Cable operation parameter prediction method and device, electronic equipment and storage medium

By using the combination of time domain convolutional network and channel attention network in cable operation parameter prediction, the problem of insufficient processing capabilities of the nonlinear time series data in the prior art is solved, and the accuracy of the prediction of cable operation parameter is improved.

CN118982109APending Publication Date: 2024-11-19ZHONGTIAN ELECTRIC POWER OPTICAL CABLES CO LTD +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411045218.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

Existing time series prediction methods, such as ARIMA and SARIMA models, have limited capturing capabilities for nonlinear relationships and cannot adapt to diverse time series data, resulting in insufficient accuracy in predicting cable operation parameters.

Method used

A pre-trained parameter prediction model is used, combining the time domain convolution network and channel attention network, and the long-term dependence of time series parameters and the importance of different channels are extracted to predict cable operation parameters.

Benefits of technology

The accuracy of cable operation parameters prediction is improved, the model's adaptability to diversified time series data is enhanced, and the prediction accuracy is ensured.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118982109A_ABST
    Figure CN118982109A_ABST
Patent Text Reader

Abstract

The invention relates to the field of cable monitoring, in particular to a cable operation parameter prediction method and device, electronic equipment and a storage medium. The cable operation parameter prediction method adopts a pre-trained parameter prediction model to predict operation parameters of a to-be-detected cable, the parameter prediction model comprises a time domain convolutional network and a channel attention network, and the method comprises the following steps: obtaining time sequence parameters of the to-be-detected cable, the time sequence parameters are historical operation parameters of the to-be-tested cable arranged according to a time sequence; performing time sequence feature extraction on the time sequence parameter based on a time domain convolutional network to obtain a target time sequence feature; determining target weight information based on the time sequence parameters and a channel attention network, wherein the target weight information represents the weight of each channel in the target time sequence characteristics; and predicting operation parameters of the cable based on the target time sequence characteristics and the target weight information to obtain predicted operation parameters of the to-be-detected cable. The accuracy of cable operation parameter prediction can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of cable monitoring, and specifically to a cable operation parameter prediction method, device, electronic device and storage medium. Background Art

[0002] Time series prediction plays an important role in cable operation monitoring. It can predict the future operating status of the cable, which is conducive to relevant staff to deal with the predicted cable operating status in a timely manner and ensure the normal operation of the cable.

[0003] At present, methods such as the Autoregressive Integrated Moving Average model (ARIMA model) and the Seasonal Autoregressive Integrated Moving Average model (SARIMA model) can be used for time series prediction.

[0004] However, these methods have limited ability to capture nonlinear relationships and cannot adapt to nonlinear time series data. In other words, these methods cannot adapt to the processing requirements of diverse time series data and cannot guarantee the prediction accuracy of diverse time series data. Summary of the invention

[0005] In view of the above, the embodiments of the present application provide a cable operating parameter prediction method, device, electronic device and storage medium, which can predict the cable operating parameters based on different types of time series data and improve the accuracy of cable operating parameter prediction.

[0006] In a first aspect, an embodiment of the present application provides a cable operation parameter prediction method, wherein the cable operation parameter prediction method uses a pre-trained parameter prediction model to predict the operation parameters of a cable to be tested, wherein the parameter prediction model includes at least one time domain convolutional network and at least one channel attention network, and the cable operation parameter prediction method includes:

[0007] Acquire time series parameters of the cable to be tested, where the time series parameters are historical operating parameters of the cable to be tested arranged in chronological order;

[0008] Based on the time domain convolutional network, time series features are extracted from the time series parameters to obtain target time series features, where the target time series features are multi-channel features;

[0009] Determining target weight information based on the time series parameter and the channel attention network, wherein the target weight information represents the weight of each channel in the target time series feature;

[0010] The operating parameters of the cable under test are predicted based on the target timing characteristics and the target weight information to obtain the predicted operating parameters of the cable under test.

[0011] The embodiment of the present application combines a time domain convolutional network and a channel attention mechanism to predict the operating parameters of the cable to be tested. The time convolutional network can better capture the long-term dependencies in the time series parameters to extract more accurate time series features. The channel attention network can focus on the importance of different channels of time series features based on the time series parameters, so that the model has stronger adaptability when facing diverse data, thereby improving the accuracy of cable operating parameter prediction.

