Evaluation method based on digital monitoring model of power system, readable storage medium

Abstract

The embodiment of the application provides an evaluation method and readable storage medium based on a power system digital supervision model, and belongs to the technical field of power digital supervision. The evaluation method comprises the following steps: acquiring historical output data of a digital supervision model in a power system and historical actual data of a supervised module; constructing a supervision data set by acquiring the historical output data of the digital supervision model and the historical actual data of the supervised module, training a digital supervision evaluation model by using the supervision data set, inputting output data of the current digital supervision model and actual data of the supervised module into the trained digital supervision evaluation model, and obtaining an evaluation value, using the evaluation value to evaluate the current digital supervision model, and further analyzing abnormal conditions of the current digital supervision model in a timely manner to ensure the reliability of the digital supervision model of the power system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power digital supervision, in particular to an evaluation method based on a power system digital supervision model and a readable storage medium. BACKGROUND

[0002] A power system is a complex system that converts energy into electric energy through power generation devices and the like, and then supplies electric energy to users through power transmission, power transformation and power distribution. In order to maintain the stability and reliability of the power system, reliable supervision of multiple modules of the power system is required. Digital supervision combines the advancement of information technology with the actual needs of supervision work, aiming to improve supervision efficiency, quality and intelligence. The modules that can be digitally supervised in the power system can include line voltage, circuit breaker current, power equipment safety parameters, etc.

[0003] At present, for the digital supervision of each module of the power system, a corresponding digital supervision model is mainly constructed, and digital prediction is performed on each module of the power system to realize digital supervision of each module of the power system and ensure the safe and stable operation of each module. However, with the continuous construction and iteration of the power system, some digital supervision models have less training data, resulting in low prediction performance accuracy. Secondly, in the training of traditional digital supervision models, known influencing factors are used for training, and unknown factors still affect the digital supervision model. Therefore, it is necessary to evaluate the digital supervision model in real time to ensure that the digital supervision model can be continuously optimized. However, there is no way to evaluate the digital supervision at present, and thus the reliability of the existing digital supervision model of the power system cannot be guaranteed.

[0004] The present application inventors found in the process of implementing the present application that the above-mentioned prior art scheme has the defect of poor reliability of the existing digital supervision model of the power system. SUMMARY

[0005] The purpose of the embodiment of the present application is to provide an evaluation method based on a power system digital supervision model and a readable storage medium, which has the function of improving the reliability of the digital supervision model of the power system.

[0006] In order to achieve the above-mentioned purpose, the embodiment of the present application provides an evaluation method based on a power system digital supervision model, comprising:

[0007] obtaining historical output data of a digital supervision model in a power system and historical actual data of a supervised module;

[0008] Preprocess the historical output data of the digital supervision model and the historical actual data of the supervised module to obtain a supervision data set;

[0009] Construct a digital supervision evaluation model;

[0010] Train the digital supervision evaluation model using the supervision data set;

[0011] Obtain output data of the digital supervision model in a preset time period and actual data of the supervised module in the preset time period;

[0012] Obtain an evaluation value of the digital supervision model according to the output data of the digital supervision model in the preset time period and the actual data of the supervised module in the preset time period;

[0013] Evaluate the current digital supervision model according to the evaluation value of the digital supervision model.

[0014] Optionally, obtaining the historical output data of all digital supervision models in the power system and the historical actual data of the supervised module includes:

[0015] Obtain the historical output data of the digital supervision model ;

[0016] Obtain the historical actual data of the supervised module ;

[0017] Divide the historical output data and the historical actual data into equal intervals according to the preset time period to obtain a historical output data set and a historical actual data set .

[0018] Optionally, preprocessing the historical output data of the digital supervision model and the historical actual data of the supervised module includes:

[0019] Obtain an error sequence of the historical output data set and the historical actual data set according to formula (1),

[0020] , (1)

[0021] wherein, is an error value of the historical output data set and the historical actual data set at the th data sequence moment, is a value of the historical output data set at the th data sequence moment, a value of a time point of a data sequence in the historical actual data set, , a total number of individuals in the historical output data set or the historical actual data set, a time point in a data sequence;

[0022] a plurality of the error sequences within 0 to T are summarized to form an error sequence set

[0023] the error sequence set is normalized.

