Model training method, operation data prediction method and storage medium

Abstract

The embodiment of the invention discloses a model training method, an operation data prediction method and a storage medium. The model training method comprises the following steps: acquiring historical operation data of target equipment in a historical time period, and feature data which corresponds to abnormal operation data in the historical operation data and is used for representing external features of the target equipment; performing correlation analysis on the abnormal operation data and the feature data to obtain a correlation coefficient; and based on the correlation coefficient and the historical operation data, determining a target prediction model used for performing operation data prediction of the target equipment. According to the technical scheme provided by the embodiment of the invention, the accuracy of running data prediction of the target equipment can be improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of data processing, and in particular to a model training method, an operation data prediction method and a storage medium. Background Art

[0002] Gas insulated switchgear (GIS) equipment is widely used in the power industry due to its advantages such as small size, convenient maintenance, high reliability and strong environmental adaptability.

[0003] However, due to the large number of components and complex structure of GIS equipment, it is currently impossible to accurately predict the operating data of GIS equipment, which needs to be solved urgently. Summary of the invention

[0004] Embodiments of the present invention provide a model training method, an operation data prediction method and a storage medium to improve the accuracy of operation data prediction of a target device.

[0005] According to one aspect of the present invention, a model training method is provided, which may include:

[0006] Acquire historical operation data of the target device within a historical time period, and feature data corresponding to abnormal operation data in the historical operation data and used to characterize external features of the target device;

[0007] Conduct correlation analysis on abnormal operation data and characteristic data to obtain correlation coefficient;

[0008] Based on the correlation coefficient and the historical operation data, a target prediction model for predicting the operation data of the target device is determined.

[0009] According to another aspect of the present invention, a method for predicting operation data is provided, which may include:

[0010] Obtain a target prediction model trained by the model training method provided by any embodiment of the present invention;

[0011] Acquire first operation data of the target device at a current time point and second operation data at a preset number of historical time points before the current time point;

[0012] Based on the first operation data, the second operation data and the target prediction model, future operation data of the target device in a future time period after the current time point is predicted.

[0013] According to another aspect of the present invention, a model training device is provided, which may include:

[0014] A feature data acquisition module, used to acquire historical operation data of the target device within a historical time period, and feature data corresponding to abnormal operation data in the historical operation data and used to characterize external features of the target device;

[0015] A correlation coefficient obtaining module is used to perform correlation analysis on abnormal operation data and characteristic data to obtain a correlation coefficient;

[0016] The target prediction model determination module is used to determine the target prediction model for predicting the operation data of the target device based on the correlation coefficient and the historical operation data.

[0017] According to another aspect of the present invention, there is provided an operation data prediction device, which may include:

[0018] A target prediction model acquisition module, used to acquire a target prediction model trained by a model training method provided by any embodiment of the present invention;

[0019] A second operation data acquisition module, used to acquire first operation data of the target device at a current time point and second operation data at a preset number of historical time points before the current time point;

[0020] The future operation data prediction module is used to predict the future operation data of the target device in a future time period after the current time point based on the first operation data, the second operation data and the target prediction model.

[0021] According to another aspect of the present invention, a computer-readable storage medium is provided, on which computer instructions are stored, and the computer instructions are used to enable a processor to implement the model training method or the running data prediction method provided by any embodiment of the present invention when executed.

[0022] The technical solution of the embodiment of the present invention obtains the historical operation data of the target device in the historical time period, and the feature data corresponding to the abnormal operation data in the historical operation data for characterizing the external characteristics of the target device, so as to comprehensively consider the external influences that the target device may be subject to when operating abnormally; performs correlation analysis on the abnormal operation data and the feature data to obtain the correlation coefficient, so as to characterize the external influences that the target device may be subject to when operating abnormally through the correlation coefficient; based on the correlation coefficient and the historical operation data, determines the target prediction model for predicting the operation data of the target device, so as to improve the accuracy of the prediction of the operation data of the target device through the target prediction model obtained through training. The above technical solution, by performing correlation analysis on the abnormal operation data and the feature data corresponding to it, and obtaining the target prediction model through the correlation coefficient and the historical operation data training, can comprehensively consider the external influences that the target device may be subject to when operating abnormally, and improve the accuracy of the prediction of the operation data of the target device.

