Operation and maintenance fault analysis method and device and computer equipment

Abstract

The embodiment of the present application relates to the technical field of artificial intelligence, and discloses an operation and maintenance fault analysis method, which comprises the following steps: obtaining a to-be-predicted operation and maintenance index sequence; the to-be-predicted operation and maintenance index sequence comprises time series data corresponding to a plurality of operation and maintenance indexes; inputting the to-be-predicted operation and maintenance index sequence into a preset operation and maintenance fault prediction model to obtain fault information of a target service; wherein the operation and maintenance fault prediction model comprises an irregular convolutional network and an LSTM network; the operation and maintenance fault prediction model inputs historical operation and maintenance fault index sequence samples into the irregular convolutional network, and inputs a one-dimensional vector obtained by processing the output convolutional result into the LSTM network to obtain training. Through the above method, the embodiment of the present application realizes the effect of accurately predicting operation and maintenance fault information.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of artificial intelligence technology, specifically to a method, apparatus, computer equipment, and computer-readable storage medium for operation and maintenance fault analysis. Background Technology

[0002] Currently, existing methods for handling operational and maintenance faults mainly involve time-series forecasting, for which research is relatively well-developed. Methods based on statistical machine learning include the Autoregressive Moving Average (ARIMA) model and regression prediction. Deep learning-based methods include RNNs and LSTMs. While ARIMA methods rely on statistical analysis of time-series data for prediction, its accuracy in fault prediction is relatively low.

[0003] Therefore, the inventors of this application have found that existing technologies and processing mechanisms cannot meet the needs of actual operation and maintenance, nor do they conform to the future development trend of IT. Summary of the Invention

[0004] In view of the above problems, embodiments of the present invention provide a method, apparatus, computer equipment and computer-readable storage medium for operation and maintenance fault analysis, which are used to solve the technical problem of inaccurate fault prediction in the prior art.

[0005] According to one aspect of the present invention, a method for analyzing operational and maintenance faults is provided, the method comprising:

[0006] Obtain the sequence of operation and maintenance indicators to be predicted; the sequence of operation and maintenance indicators to be predicted includes time series data corresponding to multiple operation and maintenance indicators;

[0007] The sequence of operation and maintenance indicators to be predicted is input into a preset operation and maintenance fault prediction model to obtain fault information of the target service; wherein, the operation and maintenance fault prediction model includes an irregular convolutional network and an LSTM network; the operation and maintenance fault prediction model inputs the historical operation and maintenance fault indicator sequence samples into the irregular convolutional network, and processes the output convolution result into a one-dimensional vector and inputs it into the LSTM network for training.

[0008] In one optional approach, before inputting the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service in a future preset time period, the method includes:

[0009] Preprocess the initial time-series data of historical operation and maintenance failure indicators to obtain the preprocessed time-series data corresponding to each historical operation and maintenance failure indicator.

[0010] Analyze the correlation between various historical operation and maintenance failure indicators and the business performance of the target business;

[0011] Based on the correlation, a preset number of historical operation and maintenance failure indicators are selected from each historical operation and maintenance failure indicator;

[0012] The preprocessed time-series data corresponding to the preset number of historical operation and maintenance fault indicators are spliced ​​together to obtain a historical operation and maintenance fault indicator sequence sample.

[0013] In one optional approach, the preprocessing of the initial time-series data of historical operation and maintenance failure indicators to obtain preprocessed time-series data corresponding to each historical operation and maintenance failure indicator further includes:

[0014] Obtain time-series data of multiple historical operation and maintenance failure indicators;

[0015] The time-series data of the historical operation and maintenance failure indicators are padded with dates to obtain the padded time-series data;

[0016] The time series data after completion is normalized using the normalization method to obtain normalized time series data.

[0017] Based on the data deviation, abnormal data in the normalized time series data are identified and replaced with preset values ​​to obtain the preprocessed time series data corresponding to the historical operation and maintenance fault indicators.

[0018] In one alternative approach, the analysis of the correlation between various historical operational failure indicators and the business performance of the target service includes:

[0019] The correlation between each of the historical operation and maintenance failure indicators and the business performance of the target service is calculated using the Pearson correlation coefficient.

[0020] In one optional approach, the step of identifying abnormal data in the normalized time-series data based on data deviation and replacing it with a preset value to obtain preprocessed time-series data corresponding to the historical operation and maintenance fault indicators includes:

[0021] Calculate the median value of all data preceding any given data in the normalized time series data;

[0022] Based on the deviation of any data point in the normalized time series data from the median value, it is determined whether the data is abnormal.

[0023] When the data is abnormal, it will be replaced with a preset value.

