3, extracting features of the ice-covered dancing data set according to a temporal convolutional network model to obtain an ice-covered dancing feature data set.

[0093] In this embodiment, the temporal convolutional network model includes one-dimensional full convolution, causal convolution, residual connection module and void convolution module. Features are extracted according to the temporal convolutional network (TCN) model, and the features are combined with the original data to generate a feature data set.

[0094] Step S4, cascading the bidirectional long short-term memory network model and the temporal convolutional network model to construct a Bi-TCN-LSTM network model, and training the Bi-TCN-LSTM network model according to the ice-covered dancing feature data set.

[0095] like Figure 2As shown in the figure, the long short-term memory network is cascaded with the TCN network to build a Bi-TCN-LSTM network model, and the LSTM neural network is used to train the forward propagation and the LSTM neural network is used to train the back propagation to predict the ice dancing of the transmission line.

[0096] In this embodiment, the bidirectional long short-term memory network model includes an input gate, a forget gate and an output gate;

[0097] The input gate determines the extent to which new information will be written into the cell state, and is used to selectively accept input data at the current moment and hidden state information at the previous moment;

[0098] The forget gate determines the information in the cell state that needs to be retained or forgotten, and is used to dynamically adjust the length of memory, discard unimportant historical information, and retain key information related to prediction;

[0099] The output gate determines the part of the cell state that affects the hidden state, thereby participating in the calculation of the next moment, and is used to selectively output information that is helpful for predicting subsequent moments based on the input data at the current moment and the updated cell state.

[0100] In an optional manner, the structure of the input gate is:

[0101] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0102]

[0103] Among them, i t is the activation value of the input gate, which controls the degree of passing of meteorological information at the current moment; represents a new candidate cell state; [h t-1 ,x t ] represents the hidden state of the previous moment and the meteorological data input at the current moment; W i , W C 、b C 、b i represents weight and bias parameters;

[0104] The structure of the forget gate is:

[0105] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0106] Among them, f t Represents the activation value of the forget gate. The forget gate is opened only when the activation value is met. f and b f are weight and bias parameters;

[0107] The structure of the output gate is:

[0108] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0109] h t =o t tanh(C t )

[0110] Among them, t Represents the activation value of the output gate, controlling the output degree, h t represents the hidden state at the current moment, W o , b o are weight and bias parameters.

[0111] Step S5, input the target weather forecast data into the trained Bi-TCN-LSTM network model to predict the ice dancing information of the transmission line.

[0112] Optionally, the prediction results of the ice dancing information of the transmission line are evaluated, and the root mean square error RMSE, the mean absolute error MAE and the goodness of fit R2 are used as evaluation indicators.

[0113]

[0114] Among them, m represents the total capacity of the ice dancing feature training set, and f i Represents the predicted value of the Bi-TCN-LSTM network, y i Indicates the actual test value, Indicates that the actual value is averaged.

[0115] In an optional manner, the method further includes:

[0116] Threshold conditions are set according to the precipitation, temperature, humidity and wind field when the transmission line galloping event occurs, so as to screen the transmission line galloping events that meet the threshold conditions, wherein the threshold conditions are:

[0117]

[0118] Among them, P is precipitation, T is temperature, S is humidity, and W is wind field.

[0119] In an optional manner, after predicting the ice dancing information of the transmission line, the method further includes:

[0120] Establishing wind excitation model judgment conditions to perform risk rating on predicted ice dancing information; the wind excitation model judgment conditions include first-level judgment conditions, second-level judgment conditions, third-level judgment conditions and fourth-level judgment conditions;

[0121] Wherein, the first-level determination condition is:

[0122] The accumulated time of temperature between -5 and 0°C and humidity > 90% in the past 24 hours and the next 24 hours is ≥ 5 hours, and the wind speed is within the range of 8 ≤ V ≤ 20 m / s for a duration of more than 10 hours;

[0123] The secondary determination conditions are:

[0124] The accumulated time of temperature between -7 and 1°C and humidity > 80% in the past 24 hours and the next 24 hours is ≥ 5 hours, and the wind speed is within the range of 8 ≤ V ≤ 20 m / s for a duration of more than 4 hours;