[0023] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 is a flow chart of a model training method provided according to an embodiment of the present invention;

[0026] Figure 2 is a flowchart of another model training method provided according to an embodiment of the present invention;

[0027] Figure 3 is a flow chart of an operation data prediction method provided according to an embodiment of the present invention;

[0028] Figure 4 is a fault diagnosis flow chart according to a specific example of a method for predicting operating data provided by an embodiment of the present invention;

[0029] Figure 5 is a schematic diagram of a specific example fault diagnosis process in an operation data prediction method provided by an embodiment of the present invention;

[0030] Figure 6 is a structural block diagram of a model training device provided according to an embodiment of the present invention;

[0031] Figure 7 is a structural block diagram of an operation data prediction device provided according to an embodiment of the present invention;

[0032] Figure 8 It is a structural diagram of an electronic device for implementing the model training method or the operation data prediction method of an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. The situations of "target", "original", etc. are similar and will not be repeated here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] Before introducing the embodiment of the present invention, the relevant contents described in the background technology are first exemplarily described to better understand the model training method described in the embodiment of the present invention.

[0036] GIS equipment has been widely recognized by the power industry due to its advantages such as small size, convenient maintenance, high reliability and strong environmental adaptability. However, with the increase in installed capacity and operating time of GIS equipment in recent years, various abnormalities and failures will inevitably occur in the equipment. Due to the sealing of GIS equipment, when an abnormality occurs inside the equipment, it is difficult to detect it directly and in time through human senses; and GIS equipment has many components, complex structure, and various types of failures. How to quickly and accurately identify the defect type and locate the defective air chamber is of great significance for carrying out GIS equipment maintenance and ensuring equipment safety.

[0037] In practical applications, workers often use online monitoring systems to monitor GIS machinery. GIS machinery monitoring data belongs to time series. Generally speaking, any time series will have irregular components. Time series without trend and seasonal components, that is, stationary time series, only contain random components. Random fluctuations can be eliminated through smoothing. Therefore, this type of prediction method is also called smoothing prediction method.

[0038] Exponential smoothing is an algorithm in time series prediction. The data predicted by this algorithm refers to a set of predicted values ​​of the same data variable arranged in the order of events. There are two characteristics of time series: one is to predict the development of future data based on the trend of historical data; the other is that the data is variable, and there are four types: trend, periodicity, randomness and comprehensiveness. According to the number of iterations of the exponential smoothing method, the number of conventional exponential smoothing methods is one to three. The first and second exponential smoothing are suitable for linear models. When the monitoring data shows nonlinear trend changes, the second exponential smoothing will have errors, so the third exponential smoothing method is introduced. The third exponential smoothing method is to perform a smoothing calculation on the basis of the second exponential smoothing. The traditional third exponential smoothing method uses static parameters and static coefficients to bring into the calculation model. The advantage of the static third exponential smoothing model is that it retains seasonal information on the basis of the second exponential smoothing, so that it can predict time series with seasonality. The third exponential smoothing adds a new parameter to represent the trend after smoothing. Since the traditional third exponential smoothing does not take into account the deviation of the outside world, the static third exponential smoothing method is widely used in stable data models. Moreover, in the traditional static triple exponential smoothing model, the smoothing coefficient is a fixed constant. Once the coefficient is determined, it cannot be dynamically modified. This leads to the inability to accurately and reliably predict the operating data of GIS equipment, which needs to be solved urgently.

[0039] Figure 1 It is a flow chart of a model training method provided in an embodiment of the present invention. This embodiment is applicable to the case of training the operation data prediction model of the target device. The method can be executed by the model training device provided in an embodiment of the present invention, and the device can be implemented by software and / or hardware. The device can be integrated on an electronic device, and the electronic device can be various user terminals or servers.

[0040] See also Figure 1 The method of the embodiment of the present invention specifically includes the following steps:

[0041] S110: Acquire historical operation data of the target device within a historical time period, and feature data corresponding to abnormal operation data in the historical operation data and used to characterize external features of the target device.

[0042] The target device can be understood as the device for which the operation data prediction is to be performed. Optionally, the target device is a gas-insulated metal-enclosed switchgear located in the power system. By predicting the operation data of the GIS device, faults can be quickly and accurately identified to ensure the safe operation of the GIS device.

[0043] Historical operation data can be understood as time series data generated during the operation of GIS equipment in historical time periods. The historical operation data includes historical operation data when the GIS equipment operates normally and abnormal operation data when the GIS equipment operates abnormally. When the GIS equipment operates abnormally, feature data representing the external characteristics of the target equipment corresponding to the abnormal operation data can be obtained, for example, feature data of external characteristics such as weather, temperature, and related voltage when the GIS equipment operates abnormally.