[0024] In one optional approach, before inputting the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service in a future preset time period, the method includes:

[0025] Construct an irregular convolutional network and an LSTM network, and assign different weights to different positions within the convolutional kernels in the irregular convolutional network;

[0026] The historical operation and maintenance fault indicator sequence samples are input into the irregular convolutional network to obtain the convolution result;

[0027] The convolution result is processed into a one-dimensional vector, and the one-dimensional vector is input into the LSTM network. The vector is then output to a one-dimensional fully connected layer through a fully connected layer to obtain the prediction result.

[0028] Based on the predicted results and the expected results corresponding to the historical operation and maintenance fault index sequence samples, the weights, as well as the parameters of the irregular convolutional network and the LSTM network, are optimized and training continues to obtain the operation and maintenance fault prediction model.

[0029] According to another aspect of the present invention, an operation and maintenance fault analysis device is provided, comprising:

[0030] The acquisition module is used to acquire the sequence of operation and maintenance indicators to be predicted; the sequence of operation and maintenance indicators to be predicted includes time-series data corresponding to multiple operation and maintenance indicators.

[0031] The prediction module is used to input the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service; wherein, the operation and maintenance fault prediction model includes an irregular convolutional network and an LSTM network; the operation and maintenance fault prediction model inputs the irregular convolutional network according to the historical operation and maintenance fault indicator sequence samples, and processes the output convolution result into a one-dimensional vector and inputs it into the LSTM network for training.

[0032] In an alternative embodiment, the device further includes:

[0033] The preprocessing module is used to preprocess the initial time-series data of historical operation and maintenance failure indicators to obtain the preprocessed time-series data corresponding to each historical operation and maintenance failure indicator.

[0034] The correlation analysis module is used to analyze the correlation between various historical operation and maintenance failure indicators and the business performance of the target business;

[0035] The filtering module is used to filter a preset number of historical operation and maintenance failure indicators from various historical operation and maintenance failure indicators based on the correlation.

[0036] The splicing module is used to splice the preprocessed time-series data corresponding to the preset number of historical operation and maintenance fault indicators to obtain a historical operation and maintenance fault indicator sequence sample.

[0037] According to another aspect of the present invention, a computer device is provided, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;

[0038] The memory is used to store at least one executable instruction, which causes the processor to perform the operation and maintenance fault analysis method.

[0039] According to another aspect of the present invention, a computer-readable storage medium is provided, the storage medium storing at least one executable instruction, which, when executed on a computer device, causes the operation of the maintenance fault analysis method described in the computer device.

[0040] This invention provides an embodiment of the method for obtaining a sequence of operation and maintenance (O&M) indicators to be predicted. This sequence includes time-series data corresponding to multiple O&M indicators. The O&M indicator sequence is then input into a pre-defined O&M fault prediction model to obtain fault information for the target service. The O&M fault prediction model includes an irregular convolutional network and an LSTM network. The O&M fault prediction model inputs historical O&M fault indicator sequence samples into the irregular convolutional network and processes the output convolutional result into a one-dimensional vector, which is then input into the LSTM network for training. This method can accurately predict O&M fault information.

[0041] The above description is merely an overview of the technical solutions of the embodiments of the present invention. In order to better understand the technical means of the embodiments of the present invention and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0042] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0043] Figure 1 A flowchart illustrating the operation and maintenance fault analysis method provided in an embodiment of the present invention is shown;

[0044] Figure 2 A flowchart illustrating the irregular convolutional network in the operation and maintenance fault analysis method provided in this embodiment of the invention is shown.

[0045] Figure 3 This diagram illustrates model training in the operation and maintenance fault analysis method provided in this embodiment of the invention.

[0046] Figure 4 A schematic diagram of the operation and maintenance fault analysis device provided in an embodiment of the present invention is shown;

[0047] Figure 5 A schematic diagram of the structure of a computer device provided in an embodiment of the present invention is shown. Detailed Implementation

[0048] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0049] Figure 1 A flowchart of an operation and maintenance fault analysis method provided in an embodiment of the present invention is shown. This method is executed by a computer device. The computer device can be a desktop computer, laptop computer, tablet computer, smart terminal device, distributed computing device, etc., and the embodiments of the present invention do not impose specific limitations. Figure 1 As shown, the method includes the following steps:

[0050] Step 110: Obtain the sequence of operation and maintenance indicators to be predicted; the sequence of operation and maintenance indicators to be predicted includes time series data corresponding to multiple operation and maintenance indicators.