[0024] Optionally, the normalization of the error sequence set comprises:

[0025] each data sequence in the error sequence set is normalized according to formula (2),

[0026] (2)

[0027] wherein, a normalized value of a time point of a data sequence in the error sequence set, a value of a time point of a data sequence in the error sequence set, a maximum value of a data sequence in the error sequence set, a minimum value of a data sequence in the error sequence set; a prediction evaluation index of each data sequence in the error sequence set is obtained, and a supervision data set is constructed with the normalized data sequence.

[0028] Optionally, the training of the digital supervision evaluation model by using the supervision data set comprises:

[0029] a fast Fourier transform is performed on each individual in the supervision data set to obtain a corresponding frequency domain feature;

[0030] the frequency domain feature is input into the digital supervision evaluation model;

[0031] an estimated value of the digital supervision evaluation model is obtained.

[0032] an estimated value of the digital supervision evaluation model is obtained.

[0033] ​​​​​​​​​The parameters of the digital supervision evaluation model are corrected based on the estimated values ​​of the digital supervision evaluation model and the corresponding predicted evaluation indicators.

[0034] Optionally, obtaining the evaluation value of the digital supervision model based on the output data of the digital supervision model within a preset time period and the actual data of the currently supervised module within the preset time period includes:

[0035] The error sequence for the current preset time period is obtained according to formula (3).

[0036] (3)

[0037] in, For the current preset time period The error sequence, For the current preset time period The output data, For the current preset time period Actual data, Numbered by integer;

[0038] According to formula (4), the normalized value of the error sequence for the current preset time period is obtained.

[0039] (4)

[0040] in, For the current preset time period The error sequence in The value after time normalization, For the current preset time period The error sequence in The value at time, For the current preset time period The maximum value of the error sequence, For the current preset time period The minimum value of the error sequence;

[0041] The normalized value of the error sequence for the current preset time period is input into the digital supervision evaluation model to obtain the predictive evaluation index of the digital supervision model for the current preset time period.

[0042] Optionally, obtaining the evaluation value of the digital supervision model based on the output data of the digital supervision model within a preset time period and the actual data of the currently supervised module within the preset time period further includes:

[0043] Before obtaining the current digital supervision model The parameter adjustment coefficients for a preset time period;

[0044] According to formula (5), a first evaluation value of the current digital supervision model is obtained,

[0045] , (5)

[0046] wherein, the first evaluation value of the current digital supervision model, the output value of the digital supervision evaluation model in the preset time period, the output value of the digital supervision evaluation model in the preset time period, the parameter adjustment coefficient of the digital supervision evaluation model in the preset time period, the integer number, the current preset time period, and ;

[0047] According to formula (6), a second evaluation value of the current digital supervision model is obtained,

[0048] , (6)

[0049] wherein, the second evaluation value of the current digital supervision model, the output value of the digital supervision evaluation model in the preset time period, the integer number, the integer number, , and , , the evaluation coefficient, and .

[0050] Optionally, the parameter adjustment coefficient of the digital supervision evaluation model in the preset time period before the current digital supervision model is obtained includes: According to formula (7), the parameter adjustment coefficient of the digital supervision evaluation model in the preset time period before the current digital supervision model is obtained,

[0051] ,

[0052] ,

[0053] ,

[0054] ,

[0055] , (7)

[0056] wherein, the parameter adjustment value, .

[0057] ​​Optionally, evaluating the current digital supervision model according to the evaluation value of the digital supervision model comprises:

[0058] determining whether the first evaluation value of the current digital supervision model is greater than or equal to a first threshold value;

[0059] in a case where the first evaluation value of the current digital supervision model is greater than or equal to the first threshold value, determining that the current digital supervision model has an abnormal influencing factor;

[0060] in a case where the first evaluation value of the current digital supervision model is less than the first threshold value, determining whether the second evaluation value of the current digital supervision model is greater than or equal to a second threshold value;

[0061] in a case where the second evaluation value of the current digital supervision model is greater than or equal to the second threshold value, determining that the current digital supervision model has a periodic influencing factor;

[0062] in a case where the second evaluation value of the current digital supervision model is less than the second threshold value, determining that the current digital supervision model is relatively optimal.

[0063] On the other hand, the present application also provides a computer readable storage medium, the computer readable storage medium stores instructions, the instructions are used for being read by a machine to make the machine execute the evaluation method as any of the above.

[0064] Through the above technical solution, the evaluation method and the readable storage medium based on the digital supervision model of the power system provided by the present application obtain the historical output data of the digital supervision model and the historical actual data of the supervised module, to construct a supervision data set, and train the digital supervision evaluation model by using the supervision data set, input the output data of the current digital supervision model and the actual data of the supervised module into the trained digital supervision evaluation model, to obtain the evaluation value, and evaluate the current digital supervision model by using the evaluation value, so that the abnormal situation of the current digital supervision model can be determined in time, and further analysis can be performed, to ensure the reliability of the digital supervision model of the power system.