[0044] S120. Perform correlation analysis on the abnormal operation data and the characteristic data to obtain a correlation coefficient.

[0045] Among them, the correlation between the feature data and the abnormal operation data can be obtained by performing a correlation analysis on the abnormal operation data and the feature data, and represented by a correlation coefficient. Optionally, the correlation coefficient is obtained based on the abnormal operation data and the feature data, including: performing one-hot encoding on the feature data to obtain an encoding result; based on the encoding result, performing a correlation analysis on the abnormal operation data and the feature data to obtain a correlation coefficient. That is, first perform one-hot encoding on the feature data, and perform a correlation analysis on the obtained encoding result with the abnormal operation data to obtain a correlation coefficient. Optionally, the Pearson correlation coefficient can be used for correlation analysis, and the specific formula is as follows:

[0046]

[0047] Among them, x is the abnormal operation data, y is the encoding result, x μ ,y μ are the means of x and y respectively, δx and δy are the standard deviations of x and y respectively.

[0048] S130 . Determine a target prediction model for predicting the operation data of the target device based on the correlation coefficient and the historical operation data.

[0049] The constructed initial model can be trained by using correlation coefficients and historical operation data, thereby determining a target prediction model for predicting the operation data of the target device.

[0050] The technical solution of the embodiment of the present invention obtains the historical operation data of the target device in the historical time period, and the feature data corresponding to the abnormal operation data in the historical operation data for characterizing the external characteristics of the target device, so as to comprehensively consider the external influences that the target device may be subject to when operating abnormally; performs correlation analysis on the abnormal operation data and the feature data to obtain the correlation coefficient, so as to characterize the external influences that the target device may be subject to when operating abnormally through the correlation coefficient; based on the correlation coefficient and the historical operation data, determines the target prediction model for predicting the operation data of the target device, so as to improve the accuracy of the prediction of the operation data of the target device through the target prediction model obtained through training. The above technical solution, by performing correlation analysis on the abnormal operation data and the feature data corresponding to it, and obtaining the target prediction model through the correlation coefficient and the historical operation data training, can comprehensively consider the external influences that the target device may be subject to when operating abnormally, and improve the accuracy of the prediction of the operation data of the target device.

[0051] Figure 2 It is a flowchart of another model training method provided in an embodiment of the present invention. This embodiment is optimized based on the above-mentioned technical solutions. In this embodiment, optionally, based on the correlation coefficient and the historical operation data, a target prediction model for predicting the operation data of the target device is determined, including: obtaining an initial prediction model established based on a triple exponential smoothing algorithm; determining a training data set for training the initial prediction model from the historical operation data; training the initial prediction model based on the training data set and the correlation coefficient, and determining the target smoothing coefficient applied by the initial prediction model during the third smoothing process; determining the target prediction model based on the target smoothing coefficient. Among them, the explanations of the terms that are the same or corresponding to the above-mentioned embodiments are not repeated here.

[0052] See also Figure 2 The method of this embodiment may specifically include the following steps:

[0053] S210: Acquire historical operation data of the target device within a historical time period, and feature data corresponding to abnormal operation data in the historical operation data and used to characterize external features of the target device.

[0054] S220: Perform correlation analysis on the abnormal operation data and the characteristic data to obtain a correlation coefficient.

[0055] S230: Obtain an initial prediction model based on a cubic exponential smoothing algorithm.

[0056] Among them, the triple exponential smoothing algorithm can retain seasonal information on the basis of quadratic exponential smoothing, so that it can predict time series with seasonality. The initial prediction model established based on the triple exponential smoothing algorithm can make the operating data predicted by the trained target prediction model more accurate.

[0057] S240: Determine a training data set for training an initial prediction model from historical operation data.

[0058] Part or all of the historical operation data may be intercepted from the historical operation data as a training data set, and the training data set may include normal operation data and abnormal operation data to cover all possible operation data.

[0059] S250: Based on the training data set and the correlation coefficient, the initial prediction model is trained to determine a target smoothing coefficient applied by the initial prediction model during the third smoothing process.

[0060] The target smoothing coefficient can be understood as the smoothing coefficient of the third exponential smoothing process in the initial prediction model obtained through training. Based on the training data set and the correlation coefficient, the initial prediction model is trained, and the smoothing coefficient of the third exponential smoothing process in the initial prediction model can be adjusted through the prediction error, thereby determining the target smoothing coefficient applied by the initial prediction model during the third smoothing process.