[0051] In this embodiment of the invention, the sequence of operation and maintenance indicators to be predicted can be time-series data corresponding to multiple operation and maintenance indicators within a current preset time period. For example, it can be the current time-series data of each operation and maintenance indicator at the current time, and the historical time-series data of each operation and maintenance indicator within a preset time period prior to the current time, forming the time-series data for the current preset time period. These operation and maintenance indicators include multiple indicators related to the target business, such as performance indicators like CPU utilization, memory utilization, disk utilization, and network latency.

[0052] Step 120: Input the sequence of operation and maintenance indicators to be predicted into the preset operation and maintenance fault prediction model to obtain the fault information of the target service.

[0053] The operation and maintenance fault prediction model includes an irregular convolutional network and an LSTM network. The operation and maintenance fault prediction model is obtained by inputting historical operation and maintenance fault index sequence samples into the irregular convolutional network and processing the output convolution result into a one-dimensional vector and inputting it into the LSTM network for training.

[0054] In this embodiment of the invention, the fault information of the target service is the fault alarm corresponding to the target service.

[0055] In this embodiment of the invention, a neural network is pre-trained using historical operation and maintenance fault indicator sequence samples to obtain an operation and maintenance fault prediction model. First, data processing is performed to obtain historical operation and maintenance fault indicator sequence samples, which include time-series data corresponding to each historical operation and maintenance fault indicator and corresponding business performance time-series data. Specifically, the following steps are included:

[0056] Step 001: Preprocess the initial time-series data of historical operation and maintenance failure indicators to obtain preprocessed time-series data corresponding to each historical operation and maintenance failure indicator. First, time-series data of multiple historical operation and maintenance failure indicators are acquired. Then, date padding is performed on the time-series data of the historical operation and maintenance failure indicators to obtain padded time-series data. Specifically, date padding is performed on the time-series data to ensure the integrity of the periodic data, and null values ​​are handled for the padded data. In this embodiment of the invention, linear interpolation can be used for null value filling. Afterwards, since the numerical ranges of different indicators are inconsistent, data normalization is required. The padded time-series data is normalized using a normalization method to obtain normalized time-series data. Finally, based on the data deviation, abnormal data in the normalized time-series data is identified and replaced with preset values ​​to obtain the preprocessed time-series data corresponding to the historical operation and maintenance failure indicators. Specifically, the median value of all data preceding any given data point in the normalized time-series data is calculated; based on the deviation of any given data point from the median value, it is determined whether the data is abnormal; if the data is abnormal, it is replaced with a preset value. This preset value can be null. When the deviation exceeds a preset deviation threshold, the data is determined to be abnormal.

[0057] The deviation can be calculated using the following formula:

[0058]

[0059] Where, x i This represents the i-th value in the time series data. This represents the median before the i-th value in the sequence.

[0060] Step 002: Analyze the correlation between various historical maintenance failure indicators and the business performance of the target service. Specifically, the Pearson correlation coefficient is used to calculate the correlation between each historical maintenance failure indicator and the business performance of the target service. This business performance can be, for example, the service response rate. Business performance data at various historical time points can be obtained, thus yielding time-series data of the business performance. The time-series data of each historical maintenance failure indicator are mapped to the corresponding business performance time series, and the Pearson correlation coefficient is used to determine the impact of each historical maintenance failure indicator on business performance. For example, if the service response rate is used as the target time series for prediction, the Pearson correlation coefficient is used to calculate the correlation between the target sequence and multiple maintenance indicators such as CPU utilization, memory utilization, disk utilization, and network latency.

[0061] Step 003: Based on the correlation, select a preset number of historical operation and maintenance failure indicators from each historical operation and maintenance failure indicator. Among them, the top 5 operation and maintenance indicators can be selected based on the correlation as the indicator data of the business performance corresponding to the target business.

[0062] Step 004: Concatenate the preprocessed time-series data corresponding to the preset number of historical operation and maintenance fault indicators to obtain a historical operation and maintenance fault indicator sequence sample.

[0063] Based on the correlation analysis results, the time-series data of the five most correlated operation and maintenance indicators at each historical moment [S1, S2, S3, S4, S5] are obtained, and horizontally concatenated to obtain the time-series data matrix of historical operation and maintenance indicators. The time-series data of business performance corresponding to the time-series data matrix of historical operation and maintenance indicators are determined to form a sample of historical operation and maintenance fault indicator sequences.

[0064] Among them, such as Figure 3 As shown, in this embodiment of the invention, before inputting the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service in a future preset time period, the method further includes:

[0065] Step 005: Construct an irregular convolutional network and an LSTM network, assigning different weights to different positions within the convolutional kernels of the irregular convolutional network. For example, Figure 2 As shown, convolutional networks are commonly used in image processing to extract pixel region features. They employ two-dimensional convolution (with a 5×5 kernel) and use max pooling during the convolutional network processing.