[0125] The three-level determination conditions are:

[0126] The accumulated time of temperature between -9 and 1°C and humidity > 70% in the past 24 hours and the next 24 hours is ≥ 4 hours, and the wind speed is within the range of 4 ≤ V ≤ 20 m / s for a duration of more than 2 hours;

[0127] The four-level determination conditions are:

[0128] The cumulative time in the past 24 hours and the next 24 hours that the temperature is between -9 and 1°C and the humidity is greater than 70% is ≥ 4 hours, and the wind speed is within the range of 4≤V≤20m / s for a duration of less than or equal to 2 hours.

[0129] In an optional manner, the risk rating of the predicted ice dancing information further includes:

[0130] Step S441, when the wind excitation judgment condition meets the first-level judgment condition, the galloping risk level is output as level one, otherwise, it goes to step S442;

[0131] Step S442, when the wind excitation judgment condition meets the second-level judgment condition, the galloping risk level is output as level 2, otherwise, the process goes to step S443;

[0132] Step S443, when the wind excitation judgment condition meets the third-level judgment condition, the galloping risk level is output as level three, otherwise, proceed to step S444;

[0133] Step S444: when the wind excitation judgment condition meets the fourth-level judgment condition, the dancing risk level is output as fourth-level, otherwise the icing risk level is output as no risk, and the step ends;

[0134] Step S445, judging in the past 24 hours and the future 24 hours, if the accumulated time of temperature between -5 and 0°C and humidity greater than 90% is ≥ 10 hours, the time of > 0°C is < 3 hours, and the wind excitation judgment condition meets the first-level judgment condition, the galloping risk level is output as level 1, otherwise, proceeding to step S336;

[0135] Step S446, if the accumulated time of the temperature between -7 and 1°C and the humidity greater than 80% in the past 24 hours and the future 24 hours is ≥10 hours, the time of >1°C during the period is <3 hours, and the wind excitation judgment condition meets the second-level judgment condition, the galloping risk level is output as the second level, otherwise, go to step S447;

[0136] Step S447, judging the past 24 hours and the future 24 hours, if the cumulative time of temperature between -9 and -5°C and humidity greater than 70% is ≥10 hours, or the cumulative time of temperature between -5 and 0°C and humidity greater than 70% is ≥5 hours, and the time of >1°C is <3 hours, and the wind excitation judgment condition meets the third-level judgment condition, the galloping risk level is output as the third level, otherwise, proceed to step S448;

[0137] Step S448, judge the past 24 hours and the future 24 hours, if the cumulative time of temperature being between -9 and -5°C and humidity being greater than 70% is ≥8 hours, or the cumulative time of temperature being between -5 and 0°C and humidity being greater than 70% is ≥5 hours, and the time of >1°C during the period is <5 hours, and the wind excitation judgment condition meets the four-level judgment condition, the dancing risk level output is level four, otherwise the dancing risk level output is no risk, and the step ends.

[0138] According to the solution provided by the present invention, a meteorological data set is generated according to historical meteorological forecast data, wherein the historical meteorological forecast data includes but is not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, and the weather conditions include but are not limited to sunny days, rainy days and snowy days; the historical ice-dancing information of the transmission line is matched and labeled with the meteorological data set to obtain an ice-dancing data set; the ice-dancing data set is feature extracted according to a temporal convolutional network model to obtain an ice-dancing feature data set; a bidirectional long short-term memory network model and the temporal convolutional network model are cascaded to construct a Bi-TCN-LSTM network model, and the Bi-TCN-LSTM network model is trained according to the ice-dancing feature data set; the target meteorological forecast data is input into the trained Bi-TCN-LSTM network model to predict the ice-dancing information of the transmission line. The present invention combines the weather forecast information of the meteorological department, fully considers the strong correlation factors of the ice dancing of transmission lines in severe weather, and has certain effectiveness and rationality. By combining the TCN model and the Bi-LSTM model, it effectively captures the long-term and short-term dependencies in the time series data, improves the prediction accuracy, ensures the authority and reliability of the data, and contributes to the maintenance and safety management of the power system.