[0061] S260: Determine a target prediction model based on the target smoothing coefficient.

[0062] The technical solution of the embodiment of the present invention determines the target smoothing coefficient applied by the initial prediction model during the third smoothing process through the training data set and the correlation coefficient, and can dynamically correct the target smoothing coefficient to improve the accuracy of the operation data prediction.

[0063] On this basis, an optional technical solution is provided, which trains the initial prediction model based on the training data set and the correlation coefficient, and determines the target smoothing coefficient applied by the initial prediction model during the third smoothing process, including: determining from the training data set the actual operating data corresponding to the prediction time point for operating data prediction using the initial prediction model; determining from the training data set the training operating data corresponding to a preset number of training time points that are before the prediction time point and adjacent to the prediction time point; based on the training operating data and the correlation coefficient, using the initial prediction model, predicting the predicted operating data corresponding to the prediction time point, and determining the target smoothing coefficient applied by the initial prediction model during the third smoothing process based on the predicted operating data and the actual operating data.

[0064] Among them, from the training data set, determine the real operating data corresponding to the prediction time point for which the initial prediction model is to be used to predict the operating data; from the training data set, determine the training operating data corresponding to a preset number of training time points before the prediction time point and adjacent to the prediction time point. For example, if the operating data of point N is to be predicted, obtain the training operating data corresponding to the three time points before point N from the training data set [X N-3 ,X N-2 ,X N-1 ]. Then, the training operation data is input into the initial prediction model and substituted into the correlation coefficient for calculation, so as to predict the predicted operation data corresponding to the predicted time point. Based on the predicted operation data and the actual operation data, the prediction error can be determined, and then the smoothing coefficient in the initial prediction model can be adjusted through the prediction error, so as to determine the target smoothing coefficient applied by the initial prediction model in the third smoothing process. Optionally, based on the training running data and the correlation coefficient, the initial prediction model is used to predict the predicted running data corresponding to the predicted time point, and based on the predicted running data and the actual running data, a target smoothing coefficient applied by the initial prediction model during the third smoothing process is determined, including: based on the training running data and the correlation coefficient, the initial prediction model is used to predict the predicted running data corresponding to the predicted time point; based on the predicted running data and the actual running data, a prediction error is determined; when the prediction error is greater than a preset error threshold, a preset value is added to the current smoothing coefficient applied by the initial prediction model during the third smoothing process to obtain an adjusted smoothing coefficient, and the adjusted smoothing coefficient is used as the current smoothing coefficient to update the initial prediction model, and the step of predicting the predicted running data corresponding to the predicted time point based on the training running data and the correlation coefficient and using the initial prediction model is repeated; when the prediction error is less than or equal to the preset error threshold, the current smoothing coefficient is used as the target smoothing coefficient applied by the initial prediction model during the third smoothing process.

[0065] The predicted operating data is obtained by predicting the training operating data and the correlation coefficient. Optionally, when predicting the predicted operating data, the predicted operating data at the predicted time point can be predicted by the training operating data, and then the predicted operating data at the predicted time point is used as the training operating data to predict the operating data at the time point after the predicted time point, so that the predicted operating data for a time period can be predicted. After that, the prediction error between the predicted operating data and the actual operating data can be determined. Optionally, the prediction error can include the root mean square error and the relative error. The formula for calculating the root mean square error is as follows:

[0066]

[0067] The formula for calculating the relative error is as follows:

[0068]

[0069] Among them, t is the forecast period, s t is the forecast operation data for the forecast period, x t is the actual operation data corresponding to each time point in the prediction period, σ is the root mean square error, δ is the relative error, Δ represents the absolute error between the actual value and the predicted value, and L is the actual operation data.

[0070] When the prediction error is greater than the preset error threshold, a preset value is added to the current smoothing coefficient applied by the initial prediction model during the third smoothing process to obtain an adjusted smoothing coefficient after adjusting the current smoothing coefficient, and the adjusted smoothing coefficient is used as the current smoothing coefficient to update the initial prediction model, and the steps of predicting the predicted running data corresponding to the predicted time point using the initial prediction model based on the training running data and the correlation coefficient are repeated; when the prediction error is less than or equal to the preset error threshold, the current smoothing coefficient is used as the target smoothing coefficient applied by the initial prediction model during the third smoothing process.

[0071] The above technical solution calculates the prediction error and continuously adjusts the smoothing coefficient in the initial prediction model based on a preset error threshold until the prediction error is less than or equal to the preset error threshold, thereby determining an accurate target smoothing coefficient and improving the accuracy of the target prediction model prediction.