[0065] Other features and advantages of the present application will be described in detail in the following specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0066] The accompanying drawings are included to provide a further understanding of the present application and constitute a part of the specification, and are used together with the following specific embodiments to explain the present application, but do not constitute a limitation on the present application. In the drawings:

[0067] Figure 1is a flowchart of an evaluation method based on a digital supervision model of a power system according to an embodiment of the present application;

[0068] Figure 2 is a flowchart of obtaining historical output data and historical actual data in an evaluation method based on a digital supervision model of a power system according to an embodiment of the present application;

[0069] Figure 3 is a flowchart of obtaining an error sequence in an evaluation method based on a digital supervision model of a power system according to an embodiment of the present application;

[0070] Figure 4 is a flowchart of constructing a supervision data set in an evaluation method based on a digital supervision model of a power system according to an embodiment of the present application;

[0071] Figure 5 is a flowchart of training a digital supervision evaluation model in an evaluation method based on a digital supervision model of a power system according to an embodiment of the present application;

[0072] Figure 6 is a flowchart of obtaining an evaluation value in an evaluation method based on a digital supervision model of a power system according to an embodiment of the present application;

[0073] Figure 7 is a flowchart of evaluating a digital supervision model in an evaluation method based on a digital supervision model of a power system according to an embodiment of the present application. DETAILED DESCRIPTION

[0074] The specific embodiments of the embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely intended to illustrate and explain the embodiments of the present application, and are not intended to limit the embodiments of the present application.

[0075] It should be noted that the acquisition, transmission, storage, use, processing, etc. of data in the technical solutions of the present application comply with relevant provisions of national laws and regulations. In the embodiments of the present application, some industry existing solutions, components, models, etc. may be mentioned, which should be considered as exemplary, and the purpose is only to illustrate the feasibility of the implementation of the technical solutions of the present application, but it does not mean that the applicant has or will necessarily use the solutions.

[0076] Figure 1 is a flowchart of an evaluation method based on a digital supervision model of a power system according to an embodiment of the present application. In Figure 1 , the evaluation method can include:

[0077] In step S10, the historical output data of the digital supervision model in the power system and the historical actual data of the supervised module are acquired. The historical output data of the digital supervision model in the power system can include historical prediction data of voltage, current, parameters and the like in the power system, i.e., time-continuous time series data. The supervised module is a module supervised by the digital supervision model, and the historical actual data output by the module is actual data output by the module during actual operation, i.e., time-continuous time series data.

[0078] In step S11, the historical output data of the digital supervision model and the historical actual data of the supervised module are preprocessed to obtain a supervision data set. The historical output data and the historical actual data can be preprocessed to obtain error sequences of the two, and the error sequences are normalized to obtain the supervision data set.

[0079] In step S12, a digital supervision evaluation model is constructed. The digital supervision evaluation model can include a CNN convolutional neural network model known to those skilled in the art.

[0080] In step S13, the digital supervision evaluation model is trained using the supervision data set. After the digital supervision evaluation model is constructed, the digital supervision evaluation model can be trained using the supervision data set.

[0081] In step S14, output data of the digital supervision model in a preset time period and actual data of the current supervised module in the preset time period are acquired. When the digital supervision model and the supervised module are normally operated, the output data and the actual data in the same preset time period are acquired.

[0082] In step S15, an evaluation value of the digital supervision model is acquired according to the output data of the digital supervision model in the preset time period and the actual data of the current supervised module in the preset time period. The output data and the actual data in the same time period are preprocessed and input into the trained digital supervision model, and then a prediction evaluation index can be output, i.e., an evaluation index of the prediction ability of the digital supervision model. Specifically, in the embodiment of the present application, the evaluation index can include 0 and 1, representing good prediction ability and poor prediction ability, respectively, and finally the evaluation index is converted into a specific evaluation value.

[0083] In step S16, the current digital supervision model is evaluated according to the evaluation value of the digital supervision model. After the evaluation value of the digital supervision model is acquired, the current digital supervision model can be evaluated to determine the reliability of the current digital supervision model.