[0066] In this invention, due to the time lag in the impact of temporal data, temporal misalignment can occur. To enable the network to recognize this misalignment, this embodiment applies an irregular convolutional network (ICNN) method to temporal feature extraction, improving prediction performance. ICNNs are used in some existing technologies to address the problem of irregular samples in target images during image recognition. This embodiment optimizes the process by assigning different weights to different positions within the convolution kernel in the ICNN. During pooling, the different positional weights significantly affect the target result. This embodiment uses an ICNN to assign different weights to different positions in the temporal data feature matrix. During the neural network training iterations, the positional weights are continuously trained and optimized, effectively reducing errors caused by temporal misalignment. To capture the temporal changes of the temporal data, an LSTM network is added after the ICNN.

[0067] Step 006: Input the historical operation and maintenance fault index sequence sample into the irregular convolutional network to obtain the convolution result.

[0068] In this embodiment of the invention, historical operation and maintenance fault index sequence samples are first input into the irregular convolutional network, and the variables at different positions in the convolution kernel are weighted to obtain the convolution result.

[0069] Step 010: The convolution result is processed into a one-dimensional vector, and the one-dimensional vector is input into the LSTM network. The vector is then passed through a fully connected layer and output to a one-dimensional fully connected layer to obtain the prediction result. Specifically, the convolution result is reshaped into a one-dimensional vector, then added to the LSTM network, and finally passed through a fully connected layer and output to a one-dimensional fully connected layer to obtain the prediction result.

[0070] Step 011: Based on the predicted results and the expected results corresponding to the historical operation and maintenance fault index sequence samples, optimize the weights and the parameters of the irregular convolutional network and LSTM network, and continue training to obtain the operation and maintenance fault prediction model.

[0071] Specifically, a loss function is calculated based on the predicted results and the expected results corresponding to the historical operation and maintenance fault index sequence samples. The loss function can be the mean squared error (MSE). The weights and the parameters of the irregular convolutional network and LSTM network are adjusted according to the loss function. Then, the historical operation and maintenance fault index sequence samples are input for training until the loss function converges or the preset number of iterations is reached, thus obtaining the operation and maintenance fault prediction model of this embodiment of the invention.

[0072] After obtaining the operation and maintenance fault prediction model, the sequence of operation and maintenance indicators to be predicted is input into the preset operation and maintenance fault prediction model to obtain the time-series data of the corresponding target service's business performance. This time-series data of the target service's business performance is the time-series data of the target service's business performance within a preset future time period. Based on this time-series data of business performance, it can be determined whether an operation and maintenance fault exists, thus obtaining the fault information of the target service. The preset future time period can refer to a preset time period after the current time. This preset future time period and the aforementioned preset current time period are periodic, and can be a periodic time such as an hour, day, or week; this embodiment of the invention does not impose specific limitations. For example, the time-series data of the target service's business performance for the next day can be predicted using the sequence of operation and maintenance indicators to be predicted for the current day. In this embodiment of the invention, when it is determined that a fault exists in the target service within a preset future time period based on the fault information of the target service, an alarm message is sent to the user.

[0073] This invention provides an embodiment of the method for obtaining a sequence of operation and maintenance (O&M) indicators to be predicted. This sequence includes time-series data corresponding to multiple O&M indicators. The O&M indicator sequence is then input into a pre-defined O&M fault prediction model to obtain fault information for the target service. The O&M fault prediction model includes an irregular convolutional network and an LSTM network. The O&M fault prediction model inputs historical O&M fault indicator sequence samples into the irregular convolutional network and processes the output convolutional result into a one-dimensional vector, which is then input into the LSTM network for training. This method can accurately predict O&M fault information.

[0074] Figure 4 A schematic diagram of the operation and maintenance fault analysis device provided in an embodiment of the present invention is shown. Figure 4 As shown, the device 300 includes:

[0075] The acquisition module 310 is used to acquire a sequence of operation and maintenance indicators to be predicted; the sequence of operation and maintenance indicators to be predicted includes time-series data corresponding to multiple operation and maintenance indicators.

[0076] The prediction module 320 is used to input the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service; wherein, the operation and maintenance fault prediction model includes an irregular convolutional network and an LSTM network; the operation and maintenance fault prediction model inputs the irregular convolutional network according to the historical operation and maintenance fault indicator sequence samples, and processes the output convolution result into a one-dimensional vector and inputs it into the LSTM network for training.

[0077] In an alternative embodiment, the device further includes:

[0078] The preprocessing module is used to preprocess the initial time-series data of historical operation and maintenance failure indicators to obtain the preprocessed time-series data corresponding to each historical operation and maintenance failure indicator.