[0012] In some embodiments, the parameter prediction model includes multiple time domain convolutional networks, and the time series feature extraction of the time series parameters based on the time domain convolutional networks to obtain the target time series features includes:

[0013] Based on the time series parameter and a first time domain convolutional network among the multiple time domain convolutional networks, determining an output time series feature of the first time domain convolutional network;

[0014] Determining the input timing characteristics of the next time domain convolutional network of the first time domain convolutional network based on the output timing characteristics of the first time domain convolutional network;

[0015] Inputting the input time series feature into the next time domain convolutional network of the first time domain convolutional network to obtain the output time series feature of the next time domain convolutional network of the first time domain convolutional network;

[0016] Repeat the steps of determining the input timing features of the next time domain convolutional network of the current time domain convolutional network based on the output timing features of the current time domain convolutional network, and inputting the input timing features into the next time domain convolutional network of the current time domain convolutional network to obtain the output timing features of the next time domain convolutional network of the current time domain convolutional network, until the output timing features of the last time domain convolutional network among the multiple time domain convolutional networks are obtained;

[0017] The target temporal features are determined based on the output temporal features of the last time-domain convolutional network.

[0018] In some embodiments, the parameter prediction model includes multiple channel attention networks, and after obtaining the output timing features of the current time domain convolutional network, further includes:

[0019] Inputting the output time series features into the channel attention network to obtain output weight information;

[0020] The determining target weight information based on the time series parameter and the channel attention network includes:

[0021] The target weight information is determined based on output weight information of the multiple channel attention networks.

[0022] In some embodiments, the parameter prediction model further includes a feature weighting module, and the step of determining the input time series features of the next time series convolutional network of the current time series convolutional network based on the output time series features of the current time series convolutional network includes:

[0023] In a case where the feature weighting module is configured between the current time-domain convolutional network and a next time-domain convolutional network of the current time-domain convolutional network, summing up the first several output weight information based on the feature weighting module to obtain summed weight information;

[0024] The first plurality of output weight information is the output weight information of the plurality of channel attention modules located before the feature weighting module in the parameter prediction model;

[0025] Based on the feature weighting module, the summed weight information and the output timing features of the current time domain convolutional network are subjected to channel weighting processing to obtain the input timing features of the next time domain convolutional network of the current time domain convolutional network.

[0026] In some embodiments, the parameter prediction model further includes a fully connected layer, and the operating parameters of the cable to be tested are predicted based on the target timing characteristics and the target weight to obtain the predicted operating parameters of the cable to be tested, including:

[0027] Performing channel weighting processing on each channel in the target time series feature based on the target weight to obtain a weighted time series feature;

[0028] Based on the weighted time series features and the fully connected layer, the predicted operating parameters are determined.

[0029] In some embodiments, the step of training the parameter prediction model includes:

[0030] Acquire a cable operation parameter sample, wherein the cable operation parameter sample includes a time series parameter sample and a prediction label, wherein the time series parameter sample includes the operation parameters of the cable arranged in chronological order, and the prediction label is the operation parameter monitored after the latest monitoring time, and the latest monitoring time is the monitoring time of the latest monitored operation parameter in the time series parameter sample;

[0031] Based on the time domain convolutional network, time series feature extraction is performed on the time series parameter samples to obtain time series feature samples, where the time series feature samples are multi-channel features;

[0032] Determine a target weight sample based on the time series parameter sample and the channel attention network, wherein the target weight sample represents the weight of each channel in the target time series feature sample;

[0033] Predicting the operating parameters of the cable based on the time series feature samples and the target weight samples to obtain predicted operating parameter samples;

[0034] The model parameters of the parameter prediction model are updated based on the predicted operating parameter samples and the predicted labels until the parameter prediction model converges.

[0035] In some embodiments, the time domain convolutional network includes a dilated causal convolution layer, a weight normalization layer, an activation function layer, and a random dropout layer.

[0036] In a second aspect, an embodiment of the present application further provides a cable operation parameter prediction device, which is used to predict the operation parameters of the cable to be tested using a pre-trained parameter prediction model, wherein the parameter prediction model includes at least one time domain convolutional network and at least one channel attention network, and the cable operation parameter prediction device includes:

[0037] A parameter acquisition module, used to acquire time series parameters of the cable to be tested, wherein the time series parameters are various historical operating parameters of the cable to be tested arranged in chronological order;

[0038] A feature extraction module, used for extracting time series features from the time series parameters based on the time domain convolutional network to obtain target time series features, wherein the target time series features are multi-channel features;

[0039] A weight acquisition module, used to determine target weight information based on the time series parameters and the channel attention network, wherein the target weight information represents the weight of each channel in the target time series feature;

[0040] A parameter prediction module is used to predict the operating parameters of the cable under test based on the target timing characteristics and the target weights to obtain the predicted operating parameters of the cable under test.

[0041] In a third aspect, an embodiment of the present application further provides an electronic device, comprising a processor and a memory, wherein the memory is used to store instructions, and the processor is used to call the instructions in the memory so that the electronic device executes the cable operation parameter prediction method of the first aspect mentioned above.

[0042] An embodiment of the present application further provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed on an electronic device, the electronic device executes the cable operation parameter prediction method described in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of the model structure of a parameter prediction model provided according to an embodiment of the present application.