[0084] In steps S10 to S16, the historical output data of the digital supervision module in the power system and the historical actual data of the supervised module are first acquired, and then data conversion is performed on the two to obtain an error sequence and construct a supervision data set, and the digital supervision evaluation model is trained using the supervision data set. The output data of the current digital supervision model in a preset time period and the actual data of the supervised module in the same preset time period are preprocessed in the same way and input into the trained digital supervision evaluation model, and then the evaluation value of the digital supervision model can be obtained. The evaluation value is used to evaluate the current digital supervision model to determine whether there are other prediction influencing factors in the current digital supervision model.

[0085] In the training of traditional digital supervision models, known influencing factors are used for training, and unknown factors still affect the digital supervision model, so real-time evaluation of the digital supervision model is needed to ensure that the digital supervision model can be continuously optimized. However, there is currently no way to evaluate the digital supervision, and thus the reliability of the existing digital supervision model of the power system cannot be guaranteed. In this embodiment of the application, the digital supervision evaluation model is trained using the supervision data set, which can accurately and effectively evaluate the current digital supervision model, and thus the abnormal situation of the current digital supervision model can be determined in a timely manner and further analyzed to ensure the reliability of the digital supervision model of the power system.

[0086] In this embodiment of the application, in order to obtain the supervision data set for training the model, the historical output data of the digital supervision model and the historical actual data of the supervised module are also needed, and multi-period division is performed, and the specific acquisition steps can be as shown in Figure 2 Specifically, in Figure 2 , the evaluation method can further include:

[0087] In step S100, the historical output data of the digital supervision model is acquired . The historical output data can include model output data in a long time range, i.e., time-continuous data.

[0088] In step S101, the historical actual data of the supervised module is acquired . The historical actual data of the supervised module can include actual output data in a long time range, i.e., time-continuous data.

[0089] In step S102, the historical output data and the historical actual data are equally spaced segmented according to a preset time period to obtain a historical output data set and a historical actual data set Wherein, after the historical output data and the historical actual data are equally divided, that is, equally divided in time period, each individual obtained is time-continuous time series data, and the individuals in the two data sets are one-to-one corresponding.

[0090] In this embodiment of the application, after the historical output data and the historical actual data are obtained, the prediction advantages and disadvantages of the digital supervision model need to be obtained according to the two, and therefore the error sequence also needs to be obtained according to the two and preprocessed. The specific processing steps can be as shown in Figure 3 . Specifically, in Figure 3 , the evaluation method can further include:

[0091] In step S110, the error sequence of the historical output data set and the historical actual data set is obtained according to formula (1),

[0092] , (1)

[0093] Wherein, is the error value of the historical output data set and the historical actual data set at the th data sequence moment, is the value of the historical output data set at the th data sequence moment, is the value of the historical actual data set at the th data sequence moment, is an integer number, and , is the total number of individuals in the historical output data set or the historical actual data set, is the moment in the data sequence. The error values of the continuous moments in the data sequence are summarized to obtain the error sequence, and the data sequence is the individual in the data set.

[0094] In step S111, a plurality of error sequences from 0 to T are summarized to form an error sequence set .

[0095] In step S112, the error sequence set is normalized. The normalization of the error sequence set can include the steps as shown in Figure 4 . Specifically, in Figure 4 , the evaluation method can further include:

[0096] In step S1120, each data sequence in the error sequence set is normalized according to formula (2),

[0097] , (2)

[0098] wherein, is the value of the i-th data sequence in the error sequence set at the time t, is the normalized value of the i-th data sequence in the error sequence set at the time t, is the value of the i-th data sequence in the error sequence set at the time t, is the maximum value of the i-th data sequence in the error sequence set, is the minimum value of the i-th data sequence in the error sequence set.

[0099] In step S1121, the prediction evaluation index of each data sequence in the error sequence set is obtained, and a supervision data set is constructed with the normalized data sequence. The prediction evaluation index can include the advantages and disadvantages of the model, corresponding to 0 and 1. Specifically, the prediction evaluation index can include but is not limited to being determined according to the mean value of the error sequence set.

[0100] In steps S110 to S112, the error sequence of the historical output data set and the historical actual data set in the same time period is calculated first, and then the error sequence of each time period is summarized respectively to obtain the error sequence set. Finally, each error sequence in the error sequence set is normalized, and the supervision data set is constructed in combination with the prediction evaluation index of each error sequence.