[0139] Figure 2 The framework structure diagram of the transmission line ice dancing prediction device based on the Bi-TCN-LSTM network model according to an embodiment of the present invention is shown, including:

[0140] A data acquisition module, used to generate a meteorological data set based on historical meteorological forecast data, wherein the historical meteorological forecast data includes but is not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, wherein the weather conditions include but are not limited to sunny days, rainy days and snowy days;

[0141] A data annotation module, used for matching and annotating the historical ice dancing information of the transmission line with the meteorological data set to obtain an ice dancing data set;

[0142] A feature extraction module, used for extracting features from the ice-covered dancing data set according to a temporal convolutional network model to obtain an ice-covered dancing feature data set;

[0143] A model building module, used to cascade a bidirectional long short-term memory network model and the temporal convolutional network model to build a Bi-TCN-LSTM network model, and train the Bi-TCN-LSTM network model according to the ice-covered dancing feature data set;

[0144] The prediction module is used to input the target meteorological forecast data into the trained Bi-TCN-LSTM network model to predict the ice dancing information of the transmission line.

[0145] According to the solution provided by the present invention, a meteorological data set is generated according to historical meteorological forecast data, wherein the historical meteorological forecast data includes but is not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, and the weather conditions include but are not limited to sunny days, rainy days and snowy days; the historical ice-dancing information of the transmission line is matched and labeled with the meteorological data set to obtain an ice-dancing data set; the ice-dancing data set is feature extracted according to a temporal convolutional network model to obtain an ice-dancing feature data set; a bidirectional long short-term memory network model and the temporal convolutional network model are cascaded to construct a Bi-TCN-LSTM network model, and the Bi-TCN-LSTM network model is trained according to the ice-dancing feature data set; the target meteorological forecast data is input into the trained Bi-TCN-LSTM network model to predict the ice-dancing information of the transmission line. The present invention combines the weather forecast information of the meteorological department, fully considers the strong correlation factors of the ice dancing of transmission lines in severe weather, and has certain effectiveness and rationality. By combining the TCN model and the Bi-LSTM model, it effectively captures the long-term and short-term dependencies in the time series data, improves the prediction accuracy, ensures the authority and reliability of the data, and contributes to the maintenance and safety management of the power system.

[0146] It is known from common technical knowledge that the present invention can be implemented by other embodiments that do not deviate from its spirit or essential features. Therefore, the above disclosed embodiments are only illustrative in all respects and are not exclusive. All changes within the scope of the present invention or within the scope equivalent to the present invention are included in the present invention.

[0147] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0148] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks 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 produce 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.

[0149] 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.

[0150] 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.

[0151] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting ice dancing of transmission lines based on a Bi-TCN-LSTM network model, characterized in that: include: Generate a meteorological data set according to historical meteorological forecast data, wherein the historical meteorological forecast data includes but is not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, and the weather conditions include but are not limited to sunny days, rainy days and snowy days; Matching and labeling the historical ice dancing information of the transmission line with the meteorological data set to obtain an ice dancing data set; Extracting features of the ice-covered dancing data set according to a temporal convolutional network model to obtain an ice-covered dancing feature data set; Cascading a bidirectional long short-term memory network model and the temporal convolutional network model to construct a Bi-TCN-LSTM network model, and training the Bi-TCN-LSTM network model according to the ice-covered dancing feature dataset; The target weather forecast data is input into the trained Bi-TCN-LSTM network model to predict the ice dancing information of the transmission lines.

2. The method for predicting ice dancing of transmission lines based on the Bi-TCN-LSTM network model according to claim 1 is characterized in that: Before generating the meteorological data set according to the historical meteorological forecast data, the method further comprises: performing outlier processing and normalization processing on the historical meteorological forecast data; Wherein, the outlier processing further includes: Step S11, performing long time series segmentation on the historical weather forecast data to obtain segmented time series data; Step S12, performing anomaly detection on the segmented time series data according to a preset time series, if the meteorological data is always zero, it is determined to be abnormal and the meteorological data is deleted; if the meteorological data is missing only at a certain sampling point, the value of the sampling point is replaced according to the average sampling value of the preset range of the sampling point; The normalization process further includes: Step S13, normalizing the meteorological data without abnormality in step S12, wherein the calculation formula for normalization is: Among them, X n represents the normalized meteorological data, X represents the initial data of the meteorological dataset, and X min Represents the smallest data in the meteorological data set, X max Represents the largest data in the meteorological dataset.