[0044] Figure 2 The present invention is a flowchart of the steps of a cable operation parameter prediction method provided according to an embodiment of the present application.

[0045] Figure 3 The present invention is a flowchart of the steps of a method for training a parameter prediction model according to an embodiment of the present application.

[0046] Figure 4 The present invention is a schematic diagram of the module structure of a cable operation parameter prediction device provided according to an embodiment of the present application.

[0047] Figure 5 The figure is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the present application is described in detail below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that the implementation methods of the present application and the features in the implementation methods can be combined with each other without conflict.

[0049] In the following description, many specific details are set forth to facilitate a full understanding of the present application. The described implementations are only part of the implementations of the present application, rather than all the implementations.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0051] It should be further noted that, in this article, the terms "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0052] In this application, "at least one" means one or more, and "more than one" means two or more than two. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and drawings of this application are used to distinguish similar objects, rather than to describe a specific order or sequence.

[0053] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0054] The cable operation parameter prediction method of the present application can be applied to one or more electronic devices. The electronic device is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to processors, microprogrammed control units (MCU), application specific integrated circuits (ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc. The electronic device can be a portable electronic device (such as a mobile phone, a tablet computer), a personal computer, a server, etc.

[0055] First, an embodiment of the present application provides a cable operation parameter prediction method, which uses a pre-trained parameter prediction model to predict the operation parameters of the cable to be tested, and the parameter prediction model includes at least one time domain convolutional network and at least one channel attention network.

[0056] refer to Figure 1 As shown, Figure 1 1 is a schematic diagram of the model structure of the parameter prediction model provided by an embodiment of the present application. According to different requirements, the structure of the time domain convolutional network of the parameter prediction model, the number of time domain convolutional networks, and the number of channel attention networks can be changed.

[0057] Among them, the time domain convolutional network can include: a dilated causal convolution layer (i.e., a Dilated CausalConvolutions layer), a weight normalization layer (i.e., a weightNorm layer), an activation function layer, such as a relu activation function, and a random inactivation layer (i.e., a dropout layer).

[0058] The dilated causal convolutional layer can capture the local and global features of the input data, expand the receptive field without increasing the number of parameters, and maintain the order of temporal information.

[0059] The weight normalization layer can normalize the features generated after the dilated causal convolution layer.

[0060] The activation function can reduce the instability of the training process and increase the nonlinear expression ability of the network.

[0061] During the parameter prediction model training process, the parameter prediction model is in training mode and the dropout layer is enabled. The dropout layer can randomly discard some neurons to reduce overfitting and improve the generalization ability of the parameter prediction model.

[0062] During the parameter prediction model inference process, the parameter prediction model is in evaluation mode, the dropout layer stops randomly dropping certain neurons, and the dropout layer is not enabled.

[0063] The above-mentioned channel attention network can adopt channel attention mechanisms such as Squeeze-and-Excitation Networks (SE block), but is not limited to this. In practical applications, it can be set according to needs. For example, an efficient channel attention mechanism (ECA) can also be used.

[0064] The channel attention network can adjust the importance of different channels at a specific time step, actively learn the correlation between dynamic features, and improve the parameter prediction model's attention to different channel information, thereby enhancing the model's ability to model time series data.

[0065] Taking the channel attention network as SE block as an example, the SE block can include a global average pooling layer, a fully connected layer and an activation function layer. The SE block is used to obtain the scalars of each feature channel of the input parameter based on the global average pooling layer. The feature channel scalar represents the average activity level of each channel. Then, each feature channel scalar is input into the fully connected layer and the activation function layer to normalize each channel to the range of (0, 1), thereby representing the weight of each channel and obtaining the output weight information of the channel attention network.

[0066] refer to Figure 2 As shown, Figure 2 This is a flowchart of the steps of cable operation parameter prediction provided by an embodiment of the present application. According to different requirements, the order of the steps in the flowchart can be changed, and some steps can be omitted. Figure 1 and Figure 2 The cable operation parameter prediction method provided in the embodiment of the present application is described.

[0067] The cable operating parameter method may include the following steps.

[0068] Step 201, obtaining time series parameters of the cable to be tested.

[0069] The time series parameters are the historical operating parameters of the cable to be tested arranged in chronological order.

[0070] The historical operating parameters may be power operating parameters monitored from the cable under test, such as the current of the cable under test, etc., or environmental parameters of the operating environment of the current under test, such as temperature, humidity, etc., or the strain condition of the cable under test.

[0071] For example, assuming that the ambient temperature of the operating environment of the cable to be tested is continuously monitored from the first day to the second day, the electronic device can collect the ambient temperature from the first day to the second day at a preset time interval to obtain a time series parameter, which can be used to predict the ambient temperature after the second day.

[0072] Step 202: extract time series features from time series parameters based on a time domain convolutional network to obtain target time series features.