[0101] In this embodiment of the present application, after the preprocessed data obtains the supervision data set for training, the digital supervision evaluation model can be trained using the supervision data set. Specifically, the training steps can be as shown in Figure 5 Specifically, in Figure 5 the evaluation method can further include:

[0102] In step S130, fast Fourier transform is performed on each individual in the supervision data set to obtain the corresponding frequency domain feature. The individuals in the supervision data set are all time domain data, and the frequency domain data can be obtained by performing fast Fourier transform on the time domain data, which can better reflect the periodic characteristics and thus improve the accuracy of model training.

[0103] In step S131, the frequency domain feature is input into the digital supervision evaluation model. After the frequency domain feature enters the digital supervision evaluation model, convolution, pooling, and full connection layer are required to output the corresponding estimated value.

[0104] In step S132, the estimated value of the digital supervision evaluation model is obtained.

[0105] ​​​​​In step S133, the parameters of the digitalized supervision evaluation model are corrected according to the estimated value of the digitalized supervision evaluation model and the corresponding predicted evaluation index.

[0106] In steps S130 to S133, the fast Fourier transform is first performed on each individual in the supervision data set to convert the time domain characteristics of the data into frequency domain characteristics. The frequency domain characteristics are input into the digitalized supervision evaluation model for convolution and pooling operations, and finally the corresponding estimated value is output. According to the estimated value, the parameters of the digitalized supervision evaluation model are corrected, so as to complete the training of the digitalized supervision evaluation model.

[0107] In this embodiment of the present application, after the training of the digitalized supervision evaluation model is completed, the evaluation value of the digitalized supervision model can be obtained according to the current digitalized supervision evaluation model. Specifically, the evaluation value obtaining step can be as shown in Figure 6 . Specifically, in Figure 6 , the evaluation method can further include:

[0108] In step S150, the error sequence of the current preset time period is obtained according to formula (3),

[0109] , (3)

[0110] wherein, is the error value (error sequence) of the current preset time period (at moment), is the output data of the current preset time period (at moment), is the actual data of the current preset time period ( moment), is an integer number. Specifically, the multiple moments within the preset time period are summarized, and the final error sequence is obtained.

[0111] In step S151, the normalized value of the error sequence of the current preset time period is obtained according to formula (4),

[0112] , (4)

[0113] wherein, is the normalized value of the error sequence of the current preset time period at moment, is the value of the error sequence of the current preset time period at moment, For the current preset time period The maximum value of the error sequence, For the current preset time period The minimum value of the error sequence.

[0114] In step S152, the normalized value of the error sequence for the current preset time period is input into the digital supervision evaluation model to obtain the predicted evaluation index of the digital supervision model for the current preset time period. Specifically, after obtaining the error sequence for the current preset time period and performing normalization, it can be input into the trained digital supervision evaluation model. The digital supervision evaluation model outputs its predicted evaluation index, which is either 0 or 1. If the output is 0, it indicates that the prediction accuracy of the current digital supervision model is high; conversely, it indicates that the prediction accuracy of the current digital supervision model is low.

[0115] In step S153, before obtaining the current digital supervision model The parameter adjustment coefficients are set for a preset time period. Among them, for the current digital supervision model... The preset time period is the consecutive period before the current preset time period. There are several preset time periods. Specifically, for multiple consecutive preset time periods, there are certain influencing factors between them. Therefore, in order to highlight the interaction and correlation between multiple consecutive preset time periods, a corresponding parameter adjustment coefficient can be set for each preset time period. Specifically, the parameter adjustment coefficient for each preset time period can be set according to formula (7).

[0116] ,

[0117] ,

[0118] ,

[0119] (7)

[0120] in, To adjust the parameter values, Specifically, each parameter adjustment coefficient in formula (7) is determined based on the previous output value. If the output values ​​of multiple consecutive preset time periods are all 1, that is, if the digital supervision model does not meet the requirements, then the value of the subsequent parameter adjustment coefficient is increased, that is, additional cumulative factors are considered, and thus it can effectively determine whether there are additional abnormal factors.

[0121] In step S154, the first evaluation value of the current digital supervision model is obtained according to formula (5).

[0122] (5)

[0123] in, This is the first evaluation value of the current digital supervision model. For the front The output value of the digital supervision and evaluation model for a preset time period. For the front The parameter tuning coefficients of the digital supervision and evaluation model for a preset time period. Numbered by integer. For the current preset time period, and Specifically, for The value represents a continuous sequence. The output / predicted evaluation values ​​for each preset time period are generally at least two. However, considering the prediction accuracy of the digital supervision model, the number of preset time periods can be appropriately increased to reduce the number of calculations for the first evaluation value, thus reducing the calculation frequency. Specifically, this first evaluation value can reflect the sudden changes, or anomalies, of the current digital supervision model within a continuous time period. Based on these anomalies, it can be determined whether there are influencing factors affecting the prediction effect within that continuous time period, thereby effectively capturing accidental factors and preventing overfitting of the digital supervision model during continuous training, or analyzing other influencing factors of the digital supervision model.