[0079] The correlation analysis module is used to analyze the correlation between various historical operation and maintenance failure indicators and the business performance of the target business;

[0080] The filtering module is used to filter a preset number of historical operation and maintenance failure indicators from various historical operation and maintenance failure indicators based on the correlation.

[0081] The splicing module is used to splice the preprocessed time-series data corresponding to the preset number of historical operation and maintenance fault indicators to obtain a historical operation and maintenance fault indicator sequence sample.

[0082] In one optional approach, the preprocessing of the initial time-series data of historical operation and maintenance failure indicators to obtain preprocessed time-series data corresponding to each historical operation and maintenance failure indicator further includes:

[0083] Obtain time-series data of multiple historical operation and maintenance failure indicators;

[0084] The time-series data of the historical operation and maintenance failure indicators are padded with dates to obtain the padded time-series data;

[0085] The time series data after completion is normalized using the normalization method to obtain normalized time series data.

[0086] Based on the data deviation, abnormal data in the normalized time series data are identified and replaced with preset values ​​to obtain the preprocessed time series data corresponding to the historical operation and maintenance fault indicators.

[0087] In one alternative approach, the analysis of the correlation between various historical operational failure indicators and the business performance of the target service includes:

[0088] The correlation between each of the historical operation and maintenance failure indicators and the business performance of the target service is calculated using the Pearson correlation coefficient.

[0089] In one optional approach, the step of identifying abnormal data in the normalized time-series data based on data deviation and replacing it with a preset value to obtain preprocessed time-series data corresponding to the historical operation and maintenance fault indicators includes:

[0090] Calculate the median value of all data preceding any given data in the normalized time series data;

[0091] Based on the deviation of any data point in the normalized time series data from the median value, it is determined whether the data is abnormal.

[0092] When the data is abnormal, it will be replaced with a preset value.

[0093] In one optional approach, before inputting the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service in a future preset time period, the method includes:

[0094] Construct an irregular convolutional network and an LSTM network, and assign different weights to different positions within the convolutional kernels in the irregular convolutional network;

[0095] The historical operation and maintenance fault indicator sequence samples are input into the irregular convolutional network to obtain the convolution result;

[0096] The convolution result is processed into a one-dimensional vector, and the one-dimensional vector is input into the LSTM network. The vector is then output to a one-dimensional fully connected layer through a fully connected layer to obtain the prediction result.

[0097] Based on the predicted results and the expected results corresponding to the historical operation and maintenance fault index sequence samples, the weights, as well as the parameters of the irregular convolutional network and the LSTM network, are optimized and training continues to obtain the operation and maintenance fault prediction model.

[0098] The specific working process of the operation and maintenance fault analysis device in this embodiment of the invention is largely the same as the process steps of the above method embodiment, and will not be repeated here.

[0099] This invention provides an embodiment of the method for obtaining a sequence of operation and maintenance (O&M) indicators to be predicted. This sequence includes time-series data corresponding to multiple O&M indicators. The O&M indicator sequence is then input into a pre-defined O&M fault prediction model to obtain fault information for the target service. The O&M fault prediction model includes an irregular convolutional network and an LSTM network. The O&M fault prediction model inputs historical O&M fault indicator sequence samples into the irregular convolutional network and processes the output convolutional result into a one-dimensional vector, which is then input into the LSTM network for training. This method can accurately predict O&M fault information.

[0100] Figure 5 The diagram shows a structural schematic of a computer device provided in an embodiment of the present invention. The specific embodiments of the present invention do not limit the specific implementation of the computer device.

[0101] like Figure 5 As shown, the computer device may include: a processor 402, a communications interface 404, a memory 406, and a communications bus 408.

[0102] The processor 402, communication interface 404, and memory 406 communicate with each other via communication bus 408. Communication interface 404 is used to communicate with other network elements such as clients or other servers. The processor 402 executes program 410, specifically performing the relevant steps described in the embodiment of the operation and maintenance fault analysis method.

[0103] Specifically, program 410 may include program code, which includes computer-executable instructions.

[0104] Processor 402 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The computer device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or they may be processors of different types, such as one or more CPUs and one or more ASICs.

[0105] Memory 406 is used to store program 410. Memory 406 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0106] Specifically, program 410 can be called by processor 402 to cause the computer device to perform the following operations:

[0107] Obtain the sequence of operation and maintenance indicators to be predicted; the sequence of operation and maintenance indicators to be predicted includes time series data corresponding to multiple operation and maintenance indicators;

[0108] The sequence of operation and maintenance indicators to be predicted is input into a preset operation and maintenance fault prediction model to obtain fault information of the target service; wherein, the operation and maintenance fault prediction model includes an irregular convolutional network and an LSTM network; the operation and maintenance fault prediction model inputs the historical operation and maintenance fault indicator sequence samples into the irregular convolutional network, and processes the output convolution result into a one-dimensional vector and inputs it into the LSTM network for training.