[0073] Among them, the target time series feature is a multi-channel feature.

[0074] In some embodiments, reference Figure 1 As shown, the parameter prediction model includes: a plurality of serially connected time domain convolutional networks. Step 202 may refer to the descriptions in (1) to (5) below. Step 202 may include:

[0075] (1) The electronic device determines the output timing characteristics of the first time-domain convolutional network in a plurality of series-connected time-domain convolutional networks based on the time series parameters.

[0076] For example, the electronic device may input the time series parameters into a first time-domain convolutional network to obtain output features of the first time-domain convolutional network.

[0077] (2) The electronic device determines the input timing characteristics of the next time domain convolutional network of the first time domain convolutional network based on the output timing characteristics of the first time domain convolutional network.

[0078] For example, the electronic device may use the output timing features of the first convolutional neural network as the input timing features of the next time-domain convolutional network of the first time-domain convolutional network.

[0079] (3) The electronic device inputs the input time series features in the above step (2) into the next time domain convolutional network of the first time domain convolutional network to obtain the output time series features of the next time domain convolutional network of the first time domain convolutional network.

[0080] (4) The electronic device repeatedly determines the input timing characteristics of the next time domain convolutional network of the current time domain convolutional network based on the output timing characteristics of the current time domain convolutional network. For example, the parameter prediction model includes Figure 1 The feature weighting module 101 shown in the figure is located between the current time domain convolutional network and the next time domain convolutional network. The electronic device can determine the input timing characteristics of the next time domain convolutional network of the current time domain convolutional network based on the feature weighting module 101 and the output characteristics of the current time domain convolutional network, and then input the input timing characteristics into the next time domain convolutional network of the current time domain convolutional network to obtain the output timing characteristics of the next time domain convolutional network of the current time domain convolutional network, until the output timing characteristics of the last time domain convolutional network in the multiple series-connected time domain convolutional networks are obtained.

[0081] That is, in some embodiments, the parameter prediction model may further include a feature weighting module 101, and in a plurality of serially connected time domain convolutional networks, a feature weighting module 101 is configured between two adjacent time domain convolutional networks. Figure 1 A feature weighting module 101 is configured between the second time domain convolutional network and the third time domain convolutional network.

[0082] The feature weighting module 101 is used to perform weighted summation on the output weight information of the first several channel attention networks to obtain summed weight information.

[0083] Among them, the first several output weight information is the output weight information of several channel attention modules located before the feature weighting module 101 in the parameter prediction model.

[0084] In the parameter prediction model, the embodiment of the present application configures a feature weighting module between two adjacent time domain convolutional networks, which can enable the parameter prediction model to adjust the weights of channels of temporal features at different depth levels, thereby enabling the parameter prediction model to focus on feature identifiers at different depth levels and improve the accuracy of parameter prediction.

[0085] In some embodiments, when the feature weighting module 101 is the first feature weighting module 101 of the parameter prediction model, the first several output weight information are the output weight information of all channel attention modules located before the first feature weighting module 101 in the parameter prediction model.

[0086] When the feature weighting module 101 is not the first feature weighting module 101 of the parameter prediction model, the first several output weight information are the output weight information of all channel attention modules located between the current feature weighting module and the previous feature weighting module in the parameter prediction model.

[0087] For example, assume that N time-domain convolutional networks connected in series and N+1 channel attention modules are provided before the first feature weighting module 101. The time series parameters are respectively input into the first time-domain convolutional network and the first channel attention network.

[0088] One output branch of the time domain convolutional network points to the next time domain convolutional network, and another output branch of the time domain convolutional network points to a channel attention network. That is, the output time domain features of N series-connected time domain convolutional networks are all input into the corresponding channel attention module to obtain the output weight information corresponding to the output time domain features.

[0089] Through the above method, the feature weighting module can obtain the output weight information of N+1 time domain convolutional networks, and sum the output weight information of N+1 time domain convolutional networks to obtain the summed weight information.

[0090] For example, refer to Figure 1 As shown, Figure 1 A feature weighting module 101 is configured between the second time domain convolutional network and the third time domain convolutional network, and two series-connected time domain convolutional networks and three channel attention modules are arranged before the feature weighting module 101.

[0091] The time series parameters are input into the first channel attention network and the first temporal convolutional network respectively.

[0092] The channel attention network is used to calculate the weights of each channel of the time series parameter to obtain output weight information corresponding to the time series parameter, such as weight information 1.

[0093] The output timing features of the first time domain convolutional network can be input into the second channel attention network to obtain output weight information corresponding to the output timing features, such as weight information 2.

[0094] The first time domain convolutional network extracts time series features based on the time series parameters to obtain the output time series features of the first time domain convolutional network.

[0095] The output timing features of the first time-domain convolutional network can also be used as the input timing features of the second time-domain convolutional network and input into the second time-domain convolutional network to obtain the output timing features of the second time-domain convolutional network.