[0124] In step S155, the second evaluation value of the current digital supervision model is obtained according to formula (6).

[0125] (6)

[0126] in, This is the second evaluation value of the current digital supervision model. For the front The output value of the digital supervision and evaluation model for a preset time period. Numbered by integer. Numbered by integer. ,and , that is Divisible , The evaluation coefficient is , and . Specifically, considering that the digital supervision model may have periodic influencing factors, the periodic data of the digital supervision model can be feature-captured, that is, the periodic variation information of the prediction effect of the digital supervision model can be obtained according to formula (6). Specifically, the adjacent three output values used for the calculation of the second evaluation value are periodically judged. In fact, the periodic variation of part of the digital supervision model has certain fluctuations, so for this case, the calculation of the second evaluation value can also be as shown in formula (8),

[0127] , (8)

[0128] wherein, is an integer number. According to formula (8), the output values of the adjacent multiple preset time periods are used as the overall periodic information, and the overall periodic information formed by the output values of the other multiple preset time periods is analyzed comprehensively to determine whether the current digital supervision model has periodic influencing factors.

[0129] In addition, for obtaining the periodic influencing factors of the digital supervision model, the value of can be dynamically adjusted, but the value of has an upper limit .Specifically, , is an integer number. The value of changes in order from 1 to , so as to fully obtain the periodic information of the current digital supervision model from multiple scales, further improving the evaluation precision of the prediction ability of the digital supervision evaluation model on the digital supervision model, and determining whether the periodic influencing factors of the digital supervision model are not considered, so as to facilitate the further optimization of the digital supervision model.

[0130] In steps S150 to S155, first, the error sequence corresponding to the output data and the actual data of the current preset time period is obtained, and then the error sequence is normalized. The normalized error sequence is input into the trained digital supervision evaluation model, and then the output value of the current preset time period, that is, the prediction evaluation index, can be obtained. Combined with the parameter adjustment coefficients and the output values of the previous preset time periods, the first evaluation value of the current digital supervision model can be obtained. Specifically, the current first evaluation value can reflect the abnormal information of the current digital supervision model, that is, the mutation information. At the same time, according to the output values of multiple preset time periods connected in multiple groups, the periodic information of the current digital supervision model, that is, the second evaluation value, can be captured. According to the abnormal information and the periodic information of the current digital supervision model, the advantages and disadvantages of the current digital supervision model can be evaluated.

[0131] In this embodiment of the present application, after the first evaluation value and the second evaluation value of the current digital supervision model are obtained, the current digital supervision model can be evaluated. Specifically, the evaluation steps can be as shown in Figure 7 Specifically, in Figure 7 , the evaluation method can further include:

[0132] In step S160, it is determined whether the first evaluation value of the current digital supervision model is greater than or equal to the first threshold value. The first threshold value can be determined according to the value of , that is, according to the number of adjacent preset time periods selected in the first evaluation value. When is greater, the value of the first threshold value is also greater, so as to ensure that the first threshold value can accurately evaluate the current first threshold value.

[0133] In step S161, in the case where the first evaluation value of the current digital supervision model is greater than or equal to the first threshold value, it is determined that the current digital supervision model has an abnormal influencing factor. If the current first evaluation value is greater than or equal to the first threshold value, it means that the current digital supervision model has an abnormal factor with a greater impact. It is further analyzed whether the abnormal factor in the plurality of time periods is an occasional event. If it is an occasional event, the data in the plurality of time periods should not be used to train the digital supervision model to prevent model overfitting. If it is not an occasional event, it is further analyzed whether to add it as a new training factor of the digital supervision model to improve the reliability of the digital supervision model. Specifically, after it is determined that the current digital supervision model has an influencing factor, it is further determined whether the second evaluation value of the current digital supervision model is greater than or equal to the second threshold value.

[0134] In step S162, in the case where the first evaluation value of the current digital supervision model is less than the first threshold value, it is determined whether the second evaluation value of the current digital supervision model is greater than or equal to the second threshold value. If the current first evaluation value is less than the first threshold value, it means that the prediction ability of the current digital supervision model is good and meets the prediction accuracy requirement. Therefore, it is further determined whether the digital supervision model has a periodic influencing factor, that is, the second evaluation value is compared with the second threshold value. Specifically, the size of the second threshold value can be adjusted according to the number of selected periods, that is, determined according to the value of . If the number of selected periods is greater, the second threshold value can be relatively large, and vice versa.