[0072] On this basis, an optional technical solution is to use an initial prediction model based on training operation data and correlation coefficients to predict the predicted operation data corresponding to the prediction time point, including: determining the initial smoothing coefficient in the initial prediction model based on the abnormal situation of the training operation data; using the initial prediction model with the determined initial smoothing coefficient and the correlation coefficient to predict the predicted operation data corresponding to the prediction time point.

[0073] Among them, the initial smoothing coefficient can be understood as the initial smoothing coefficient in the initial prediction model. Optionally, the initial smoothing coefficient can be determined by the abnormal situation of the training operation data, such as the training operation data is normal operation data or abnormal operation data. Optionally, the initial smoothing coefficient can give priority to the empirical value golden ratio coefficients 0.618 and 0.382, and 0.618 is used as the initial smoothing coefficient when the training operation data is abnormal operation data, and 0.382 is used as the initial smoothing coefficient when the training operation data is normal operation data. Finally, the initial prediction model with the determined initial smoothing coefficient and the correlation coefficient are used to predict the predicted operation data corresponding to the prediction time point.

[0074] The above technical solution determines the initial smoothing coefficient in the initial prediction model based on the abnormal situation of the training operation data, which can effectively reduce the repetitive process of determining the target smoothing coefficient and reduce the calculation amount of model training.

[0075] Figure 3 1 is a flowchart of an operation data prediction method provided in an embodiment of the present invention. This embodiment is applicable to the case of predicting the operation data of the target device. The method can be executed by the operation data prediction device provided in an embodiment of the present invention, and the device can be implemented by software and / or hardware. The device can be integrated in an electronic device, and the electronic device can be various user terminals or servers.

[0076] See also Figure 6 The method of the embodiment of the present invention specifically includes the following steps:

[0077] S310: Obtain a target prediction model trained by the model training method provided by any embodiment of the present invention.

[0078] S320: Acquire first operation data of the target device at the current time point and second operation data at a preset number of historical time points before the current time point.

[0079] Among them, before predicting the operating data of the future time period through the target prediction model, it is necessary to obtain the first operating data of the target device at the current time point and the second operating data of a preset number of historical time points before the current time point, so as to improve the accuracy of the prediction through the operating data closest to the prediction time period.

[0080] S330: Based on the first operation data, the second operation data and the target prediction model, predict and obtain future operation data of the target device in a future time period after the current time point.

[0081] Among them, the first operation data and the second operation data can be input into the target prediction model to predict the predicted operation data of the target device at a time point after the current time point. Optionally, the predicted operation data at the time point after the current time point can be used as the input of the target prediction model, and then the operation data of the next time point after the current time point can be predicted, so that the future operation data of the future time period can be obtained. Optionally, after predicting the future operation data of the future time period after the current time point, it also includes: based on the future operation data, diagnosing the fault of the target device. The operation status of the target device in the future time period can be obtained by the operation data, based on which the possible faults of the target device can be judged, and the target device can be adjusted in time to achieve timely discovery and effective prevention of faults.

[0082] The technical solution of the embodiment of the present invention can accurately predict the future operation data of the target device in the future time period through the target prediction model obtained by training.

[0083] In order to better understand the above-mentioned technical solutions as a whole, the following is an exemplary description of them in combination with specific examples. In this embodiment, the target device is a GIS device, and the flowchart of GIS device fault diagnosis is as follows: Figure 4 As shown in the figure, the process diagram of GIS equipment fault diagnosis is as follows Figure 5 As shown, the specific steps are as follows:

[0084] S1. Obtain historical operation data of GIS equipment and characteristic data corresponding to abnormal operation data in the historical operation data.

[0085] S2. Perform correlation analysis on abnormal operation data and characteristic data to obtain the influence of characteristic data on GIS equipment and present it in the form of correlation coefficient.

[0086] S3. Adjust the initial smoothing coefficient in the initial prediction model based on the correlation coefficient and historical operation data to determine the target smoothing coefficient.