[0096] The output time series features of the second time domain convolutional network can be input into the third channel attention network to obtain output weight information, such as weight information 3.

[0097] Through the above method, the feature weighting module can obtain the output weight information of the three time domain convolutional networks, such as weight information 1, weight information 2 and weight information 3, and then add the output weight information of the three time domain convolutional networks to obtain the summed weight information.

[0098] Next, the electronic device can perform channel weighting processing on the summed weight information and the output timing features of the current time domain convolutional network based on the feature weighting module to obtain the input timing features of the next time domain convolutional network of the current time domain convolutional network.

[0099] That is, the electronic device can add the weight of each channel represented in the summed weight information to each channel feature in the output time series feature of the current time domain convolutional network to obtain the weighted time series feature; and use the weighted time series feature as the input time series feature of the next time domain convolutional network of the current time domain convolutional network. Figure 1 As shown, the electronic device can obtain the input timing features to be input into the third time domain convolutional network based on this.

[0100] In some embodiments, the input time series feature can also be input into the next channel attention network to extract the output weight information of the time series feature to facilitate subsequent channel weighting. Figure 2 The fourth channel attention network shown.

[0101] (5) The electronic device determines the target timing characteristics based on the output timing characteristics of the last time-domain convolutional network.

[0102] The electronic device may use the output time series feature of the last time domain convolutional network as the target time series feature. For example, the electronic device may Figure 1 The output temporal features of the fourth time-domain convolutional network shown are used as target temporal features.

[0103] Step 203, determine target weight information based on time series parameters and channel attention network.

[0104] Among them, the target weight information represents the weight of each channel in the target time series feature.

[0105] In some embodiments, step 203 may include: the electronic device may determine target weight information based on output weight information of multiple channel attention networks.

[0106] For example, refer to Figure 1 As shown, after the last time-domain convolutional network of the parameter prediction module, a feature weighting module 101 is also configured. The electronic device can add the output weight information of the multiple channel attention networks based on the feature weighting module 101 to obtain the target weight information.

[0107] Among them, the output weight information of the multiple channel attention networks can be the output weight information of the channel attention network between the last feature weighting module 101 and the previous feature weighting module 101.

[0108] For example, refer to Figure 1 As shown, the electronic device may add up the weights of each channel in weight information 4, weight information 5 and weight information 6 to obtain target weight information.

[0109] In the parameter prediction module, if there is no other feature weighting module 101 before the last feature weighting module 101, the electronic device may sum the output weight information of all channel attention networks in the parameter prediction model to obtain the target weight information.

[0110] The above takes the example of a parameter prediction model including multiple channel attention networks and multiple time domain convolutional networks to illustrate the implementation details of step 202 and step 203. In actual application, the number of channel attention modules and time domain convolutional networks included in the parameter prediction model can be configured according to actual application requirements.

[0111] For example, the parameter prediction model may include a channel attention module and a time domain convolutional network. In step 202, the electronic device may input the time series parameters into the time domain convolutional network, and use the output time series features of the time domain convolutional network as the target time series features. In step 203, the electronic device may also input the time series parameters into the channel attention network, and use the output weight information of the channel attention network as the target weight information. Alternatively, the electronic device may input the target time series features into the channel attention network and use the output weight information of the channel attention network as the target weight information.

[0112] For another example, the parameter prediction model may include two channel attention modules and a time domain convolutional network. In step 202, the electronic device may input the time series parameters into the time domain convolutional network, and use the output time series features of the time domain convolutional network as the target time series features. In step 203, the electronic device may input the time series parameters into the first channel attention network to obtain weight information 1. The electronic device also inputs the target time series features into the second channel attention network to obtain weight information 2. Then, the electronic device adds the weight information 1 and the weight information 2 to obtain the target weight.

[0113] The number of channel attention modules and time domain convolutional networks included in the above parameter prediction model is only an example, and can be configured according to actual application requirements in actual application. The position and number of feature weighting modules in the parameter prediction model can also be set according to actual application requirements, and the embodiments of this application are not limited to this.

[0114] Step 204 , predicting the operating parameters of the cable to be tested based on the target timing characteristics and the target weight information to obtain the predicted operating parameters of the cable to be tested.

[0115] In some embodiments, step 204 may include: performing channel weighting processing on each channel in the target timing feature based on the target weight to obtain a weighted timing feature, and then determining the predicted operating parameters based on the weighted timing feature and the fully connected layer.

[0116] For example, the electronic device inputs the weighted time series features into the fully connected layer to obtain the predicted operating parameters.

[0117] In some embodiments, after obtaining the predicted operating parameters, the electronic device may also compare the predicted operating parameters with preset operating parameter standards, and may issue an alarm if the predicted operating parameters do not meet the operating parameter standards.