[0109] In one optional approach, before inputting the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service in a future preset time period, the method includes:

[0110] Preprocess the initial time-series data of historical operation and maintenance failure indicators to obtain the preprocessed time-series data corresponding to each historical operation and maintenance failure indicator.

[0111] Analyze the correlation between various historical operation and maintenance failure indicators and the business performance of the target business;

[0112] Based on the correlation, a preset number of historical operation and maintenance failure indicators are selected from each historical operation and maintenance failure indicator;

[0113] The preprocessed time-series data corresponding to the preset number of historical operation and maintenance fault indicators are spliced ​​together to obtain a historical operation and maintenance fault indicator sequence sample.

[0114] In one optional approach, the preprocessing of the initial time-series data of historical operation and maintenance failure indicators to obtain preprocessed time-series data corresponding to each historical operation and maintenance failure indicator further includes:

[0115] Obtain time-series data of multiple historical operation and maintenance failure indicators;

[0116] The time-series data of the historical operation and maintenance failure indicators are padded with dates to obtain the padded time-series data;

[0117] The time series data after completion is normalized using the normalization method to obtain normalized time series data.

[0118] Based on the data deviation, abnormal data in the normalized time series data are identified and replaced with preset values ​​to obtain the preprocessed time series data corresponding to the historical operation and maintenance fault indicators.

[0119] In one alternative approach, the analysis of the correlation between various historical operational failure indicators and the business performance of the target service includes:

[0120] The correlation between each of the historical operation and maintenance failure indicators and the business performance of the target service is calculated using the Pearson correlation coefficient.

[0121] In one optional approach, the step of identifying abnormal data in the normalized time-series data based on data deviation and replacing it with a preset value to obtain preprocessed time-series data corresponding to the historical operation and maintenance fault indicators includes:

[0122] Calculate the median value of all data preceding any given data in the normalized time series data;

[0123] Based on the deviation of any data point in the normalized time series data from the median value, it is determined whether the data is abnormal.

[0124] When the data is abnormal, it will be replaced with a preset value.

[0125] In one optional approach, before inputting the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service in a future preset time period, the method includes:

[0126] Construct an irregular convolutional network and an LSTM network, and assign different weights to different positions within the convolutional kernels in the irregular convolutional network;

[0127] The historical operation and maintenance fault indicator sequence samples are input into the irregular convolutional network to obtain the convolution result;

[0128] The convolution result is processed into a one-dimensional vector, and the one-dimensional vector is input into the LSTM network. The vector is then output to a one-dimensional fully connected layer through a fully connected layer to obtain the prediction result.

[0129] Based on the predicted results and the expected results corresponding to the historical operation and maintenance fault index sequence samples, the weights, as well as the parameters of the irregular convolutional network and the LSTM network, are optimized and training continues to obtain the operation and maintenance fault prediction model.

[0130] This invention provides an embodiment of the method for obtaining a sequence of operation and maintenance (O&M) indicators to be predicted. This sequence includes time-series data corresponding to multiple O&M indicators. The O&M indicator sequence is then input into a pre-defined O&M fault prediction model to obtain fault information for the target service. The O&M fault prediction model includes an irregular convolutional network and an LSTM network. The O&M fault prediction model inputs historical O&M fault indicator sequence samples into the irregular convolutional network and processes the output convolutional result into a one-dimensional vector, which is then input into the LSTM network for training. This method can accurately predict O&M fault information.

[0131] This invention provides a computer-readable storage medium storing at least one executable instruction that, when executed on a computer device, causes the computer device to perform the maintenance fault analysis method in any of the above method embodiments.

[0132] Executable instructions can be used to cause computer devices to perform the following operations:

[0133] Obtain the sequence of operation and maintenance indicators to be predicted; the sequence of operation and maintenance indicators to be predicted includes time series data corresponding to multiple operation and maintenance indicators;

[0134] The sequence of operation and maintenance indicators to be predicted is input into a preset operation and maintenance fault prediction model to obtain fault information of the target service; wherein, the operation and maintenance fault prediction model includes an irregular convolutional network and an LSTM network; the operation and maintenance fault prediction model inputs the historical operation and maintenance fault indicator sequence samples into the irregular convolutional network, and processes the output convolution result into a one-dimensional vector and inputs it into the LSTM network for training.

[0135] In one optional approach, before inputting the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service in a future preset time period, the method includes:

[0136] Preprocess the initial time-series data of historical operation and maintenance failure indicators to obtain the preprocessed time-series data corresponding to each historical operation and maintenance failure indicator.