[0135] In step S163, in the case that the second evaluation value of the current digital supervision model is greater than or equal to the second threshold value, it is determined that the current digital supervision model has a periodic influencing factor. If the second evaluation value is greater than or equal to the second threshold value, it indicates that the current digital supervision model has a periodic influencing factor, and further analysis is needed to determine whether the periodic influencing factor can be added as a new training factor to further improve the reliability of the digital supervision model.

[0136] In step S164, in the case that the second evaluation value of the current digital supervision model is less than the second threshold value, it is determined that the current digital supervision model is better. If the second evaluation value is less than the second threshold value, it indicates that the current digital supervision model does not have a periodic influencing factor, and the current digital supervision model is determined to be better and has better prediction ability.

[0137] In steps S160 to S164, the first evaluation value of the current digital supervision model is first compared with the first threshold value. If the first evaluation value is greater than or equal to the first threshold value, it indicates that the current digital supervision model has an abnormal influencing factor. If the first evaluation value is less than the first threshold value, it indicates that there is no abnormal factor, and further determination is needed to determine whether there is a periodic influencing factor. Then, the second evaluation value is compared with the second threshold value. If the second evaluation value is greater than or equal to the second threshold value, it indicates that the current digital supervision model has a periodic influencing factor, otherwise, it does not have a periodic influencing factor, and the digital supervision model is determined to be better.

[0138] In this embodiment of the present application, the evaluation order of the first evaluation value and the second evaluation value can be determined by parallel judgment or sequential judgment, that is, two judgments are needed.

[0139] On the other hand, the present application also provides a computer readable storage medium, which stores instructions for being read by a machine to make the machine execute the evaluation method of any one of the above.

[0140] Through the above technical solution, the evaluation method and the readable storage medium based on the digital supervision model of the power system provided by the present application obtain the historical output data of the digital supervision model and the historical actual data of the supervised module to construct a supervision data set, and train the digital supervision evaluation model using the supervision data set. The output data of the current digital supervision model and the actual data of the supervised module are input into the trained digital supervision evaluation model to obtain the evaluation value. The evaluation value is used to evaluate the current digital supervision model, so that the abnormal situation of the current digital supervision model can be determined in time, and further analysis can be performed to ensure the reliability of the digital supervision model of the power system.

[0141] Those skilled in the art will appreciate that embodiments of the application can be readily used as software, hardware, or a combination of software and hardware. In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0142] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 means for performing each of the functions specified in the flowchart block or blocks.

[0143] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 means for performing each of the functions specified in the flowchart block or blocks.

[0144] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 means for performing each of the functions specified in the flowchart block or blocks.

[0145] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0146] The memory can include non-persistent memory and / or volatile memory, such as random access memory (RAM) about which the processor can execute instructions. The memory can also include non-volatile memory, such as read only memory (ROM) or flash RAM, about which permanent information can be stored; such information can not change much and / or can only change slowly.

[0147] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible by a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0148] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover non-exclusive inclusions, so that a process, method, article or apparatus that includes a list of elements does not only include those elements, but also includes other elements not explicitly listed, or further includes elements inherent in such a process, method, article or apparatus. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.

[0149] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the scope of claims of the present application.