[0087] 1. Determine the prediction time point for the operation data prediction of the initial prediction model. Based on the historical operation data, determine the training operation data corresponding to the three time points before the prediction time point. Use 0.618 as the initial smoothing coefficient when the training operation data is abnormal operation data, and use 0.382 as the initial smoothing coefficient when the training operation data is normal operation data. Create a coefficient value array F[m,n], store the initial value of the smoothing coefficient α, and the correlation coefficient P, where m is the predicted GIS machine serial number and n is the predicted time point;

[0088] 2. Determine the time period T to be predicted, bring the training operation data and F[m,n] into the initial prediction model, and predict the operation data corresponding to the time period T. The specific formula is as follows:

[0089]

[0090] x t+T =A t +B t T+C t T 2

[0091] A t =3S t (1) -3S t (2) +S t (3)

[0092]

[0093] Among them, X N-3 ,X N-2 ,X N-1 The data of the first three time points of the forecast data, each time the forecast value is calculated, the three forecast parameters must be updated; constant A t , B t , C t is the exponential smoothing coefficient of the tth period. Due to the self-updating of coefficients and parameters, each prediction is different; are the first, second, and third exponential smoothing values ​​of the tth period, α m，n is the weight coefficient under m and n, that is, the correlation coefficient P, x t is the actual value of the current t cycle, x t+T It is the predicted value for the future time period t+T.

[0094] 3. Calculate the prediction error based on the predicted operation data and the actual operation data, set the interval of α to 0.01, and when the prediction error is greater than the preset error threshold, add 0.01 to α and perform the above prediction process again until the prediction error is less than or equal to the preset error threshold to obtain the target smoothing coefficient α.

[0095] S4. Based on the target prediction model with the target smoothing coefficient determined, the future operation data of the GIS equipment is predicted to obtain the fault diagnosis result.

[0096] 1. Predict the future operation data of the GIS equipment based on the target prediction model. When the number of predictions i meets the number requirement, recalculate the correlation coefficient, and continue to point to step S3 based on the recalculated correlation coefficient.

[0097] 2. Obtain fault diagnosis results based on the predicted future operating data.

[0098] The above specific examples can improve the accuracy of operation data prediction of GIS equipment.

[0099] Figure 6 This is a structural block diagram of a model training device provided in an embodiment of the present invention, and the device is used to execute the model training method provided in any of the above embodiments. The device and the model training method of the above embodiments belong to the same inventive concept, and the details not described in detail in the embodiment of the model training device can refer to the embodiment of the above model training method. Figure 6 The device may specifically include: a feature data acquisition module 410, a correlation coefficient acquisition module 420 and a target prediction model determination module 430.

[0100] The feature data acquisition module 410 is used to acquire the historical operation data of the target device in the historical time period, and the feature data corresponding to the abnormal operation data in the historical operation data and used to characterize the external characteristics of the target device;

[0101] A correlation coefficient obtaining module 420 is used to perform correlation analysis on the abnormal operation data and the characteristic data to obtain a correlation coefficient;

[0102] The target prediction model determination module 430 is used to determine a target prediction model for predicting the operation data of the target device based on the correlation coefficient and the historical operation data.

[0103] Optionally, the target prediction model determination module 430 includes:

[0104] The initial prediction model acquisition submodule is used to obtain the initial prediction model established based on the cubic exponential smoothing algorithm;

[0105] A training data set determination submodule is used to determine a training data set for training an initial prediction model from historical operation data;

[0106] A target smoothing coefficient determination submodule is used to train the initial prediction model based on the training data set and the correlation coefficient, and determine the target smoothing coefficient applied by the initial prediction model during the third smoothing process;

[0107] The target prediction model determination submodule is used to determine the target prediction model based on the target smoothing coefficient.

[0108] On this basis, an optional target smoothing coefficient determination submodule includes:

[0109] A real operation data determination unit, used to determine, from the training data set, the real operation data corresponding to the prediction time point for operation data prediction using the initial prediction model;

[0110] A training operation data determination unit, used to determine, from the training data set, training operation data corresponding to a preset number of training time points that are before the predicted time point and adjacent to the predicted time point;

[0111] The target smoothing coefficient determination unit is used to predict the predicted running data corresponding to the predicted time point based on the training running data and the correlation coefficient using the initial prediction model, and determine the target smoothing coefficient applied by the initial prediction model during the third smoothing process based on the predicted running data and the actual running data.

[0112] On this basis, an optional target smoothing coefficient determination unit includes:

[0113] An initial smoothing coefficient determination subunit, used to determine an initial smoothing coefficient in an initial prediction model based on abnormal conditions of the training operation data;

[0114] The predicted operation data obtaining subunit is used to predict the predicted operation data corresponding to the predicted time point by using the initial prediction model with the determined initial smoothing coefficient and the correlation coefficient.