[0137] Analyze the correlation between various historical operation and maintenance failure indicators and the business performance of the target business;

[0138] Based on the correlation, a preset number of historical operation and maintenance failure indicators are selected from each historical operation and maintenance failure indicator;

[0139] The preprocessed time-series data corresponding to the preset number of historical operation and maintenance fault indicators are spliced ​​together to obtain a historical operation and maintenance fault indicator sequence sample.

[0140] In one optional approach, the preprocessing of the initial time-series data of historical operation and maintenance failure indicators to obtain preprocessed time-series data corresponding to each historical operation and maintenance failure indicator further includes:

[0141] Obtain time-series data of multiple historical operation and maintenance failure indicators;

[0142] The time-series data of the historical operation and maintenance failure indicators are padded with dates to obtain the padded time-series data;

[0143] The time series data after completion is normalized using the normalization method to obtain normalized time series data.

[0144] Based on the data deviation, abnormal data in the normalized time series data are identified and replaced with preset values ​​to obtain the preprocessed time series data corresponding to the historical operation and maintenance fault indicators.

[0145] In one alternative approach, the analysis of the correlation between various historical operational failure indicators and the business performance of the target service includes:

[0146] The correlation between each of the historical operation and maintenance failure indicators and the business performance of the target service is calculated using the Pearson correlation coefficient.

[0147] In one optional approach, the step of identifying abnormal data in the normalized time-series data based on data deviation and replacing it with a preset value to obtain preprocessed time-series data corresponding to the historical operation and maintenance fault indicators includes:

[0148] Calculate the median value of all data preceding any given data in the normalized time series data;

[0149] Based on the deviation of any data point in the normalized time series data from the median value, it is determined whether the data is abnormal.

[0150] When the data is abnormal, it will be replaced with a preset value.

[0151] In one optional approach, before inputting the sequence of operation and maintenance indicators to be predicted into a preset operation and maintenance fault prediction model to obtain fault information of the target service in a future preset time period, the method includes:

[0152] Construct an irregular convolutional network and an LSTM network, and assign different weights to different positions within the convolutional kernels in the irregular convolutional network;

[0153] The historical operation and maintenance fault indicator sequence samples are input into the irregular convolutional network to obtain the convolution result;

[0154] The convolution result is processed into a one-dimensional vector, and the one-dimensional vector is input into the LSTM network. The vector is then output to a one-dimensional fully connected layer through a fully connected layer to obtain the prediction result.

[0155] Based on the predicted results and the expected results corresponding to the historical operation and maintenance fault index sequence samples, the weights, as well as the parameters of the irregular convolutional network and the LSTM network, are optimized and training continues to obtain the operation and maintenance fault prediction model.

[0156] This invention provides an embodiment of the method for obtaining a sequence of operation and maintenance (O&M) indicators to be predicted. This sequence includes time-series data corresponding to multiple O&M indicators. The O&M indicator sequence is then input into a pre-defined O&M fault prediction model to obtain fault information for the target service. The O&M fault prediction model includes an irregular convolutional network and an LSTM network. The O&M fault prediction model inputs historical O&M fault indicator sequence samples into the irregular convolutional network and processes the output convolutional result into a one-dimensional vector, which is then input into the LSTM network for training. This method can accurately predict O&M fault information.

[0157] This invention provides an operation and maintenance fault analysis device for executing the above-described operation and maintenance fault analysis method.

[0158] This invention provides a computer program that can be called by a processor to cause a computer device to execute the operation and maintenance fault analysis method in any of the above method embodiments.

[0159] This invention provides a computer program product, which includes a computer program stored on a computer-readable storage medium. The computer program includes program instructions, which, when executed on a computer, cause the computer to perform the operation and maintenance fault analysis method in any of the above method embodiments.

[0160] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, the embodiments of the present invention are not directed to any particular programming language. It should be understood that the content of the invention described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of the invention.

[0161] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0162] Similarly, it should be understood that, in order to streamline the invention and aid in understanding one or more of the various aspects of the invention, features of the embodiments of the invention are sometimes grouped together in a single embodiment, figure, or description thereof in the above description of exemplary embodiments of the invention. However, this disclosure should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim.