Claims

1. An evaluation method based on a digital supervision model of a power system, characterized in that, include: Acquire historical output data of the digital supervision model in the power system and historical actual data of the supervised modules; The historical output data of the digital supervision model and the historical actual data of the supervised module are preprocessed to obtain a supervision data set. Construct a digital supervision and evaluation model; The digital supervision evaluation model is trained using the aforementioned supervision data set; Obtain the output data of the digital supervision model within a preset time period and the actual data of the currently supervised module within the preset time period; The evaluation value of the digital supervision model is obtained based on the output data of the digital supervision model during a preset time period and the actual data of the supervised module during the preset time period. The current digital supervision model is evaluated based on its evaluation value. The evaluation value of the digital supervision model is obtained based on the output data of the digital supervision model during a preset time period and the actual data of the supervised module during the preset time period, including: Before obtaining the current digital supervision model The parameter adjustment coefficients for a preset time period; The first evaluation value of the current digital supervision model is obtained according to formula (5). ，（5） in, This is the first evaluation value of the digital supervision model described above. For the front The output value of the digital supervision and evaluation model for a preset time period. For the front The parameter tuning coefficients of the digital supervision and evaluation model for a preset time period. Numbered by integer. For the current preset time period, and ; The second evaluation value of the current digital supervision model is obtained according to formula (6). ，（6） in, This is the second evaluation value of the digital supervision model described above. For the front The output value of the digital supervision and evaluation model for a preset time period. Numbered by integer. Numbered by integer. ,and , This is the evaluation coefficient; Evaluating the current digital supervision model based on its evaluation value includes: Determine whether the first evaluation value of the current digital supervision model is greater than or equal to the first threshold. If the first evaluation value of the current digital supervision model is greater than or equal to the first threshold, it is determined that the current digital supervision model has abnormal influencing factors. If the first evaluation value of the current digital supervision model is less than the first threshold, then determine whether the second evaluation value of the current digital supervision model is greater than or equal to the second threshold. If the second evaluation value of the current digital supervision model is greater than or equal to the second threshold, it is determined that the current digital supervision model has periodic influencing factors. If the second evaluation value of the current digital supervision model is less than the second threshold, the current digital supervision model is determined to be superior.

2. The evaluation method according to claim 1, characterized in that, Acquiring historical output data of all digital supervision models in the power system and historical actual data of the supervised modules includes: Obtain the historical output data of the digital supervision model. ; Obtain the historical actual data of the supervised module. ; The historical output data and the historical actual data are divided into equal intervals according to the preset time period to obtain the historical output dataset. and historical real-world datasets .

3. The evaluation method according to claim 2, characterized in that, Preprocessing the historical output data of the digital supervision model and the historical actual data of the supervision module includes: The error sequences of the historical output dataset and the historical actual dataset are obtained according to formula (1). ，（1） in, For the historical output dataset and the historical actual dataset, the first... Data sequences Error value at time, For the historical output dataset, the first Data sequences The value at time, For the historical actual dataset, the first Data sequences The value at time, The number is an integer, and , The total number of individuals in the historical output dataset or the historical actual dataset. For each moment in the data sequence; Multiple error sequences within the range of 0 to T are summarized to form an error sequence set. ; The error sequence set is then normalized.

4. The evaluation method according to claim 3, characterized in that, Normalizing the error sequence set includes: According to formula (2), each data sequence in the error sequence set is normalized. ，（2） in, For the error sequence set, the first... Data sequences The value after time normalization, For the error sequence set, the first... Data sequences The value at time, For the error sequence set, the first... The maximum value of a data sequence. For the error sequence set, the first... The minimum value of a data sequence; Obtain the prediction and evaluation index for each data sequence in the error sequence set, and construct a supervision data set with the normalized data sequence.

5. The evaluation method according to claim 4, characterized in that, Training the digital supervision evaluation model using the aforementioned supervision data set includes: Perform a Fast Fourier Transform on each individual data point in the monitoring dataset to obtain the corresponding frequency domain features; The frequency domain features are input into the digital supervision and evaluation model; Obtain the estimated value of the digital supervision and evaluation model; The parameters of the digital supervision evaluation model are corrected based on the estimated values ​​of the digital supervision evaluation model and the corresponding predicted evaluation indicators.

6. The evaluation method according to claim 5, characterized in that, Obtaining the evaluation value of the digital supervision model based on the output data of the digital supervision model over a preset time period and the actual data of the supervised module over the preset time period also includes: The error sequence for the current preset time period is obtained according to formula (3). ，（3） in, For the current preset time period The error sequence, For the current preset time period The output data, For the current preset time period Actual data, Numbered by integer; According to formula (4), the normalized value of the error sequence for the current preset time period is obtained. ，（4） in, For the current preset time period The error sequence in The value after time normalization, For the current preset time period The error sequence in The value at time, For the current preset time period The maximum value of the error sequence, For the current preset time period The minimum value of the error sequence; The normalized value of the error sequence for the current preset time period is input into the digital supervision evaluation model to obtain the predictive evaluation index of the digital supervision model for the current preset time period.

7. The evaluation method according to claim 1, characterized in that, Before obtaining the current digital supervision model The parameter adjustment coefficients for each preset time period include: According to formula (7), the current digital supervision model is obtained before The parameter adjustment coefficients for a preset time period. ， ， ， ，（7） in, To adjust the parameter values, .

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that are read by a machine to cause the machine to perform the evaluation method as described in any one of claims 1-7.

Images