[0115] Alternatively, the target smoothing coefficient determination unit further includes:

[0116] The prediction operation data prediction subunit is used to predict the prediction operation data corresponding to the prediction time point based on the training operation data and the correlation coefficient using the initial prediction model;

[0117] A prediction error determination subunit, for determining a prediction error based on the predicted operation data and the actual operation data;

[0118] The initial prediction model updating subunit is used to add a preset value to the current smoothing coefficient applied by the initial prediction model during the third smoothing process to obtain an adjusted smoothing coefficient when the prediction error is greater than a preset error threshold, and use the adjusted smoothing coefficient as the current smoothing coefficient to update the initial prediction model, and repeatedly perform the steps of predicting the prediction operation data corresponding to the prediction time point using the initial prediction model based on the training operation data and the correlation coefficient;

[0119] The target smoothing coefficient determination subunit is used to use the current smoothing coefficient as the target smoothing coefficient applied by the initial prediction model during the third smoothing process when the prediction error is less than or equal to the preset error threshold.

[0120] Another optional technical solution, the correlation coefficient obtaining module 420 includes:

[0121] The encoding result obtaining submodule is used to perform one-hot encoding on the feature data to obtain the encoding result;

[0122] The correlation coefficient obtaining submodule is used to perform correlation analysis on abnormal operation data and characteristic data based on the encoding result to obtain the correlation coefficient.

[0123] Based on any of the above devices, the optional target device is a gas-insulated metal-enclosed switchgear located in a power system.

[0124] The model training device provided by the embodiment of the present invention obtains the historical operation data of the target device in the historical time period through the feature data acquisition module, and the feature data corresponding to the abnormal operation data in the historical operation data for characterizing the external characteristics of the target device, so as to comprehensively consider the external influences that the target device may be subject to when operating abnormally; through the correlation coefficient acquisition module, the abnormal operation data and the feature data are subjected to correlation analysis to obtain the correlation coefficient, so as to characterize the external influences that the target device may be subject to when operating abnormally through the correlation coefficient; through the target prediction model determination module, based on the correlation coefficient and the historical operation data, the target prediction model used for predicting the operation data of the target device is determined, so as to improve the accuracy of the prediction of the operation data of the target device through the target prediction model obtained through the training. The above-mentioned device can comprehensively consider the external influences that the target device may be subject to when operating abnormally, and improve the accuracy of the prediction of the operation data of the target device, by performing correlation analysis on the abnormal operation data and the feature data corresponding to it, and obtaining the target prediction model through the correlation coefficient and the historical operation data training.

[0125] The model training device provided in the embodiment of the present invention can execute the model training method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0126] It is worth noting that in the embodiment of the above-mentioned model training device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0127] Figure 7 This is a structural block diagram of an operation data prediction device provided in an embodiment of the present invention. The device is used to execute the operation data prediction method provided in any of the above embodiments. The device and the operation data prediction method of the above embodiments belong to the same inventive concept. For details not described in detail in the embodiment of the operation data prediction device, reference can be made to the embodiment of the above operation data prediction method. Figure 7 The device may specifically include: a target prediction model acquisition module 510, a second operation data acquisition module 520 and a future operation data acquisition module 530. Among them,

[0128] A target prediction model acquisition module 510 is used to acquire a target prediction model trained by a model training method provided by any embodiment of the present invention;

[0129] The second operation data acquisition module 520 is used to acquire the first operation data of the target device at the current time point and the second operation data at a preset number of historical time points before the current time point;

[0130] The future operation data obtaining module 530 is used to predict and obtain the future operation data of the target device in a future time period after the current time point based on the first operation data, the second operation data and the target prediction model.

[0131] On this basis, optionally, after predicting the future operation data of the future time period after the current time point, it also includes:

[0132] The fault diagnosis module is used to perform fault diagnosis on the target device based on future operation data.

[0133] The operation data prediction device provided by the embodiment of the present invention obtains the target prediction model trained by the model training method provided by any embodiment of the present invention through the target prediction model acquisition module; obtains the first operation data of the target device at the current time point and the second operation data at a preset number of historical time points before the current time point through the second operation data acquisition module; and predicts the future operation data of the target device in the future time period after the current time point based on the first operation data, the second operation data and the target prediction model through the future operation data acquisition module. The above-mentioned operation data prediction device can realize the accurate prediction of the operation data of the target device.

[0134] The operation data prediction device provided in the embodiment of the present invention can execute the operation data prediction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0135] It is worth noting that in the embodiment of the above-mentioned operation data prediction device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0136] Figure 8 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0137] like Figure 8As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0138] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0139] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as a model training method or an operation data prediction method.