[0118] For example, assuming that the operating parameter is the ambient temperature of the operating environment of the cable to be tested, the operating parameter standard includes that the ambient temperature of the cable does not exceed the preset temperature warning threshold. The predicted operating parameter is the predicted ambient temperature of the cable operating environment at the next moment. If the predicted operating parameter exceeds the preset temperature warning threshold, the electronic system can issue an alarm to provide feedback to the relevant staff.

[0119] The present application also provides a method for training a parameter prediction model. The training method can be used to train the parameter prediction model in any of the above embodiments. Figure 3 As shown, the parameter prediction model may include:

[0120] Step 301, obtaining a cable operation parameter sample.

[0121] Specifically, the electronic device can obtain cable operating parameter samples.

[0122] The cable operation parameter samples include time series parameter samples and prediction labels.

[0123] The time series parameter sample includes operating parameters of the cable arranged in time sequence.

[0124] The prediction label is the operating parameter monitored after the latest monitoring time, and the latest monitoring time is the monitoring time of the latest monitored operating parameter in the time series parameter sample.

[0125] For example, the latest monitored operating parameter in the time series parameter sample is monitored at 1:00 a.m., and the prediction label may be the operating parameter actually monitored at 1:15 a.m.

[0126] The steps for collecting cable operating parameter samples may include:

[0127] Step 1: Collect the operating parameters of the cable using a circulating current collector, a temperature sensor collector, a humidity sensor collector, a light sensor, such as Brillouin optical time domain analysis (ie, BOTDA), etc. The operating parameters may include, but are not limited to, collecting current, temperature, humidity, strain, etc.

[0128] Step 2: Preprocess the collected data, for example, remove abnormal data such as incomplete and empty data. The sampling rate of the operating parameters can be set according to the actual application requirements. For example, the sampling rate of each operating parameter is uniformly configured as: the unit collection time is 1 minute interval.

[0129] Step 3: Classify the collected operating parameters and divide the data set according to a certain ratio. For example, 70% is used as a training set and 30% is used as a test set.

[0130] Step 302: extract time series features from the time series parameter samples based on the time domain convolutional network to obtain time series feature samples.

[0131] Step 303, determine the target weight sample based on the time series parameter sample and the channel attention network.

[0132] Step 304 , predicting the operating parameters of the cable based on the time series feature samples and the target weight samples to obtain predicted operating parameter samples.

[0133] The implementation details of the above steps 302 to 304 may refer to steps 201 to 204 and will not be repeated here.

[0134] Step 305 , updating the model parameters of the parameter prediction model based on the predicted operating parameter samples and the predicted labels until the parameter prediction model converges.

[0135] In some embodiments, step 305 may include: the electronic device may obtain a loss value based on a preset loss function, a predicted operating parameter sample and a predicted label, and update the model parameters of the parameter prediction model based on the loss value, such as updating the model parameters of the parameter prediction model based on the loss value and the gradient descent method.

[0136] For example, the loss value can be calculated as follows:

[0137]

[0138] Among them, the preset loss function is recorded as F R2 , the sample of predicted operating parameters is recorded as The predicted label is denoted as y i .

[0139] The parameter prediction model of the embodiment of the present application uses a time-domain convolutional network for parameter prediction. The time-domain convolutional network not only has good parallelization capability and computational efficiency, and can more efficiently process large-scale time series parameters and reduce the time cost of training and inference processes, but also the time-domain convolutional network can model long-term dependencies and has good processing performance when faced with time series parameters with long-term dependencies.

[0140] In addition, the parameter prediction model also introduces a channel attention network, which combines the channel attention network with the time domain convolutional network, and can more flexibly model local and global features in time series parameters. The channel attention network has the ability to weight feature channels according to the importance of the channels, which can enable the parameter prediction model to better pay attention to important channels and understand and utilize the correlation between different channel features, thereby ensuring that the parameter prediction model has strong adaptability to diverse time series data such as linear and nonlinear, thereby improving the accuracy of cable operation parameter prediction.

[0141] Moreover, the embodiments of the present application can also use multiple convolutional neural networks to extract the timing characteristic parameters of the cable to be tested, which can enhance the expression ability of the target timing characteristics. The present application uses multiple channel attention networks to process the output timing characteristics of different convolutional neural networks, which can further improve the expression ability of the timing characteristics used to generate predicted operating parameters.

[0142] Based on the same idea as the cable operation parameter prediction method in the above embodiment, the present application also provides a cable operation parameter prediction device, which can be used to execute the above cable operation parameter prediction method. For ease of explanation, the structural schematic diagram of the cable operation parameter prediction device embodiment only shows the parts related to the embodiment of the present application. It can be understood by those skilled in the art that the illustrated structure does not constitute a limitation on the device, and may include more or fewer components than shown in the diagram, or combine certain components, or arrange the components differently.