3. The method for predicting ice dancing of transmission lines based on the Bi-TCN-LSTM network model according to claim 1 or 2, characterized in that: The matching and labeling of the historical ice dancing information of the transmission line with the meteorological data set to obtain the ice dancing data set further includes: Step S21: extracting corresponding meteorological parameters from the meteorological data set according to the time and location of each ice covering and dancing event; wherein the meteorological parameters include temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions; Step S22: Match and annotate each historical ice cover and dancing event with the corresponding meteorological data to obtain a feature label data set; wherein the annotated content includes time, location, ice cover thickness and dancing amplitude.

4. The method for predicting ice dancing of transmission lines based on the Bi-TCN-LSTM network model according to claim 1 is characterized in that: The temporal convolutional network model includes a one-dimensional full convolution, a causal convolution, a residual connection module and a hole convolution module.

5. The method for predicting ice dancing of transmission lines based on the Bi-TCN-LSTM network model according to claim 1 is characterized in that: The bidirectional long short-term memory network model includes an input gate, a forget gate and an output gate; The input gate determines the extent to which new information will be written into the cell state, and is used to selectively accept input data at the current moment and hidden state information at the previous moment; The forget gate determines the information in the cell state that needs to be retained or forgotten, and is used to dynamically adjust the length of memory, discard unimportant historical information, and retain key information related to prediction; The output gate determines the part of the cell state that affects the hidden state, thereby participating in the calculation of the next moment, and is used to selectively output information that is helpful for predicting subsequent moments based on the input data at the current moment and the updated cell state.

6. The method for predicting ice dancing of transmission lines based on the Bi-TCN-LSTM network model according to claim 5 is characterized in that: The structure of the input gate is: i t =σ(W i ·[h t-1 ,x t ]+b i ); Among them, i t is the activation value of the input gate, which controls the degree of passing of meteorological information at the current moment; represents a new candidate cell state; [h t-1 ,x t ] represents the hidden state of the previous moment and the meteorological data input at the current moment; W i , W C 、b C 、b i represents weight and bias parameters; The structure of the forget gate is: f t =σ(W f ·[h t-1 ,x t ]+b f ) Among them, f t Represents the activation value of the forget gate. The forget gate is opened only when the activation value is met. f and b f are weight and bias parameters; The structure of the output gate is: the t =σ(W o ·[h t-1 ,x t ]+b o ) h t =o t ·tanh(C t ) Among them, t Represents the activation value of the output gate, controlling the output degree, h t represents the hidden state at the current moment, W o , b o are weight and bias parameters.

7. The method for predicting ice dancing of transmission lines based on the Bi-TCN-LSTM network model according to claim 1 is characterized in that: The method further comprises: Threshold conditions are set according to the precipitation, temperature, humidity and wind field when the transmission line galloping event occurs, so as to screen the transmission line galloping events that meet the threshold conditions, wherein the threshold conditions are: Among them, P is precipitation, T is temperature, S is humidity, and W is wind field.

8. The method for predicting ice dancing of transmission lines based on the Bi-TCN-LSTM network model according to claim 1 is characterized in that: After predicting the ice dancing information of the transmission line, the method further includes: Establishing wind excitation model judgment conditions to perform risk rating on predicted ice dancing information; the wind excitation model judgment conditions include first-level judgment conditions, second-level judgment conditions, third-level judgment conditions and fourth-level judgment conditions; Wherein, the first-level determination condition is: The accumulated time of temperature between -5 and 0°C and humidity > 90% in the past 24 hours and the next 24 hours is ≥ 5 hours, and the wind speed is within the range of 8 ≤ V ≤ 20 m / s for a duration of more than 10 hours; The secondary determination conditions are: The accumulated time of temperature between -7 and 1°C and humidity > 80% in the past 24 hours and the next 24 hours is ≥ 5 hours, and the wind speed is within the range of 8 ≤ V ≤ 20 m / s for a duration of more than 4 hours; The three-level determination conditions are: The accumulated time of temperature between -9 and 1°C and humidity > 70% in the past 24 hours and the next 24 hours is ≥ 4 hours, and the wind speed is within the range of 4 ≤ V ≤ 20 m / s for a duration of more than 2 hours; The four-level determination conditions are: The cumulative time in the past 24 hours and the next 24 hours that the temperature is between -9 and 1°C and the humidity is greater than 70% is ≥ 4 hours, and the wind speed is within the range of 4≤V≤20m / s for a duration of less than or equal to 2 hours.