[0140] In some embodiments, the model training method or the running data prediction method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the model training method or the running data prediction method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the model training method or the running data prediction method in any other appropriate manner (e.g., by means of firmware).

[0141] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0142] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0143] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0144] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0145] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0146] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0147] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0148] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A model training method, characterized in that: include: Acquire historical operation data of the target device within a historical time period, and feature data corresponding to abnormal operation data in the historical operation data and used to characterize external features of the target device; Performing correlation analysis on the abnormal operation data and the characteristic data to obtain a correlation coefficient; Based on the correlation coefficient and the historical operation data, a target prediction model for predicting the operation data of the target device is determined.

2. The method according to claim 1, characterized in that The step of determining a target prediction model for predicting the operation data of the target device based on the correlation coefficient and the historical operation data includes: Obtain the initial forecast model based on the triple exponential smoothing algorithm; Determining a training data set for training the initial prediction model from the historical operation data; Based on the training data set and the correlation coefficient, the initial prediction model is trained to determine a target smoothing coefficient applied by the initial prediction model during the third smoothing process; Based on the target smoothing coefficient, a target prediction model is determined.

3. The method according to claim 2, characterized in that The step of training the initial prediction model based on the training data set and the correlation coefficient and determining a target smoothing coefficient applied by the initial prediction model during the third smoothing process includes: Determining, from the training data set, actual operating data corresponding to a prediction time point for operating data prediction using the initial prediction model; Determining, from the training data set, training operation data corresponding to a preset number of training time points that are before the predicted time point and adjacent to the predicted time point; Based on the training operation data and the correlation coefficient, the initial prediction model is used to predict the predicted operation data corresponding to the predicted time point, and based on the predicted operation data and the actual operation data, the target smoothing coefficient applied by the initial prediction model during the third smoothing process is determined.

4. The method according to claim 3, characterized in that The step of predicting the predicted operating data corresponding to the predicted time point based on the training operating data and the correlation coefficient using the initial prediction model includes: Determining an initial smoothing coefficient in the initial prediction model based on the abnormality of the training operation data; The initial prediction model for which the initial smoothing coefficient has been determined and the correlation coefficient are used to predict the predicted operating data corresponding to the predicted time point.

5. The method according to claim 3, characterized in that: The method of predicting the predicted operation data corresponding to the predicted time point by using the initial prediction model based on the training operation data and the correlation coefficient, and determining the target smoothing coefficient applied by the initial prediction model during the third smoothing process based on the predicted operation data and the real operation data, comprises: Based on the training operation data and the correlation coefficient, using the initial prediction model, predicting the predicted operation data corresponding to the predicted time point; determining a prediction error based on the predicted operating data and the actual operating data; In the case where the prediction error is greater than the preset error threshold, a preset value is added to the current smoothing coefficient applied by the initial prediction model during the third smoothing process to obtain an adjusted smoothing coefficient, and the adjusted smoothing coefficient is used as the current smoothing coefficient to update the initial prediction model, and the step of predicting the predicted running data corresponding to the predicted time point using the initial prediction model based on the training running data and the correlation coefficient is repeatedly performed; When the prediction error is less than or equal to the preset error threshold, the current smoothing coefficient is used as the target smoothing coefficient applied by the initial prediction model during the third smoothing process.

6. The method according to claim 1, characterized in that The obtaining of a correlation coefficient based on the abnormal operation data and the characteristic data includes: Performing one-hot encoding on the feature data to obtain an encoding result; Based on the encoding result, a correlation analysis is performed on the abnormal operation data and the characteristic data to obtain a correlation coefficient.

7. The method according to any one of claims 1 to 6, characterized in that: The target device is a gas-insulated metal-enclosed switchgear located in a power system.

8. A method for predicting operating data, characterized in that: include: Obtaining a target prediction model trained according to the model training method of any one of claims 1 to 7; Acquire first operation data of the target device at a current time point and second operation data at a preset number of historical time points before the current time point; Based on the first operating data, the second operating data and the target prediction model, future operating data of the target device in a future time period after the current time point is predicted.

9. The method according to claim 8, characterized in that After the prediction obtains the future operation data of the future time period after the current time point, the method further includes: Based on the future operation data, fault diagnosis is performed on the target device.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, which are used to enable a processor to implement the model training method as described in any one of claims 1-7, or the operation data prediction method as described in claim 8 or 9 when executed.

Images