[0143] like Figure 4 As shown, the cable operation parameter prediction device includes a parameter acquisition module 401, a feature extraction module 402, a weight acquisition module 403 and a parameter prediction module 404. In some embodiments, the above modules can be programmable software instructions stored in a memory and can be called and executed by a processor. It is understood that in other embodiments, the above modules can also be program instructions or firmware solidified in the processor.

[0144] The parameter acquisition module 401 is used to acquire the time series parameters of the cable to be tested, where the time series parameters are various historical operating parameters of the cable to be tested arranged in chronological order.

[0145] The feature extraction module 402 is used to extract time series features from the time series parameters based on the time domain convolutional network to obtain target time series features, where the target time series features are multi-channel features.

[0146] The weight acquisition module 403 is used to determine the target weight information based on the time series parameters and the channel attention network, and the target weight information represents the weight of each channel in the target time series feature.

[0147] The parameter prediction module 404 is used to predict the operating parameters of the cable based on the target timing characteristics and the target weight information to obtain the predicted operating parameters of the cable to be tested.

[0148] Figure 5 This is a schematic diagram of an embodiment of an electronic device of the present application.

[0149] The electronic device 100 includes a memory 20, a processor 30, and a computer program 40 stored in the memory 20 and executable on the processor 30. When the processor 30 executes the computer program 40, the steps in the above-mentioned cable operating parameter method embodiment are implemented, for example Figure 2 Steps 201 to 204 are shown.

[0150] Exemplarily, the computer program 40 can also be divided into one or more modules / units, which are stored in the memory 20 and executed by the processor 30. The one or more modules / units can be a series of computer program instruction segments that can perform specific functions, and the instruction segments are used to describe the execution process of the computer program 40 in the electronic device 100. For example, it can be divided into Figure 4 The shown module includes a parameter acquisition module 401 , a feature extraction module 402 , a weight acquisition module 403 and a parameter prediction module 404 .

[0151] Those skilled in the art will appreciate that the schematic diagram is merely an example of the electronic device 100 and does not constitute a limitation on the electronic device 100. The electronic device 100 may include more or fewer components than shown in the diagram, or a combination of certain components, or different components. For example, the electronic device 100 may also include input and output devices, network access devices, buses, etc.

[0152] The processor 30 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, a single-chip microcomputer, or the processor 30 may also be any conventional processor, etc.

[0153] The memory 20 can be used to store the computer program 40 and / or the module / unit. The processor 30 realizes various functions of the electronic device 100 by running or executing the computer program and / or the module / unit stored in the memory 20 and calling the data stored in the memory 20. The memory 20 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data (such as audio data) created according to the use of the electronic device 100, etc. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other non-volatile solid-state storage devices.

[0154] If the module / unit integrated in the electronic device 100 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0155] In the several embodiments provided in this application, it should be understood that the disclosed electronic device and method can be implemented in other ways. For example, the electronic device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and there may be other division methods in actual implementation.

[0156] In addition, each functional unit in each embodiment of the present application may be integrated into the same processing unit, each unit may exist physically separately, or two or more units may be integrated into the same unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0157] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the spirit or basic features of the present application. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or electronic devices stated in the electronic device claim can also be implemented by the same unit or electronic device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

[0158] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the above embodiments, a person of ordinary skill in the art should understand that the technical solution of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present application.

Claims

1. A cable operation parameter prediction method, characterized in that: The cable operation parameter prediction method adopts a pre-trained parameter prediction model to predict the operation parameters of the cable to be tested, and the parameter prediction model includes at least one time domain convolutional network and at least one channel attention network. The cable operation parameter prediction method includes: Acquire time series parameters of the cable to be tested, where the time series parameters are historical operating parameters of the cable to be tested arranged in chronological order; Based on the time domain convolutional network, time series features are extracted from the time series parameters to obtain target time series features, where the target time series features are multi-channel features; Determining target weight information based on the time series parameter and the channel attention network, wherein the target weight information represents the weight of each channel in the target time series feature; The operating parameters of the cable under test are predicted based on the target timing characteristics and the target weight information to obtain the predicted operating parameters of the cable under test.

2. The cable operation parameter prediction method according to claim 1, characterized in that: The parameter prediction model includes a plurality of time domain convolutional networks, and the time series feature extraction of the time series parameters based on the time domain convolutional networks to obtain target time series features includes: Based on the time series parameter and a first time domain convolutional network among the multiple time domain convolutional networks, determining an output time series feature of the first time domain convolutional network; Determining the input timing characteristics of the next time domain convolutional network of the first time domain convolutional network based on the output timing characteristics of the first time domain convolutional network; Inputting the input time series feature into the next time domain convolutional network of the first time domain convolutional network to obtain the output time series feature of the next time domain convolutional network of the first time domain convolutional network; Repeat the steps of determining the input timing features of the next time domain convolutional network of the current time domain convolutional network based on the output timing features of the current time domain convolutional network, and inputting the input timing features into the next time domain convolutional network of the current time domain convolutional network to obtain the output timing features of the next time domain convolutional network of the current time domain convolutional network, until the output timing features of the last time domain convolutional network among the multiple time domain convolutional networks are obtained; The target temporal features are determined based on the output temporal features of the last time-domain convolutional network.