[0163] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0164] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. An operation and maintenance fault analysis method, characterized in that, The method includes: An irregular convolutional network and an LSTM network are constructed, and different weights are assigned to different positions within the convolutional kernel of the irregular convolutional network. Historical maintenance failure index sequence samples are input into the irregular convolutional network to obtain the convolution result. The convolution result is processed into a one-dimensional vector, and the one-dimensional vector is input into the LSTM network, passed through a fully connected layer, and output to a one-dimensional fully connected layer to obtain the prediction result. Based on the prediction result and the expected result corresponding to the historical maintenance failure index sequence samples, the weights and the parameters of the irregular convolutional network and the LSTM network are optimized, and training continues to obtain the maintenance failure prediction model. Obtain the sequence of operation and maintenance indicators to be predicted; the sequence of operation and maintenance indicators to be predicted includes time series data corresponding to multiple operation and maintenance indicators; input the sequence of operation and maintenance indicators to be predicted into the operation and maintenance fault prediction model to obtain the fault information of the target service.

2. The method according to claim 1, characterized in that, Before inputting the sequence of operation and maintenance indicators to be predicted into the operation and maintenance fault prediction model to obtain the fault information of the target service, the method includes: Preprocess the initial time-series data of historical operation and maintenance failure indicators to obtain the preprocessed time-series data corresponding to each historical operation and maintenance failure indicator. Analyze the correlation between various historical operation and maintenance failure indicators and the business performance of the target business; Based on the correlation, a preset number of historical operation and maintenance failure indicators are selected from each historical operation and maintenance failure indicator; The preprocessed time-series data corresponding to the preset number of historical operation and maintenance fault indicators are spliced ​​together to obtain a historical operation and maintenance fault indicator sequence sample.

3. The method according to claim 2, characterized in that, The step of preprocessing the initial time-series data of historical operation and maintenance failure indicators to obtain preprocessed time-series data corresponding to each historical operation and maintenance failure indicator further includes: Obtain time-series data of multiple historical operation and maintenance failure indicators; The time-series data of the historical operation and maintenance failure indicators are padded with dates to obtain the padded time-series data; The time series data after completion is normalized using the normalization method to obtain normalized time series data. Based on the data deviation, abnormal data in the normalized time series data are identified and replaced with preset values ​​to obtain the preprocessed time series data corresponding to the historical operation and maintenance fault indicators.

4. The method according to claim 2, characterized in that, The analysis of the correlation between various historical operational failure indicators and the service performance of the target service includes: The correlation between each of the historical operation and maintenance failure indicators and the business performance of the target service is calculated using the Pearson correlation coefficient.

5. The method according to claim 3, characterized in that, The step of identifying abnormal data in the normalized time-series data based on data deviation and replacing them with preset values ​​to obtain preprocessed time-series data corresponding to the historical operation and maintenance fault indicators includes: Calculate the median value of all data preceding any given data in the normalized time series data; Based on the deviation of any data point in the normalized time series data from the median value, it is determined whether the data is abnormal. When the data is abnormal, it will be replaced with a preset value.

6. A maintenance fault analysis device, characterized in that, The device includes: The acquisition module is used to acquire the sequence of operation and maintenance indicators to be predicted; the sequence of operation and maintenance indicators to be predicted includes time-series data corresponding to multiple operation and maintenance indicators. The prediction module is used to input the sequence of operation and maintenance indicators to be predicted into the operation and maintenance fault prediction model to obtain the fault information of the target service. The device is further configured to: construct an irregular convolutional network and an LSTM network, assign different weights to different positions within the convolutional kernel of the irregular convolutional network; input historical maintenance failure index sequence samples into the irregular convolutional network to obtain convolution results; process the convolution results into a one-dimensional vector, input the one-dimensional vector into the LSTM network, output it through a fully connected layer to a one-dimensional fully connected layer to obtain prediction results; optimize the weights and the parameters of the irregular convolutional network and the LSTM network based on the prediction results and the expected results corresponding to the historical maintenance failure index sequence samples, and continue training to obtain an maintenance failure prediction model.

7. The apparatus according to claim 6, characterized in that, The device further includes: The preprocessing module is used to preprocess the initial time-series data of historical operation and maintenance failure indicators to obtain the preprocessed time-series data corresponding to each historical operation and maintenance failure indicator. The correlation analysis module is used to analyze the correlation between various historical operation and maintenance failure indicators and the business performance of the target business; The filtering module is used to filter a preset number of historical operation and maintenance failure indicators from various historical operation and maintenance failure indicators based on the correlation. The splicing module is used to splice the preprocessed time-series data corresponding to the preset number of historical operation and maintenance fault indicators to obtain a historical operation and maintenance fault indicator sequence sample.

8. A computer device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction, which causes the processor to perform the operation and maintenance fault analysis method as described in any one of claims 1-5.

9. A computer-readable storage medium, characterized in that, The storage medium stores at least one executable instruction, which, when executed on a computer device, causes the computer device to perform the operation of the maintenance fault analysis method as described in any one of claims 1-5.

Images