9. The method for predicting ice dancing of transmission lines based on the Bi-TCN-LSTM network model according to claim 8 is characterized in that: The risk rating of the predicted ice dancing information further includes: Step S441, when the wind excitation judgment condition meets the first-level judgment condition, the galloping risk level is output as level one, otherwise, it goes to step S442; Step S442, when the wind excitation judgment condition meets the second-level judgment condition, the galloping risk level is output as level 2, otherwise, the process goes to step S443; Step S443, when the wind excitation judgment condition meets the third-level judgment condition, the galloping risk level is output as level three, otherwise, proceed to step S444; Step S444: when the wind excitation judgment condition meets the fourth-level judgment condition, the dancing risk level is output as fourth-level, otherwise the icing risk level is output as no risk, and the step ends; Step S445, judging in the past 24 hours and the future 24 hours, if the accumulated time of temperature between -5 and 0°C and humidity greater than 90% is ≥ 10 hours, the time of > 0°C is < 3 hours, and the wind excitation judgment condition meets the first-level judgment condition, the galloping risk level is output as level 1, otherwise, proceeding to step S336; Step S446, if the accumulated time of the temperature between -7 and 1°C and the humidity greater than 80% in the past 24 hours and the future 24 hours is ≥10 hours, the time of >1°C during the period is <3 hours, and the wind excitation judgment condition meets the second-level judgment condition, the galloping risk level is output as the second level, otherwise, go to step S447; Step S447, judging the past 24 hours and the future 24 hours, if the cumulative time of temperature between -9 and -5°C and humidity greater than 70% is ≥10 hours, or the cumulative time of temperature between -5 and 0°C and humidity greater than 70% is ≥5 hours, and the time of >1°C is <3 hours, and the wind excitation judgment condition meets the third-level judgment condition, the galloping risk level is output as the third level, otherwise, proceed to step S448; Step S448, judge the past 24 hours and the future 24 hours, if the cumulative time of temperature being between -9 and -5°C and humidity being greater than 70% is ≥8 hours, or the cumulative time of temperature being between -5 and 0°C and humidity being greater than 70% is ≥5 hours, and the time of >1°C during the period is <5 hours, and the wind excitation judgment condition meets the four-level judgment condition, the dancing risk level output is level four, otherwise the dancing risk level output is no risk, and the step ends.

10. A transmission line ice dancing prediction device based on Bi-TCN-LSTM network model, characterized in that: The device comprises: A data acquisition module, used to generate a meteorological data set based on historical meteorological forecast data, wherein the historical meteorological forecast data includes but is not limited to temperature, humidity, precipitation, wind speed, wind direction, air pressure and weather conditions, wherein the weather conditions include but are not limited to sunny days, rainy days and snowy days; A data annotation module, used for matching and annotating the historical ice dancing information of the transmission line with the meteorological data set to obtain an ice dancing data set; A feature extraction module, used for extracting features from the ice-covered dancing data set according to a temporal convolutional network model to obtain an ice-covered dancing feature data set; A model building module, used to cascade a bidirectional long short-term memory network model and the temporal convolutional network model to build a Bi-TCN-LSTM network model, and train the Bi-TCN-LSTM network model according to the ice-covered dancing feature data set; The prediction module is used to input the target meteorological forecast data into the trained Bi-TCN-LSTM network model to predict the ice dancing information of the transmission line.

Images