3. The cable operation parameter prediction method according to claim 2, characterized in that: The parameter prediction model includes multiple channel attention networks, and after obtaining the output time series features of the current time domain convolutional network, further includes: Inputting the output time series features into the channel attention network to obtain output weight information; The determining target weight information based on the time series parameter and the channel attention network includes: The target weight information is determined based on output weight information of the multiple channel attention networks.

4. The cable operation parameter prediction method according to claim 3, characterized in that: The parameter prediction model also includes a feature weighting module, and the step of determining the input time series feature of the next time series convolutional network of the current time series convolutional network based on the output time series feature of the current time series convolutional network includes: In a case where the feature weighting module is configured between the current time-domain convolutional network and a next time-domain convolutional network of the current time-domain convolutional network, summing up the first several output weight information based on the feature weighting module to obtain summed weight information; The first plurality of output weight information is the output weight information of the plurality of channel attention modules located before the feature weighting module in the parameter prediction model; Based on the feature weighting module, the summed weight information and the output timing features of the current time domain convolutional network are subjected to channel weighting processing to obtain the input timing features of the next time domain convolutional network of the current time domain convolutional network.

5. The cable operation parameter prediction method according to claim 1, characterized in that: The parameter prediction model further includes a fully connected layer, and the operating parameters of the cable to be tested are predicted based on the target timing characteristics and the target weight to obtain the predicted operating parameters of the cable to be tested, including: Performing channel weighting processing on each channel in the target time series feature based on the target weight to obtain a weighted time series feature; Based on the weighted time series features and the fully connected layer, the predicted operating parameters are determined.

6. The cable operation parameter prediction method according to claim 1, characterized in that: The training steps of the parameter prediction model include: Acquire a cable operation parameter sample, wherein the cable operation parameter sample includes a time series parameter sample and a prediction label, wherein the time series parameter sample includes the operation parameters of the cable arranged in chronological order, and the prediction label is the operation parameter monitored after the latest monitoring time, and the latest monitoring time is the monitoring time of the latest monitored operation parameter in the time series parameter sample; Based on the time domain convolutional network, time series feature extraction is performed on the time series parameter samples to obtain time series feature samples, where the time series feature samples are multi-channel features; Determine a target weight sample based on the time series parameter sample and the channel attention network, wherein the target weight sample represents the weight of each channel in the target time series feature sample; Predicting the operating parameters of the cable based on the time series feature samples and the target weight samples to obtain predicted operating parameter samples; The model parameters of the parameter prediction model are updated based on the predicted operating parameter samples and the predicted labels until the parameter prediction model converges.

7. The cable operation parameter prediction method according to any one of claims 1 to 6, characterized in that: The time domain convolutional network includes a dilated causal convolution layer, a weight normalization layer, an activation function layer and a random inactivation layer.

8. A cable operation parameter prediction device, characterized in that: The method is used to predict the operating parameters of the cable to be tested by using a pre-trained parameter prediction model, wherein the parameter prediction model includes at least one time domain convolutional network and at least one channel attention network, and the cable operating parameter prediction device includes: A parameter acquisition module, used to acquire time series parameters of the cable to be tested, wherein the time series parameters are various historical operating parameters of the cable to be tested arranged in chronological order; A feature extraction module, used for extracting time series features from the time series parameters based on the time domain convolutional network to obtain target time series features, wherein the target time series features are multi-channel features; A weight acquisition module, used to determine target weight information based on the time series parameters and the channel attention network, wherein the target weight information represents the weight of each channel in the target time series feature; A parameter prediction module is used to predict the operating parameters of the cable under test based on the target timing characteristics and the target weights to obtain the predicted operating parameters of the cable under test.

9. An electronic device, comprising a processor and a memory, characterized in that: The memory is used to store instructions, and the processor is used to call the instructions in the memory, so that the electronic device executes the cable operation parameter prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed on an electronic device, the electronic device executes the cable operation parameter prediction method according to any one of claims 1 to 7.

Citation Information

Patent Citations

  • Civil aircraft system-level anomaly precursor identification method based on real-time flight data

    CN116361728A

  • Processing monitoring data filling method based on channel attention and time domain convolutional network

    CN116541804A

  • Drought state prediction method, drought state prediction model training method and equipment

    CN117094452A

  • Traffic flow prediction method and system based on multi-time-sequence convolutional gated graph neural network

    CN117238135A