CN112949951A - Data prediction method, data prediction device, electronic equipment and storage medium - Google Patents
Data prediction method, data prediction device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN112949951A CN112949951A CN202110476640.XA CN202110476640A CN112949951A CN 112949951 A CN112949951 A CN 112949951A CN 202110476640 A CN202110476640 A CN 202110476640A CN 112949951 A CN112949951 A CN 112949951A
- Authority
- CN
- China
- Prior art keywords
- index
- data
- abnormal
- value
- sequence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 61
- 230000002159 abnormal effect Effects 0.000 claims abstract description 261
- 238000001514 detection method Methods 0.000 claims abstract description 45
- 230000006870 function Effects 0.000 claims description 25
- 238000012549 training Methods 0.000 claims description 21
- 238000007781 pre-processing Methods 0.000 claims description 11
- 230000009466 transformation Effects 0.000 claims description 7
- 230000005856 abnormality Effects 0.000 description 9
- 230000005611 electricity Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000010606 normalization Methods 0.000 description 3
- 238000012790 confirmation Methods 0.000 description 2
- 230000010339 dilation Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000001594 aberrant effect Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000008713 feedback mechanism Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011551 log transformation method Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Tourism & Hospitality (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Theoretical Computer Science (AREA)
- Development Economics (AREA)
- General Physics & Mathematics (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention provides a data prediction method, a data prediction device, electronic equipment and a storage medium. According to the embodiment of the invention, the index data sequence corresponding to each first index is respectively detected by using the abnormal detection model to obtain the abnormal data sequence, the abnormal index characteristic sequence is determined according to the abnormal data sequence, the abnormal data sequence and the abnormal index characteristic sequence are input into the trained time convolution network model to obtain the next predicted data value of the abnormal index output by the time convolution network model, so that the predicted results of a plurality of indexes can be automatically obtained in real time, and better support is provided for auxiliary decision making.
Description
Technical Field
The present invention relates to the field of data processing technologies, and in particular, to a data prediction method and apparatus, an electronic device, and a storage medium.
Background
In the process of city development, the management and operation monitoring of cities are also being transformed digitally. Managers of cities need to assist in decision making based on index data covering various areas of city operation.
In the related art, index data corresponding to a plurality of fields can be accessed in the same platform and displayed. The information that this kind of technique can provide is very limited, can not provide strong support for the aid decision-making.
Disclosure of Invention
In order to overcome the problems in the related art, the invention provides a data prediction method, a data prediction device, an electronic device and a storage medium.
According to a first aspect of the embodiments of the present invention, there is provided a data prediction method, including:
respectively detecting the index data sequences corresponding to the first indexes by using an abnormal detection model to obtain abnormal data sequences; each data in the index data sequence is arranged according to the time sequence of data generation;
determining an abnormal index characteristic sequence according to the abnormal data sequence;
and inputting the abnormal data sequence and the abnormal index characteristic sequence into a trained time convolution network model to obtain a next predicted data value of the abnormal index output by the time convolution network model, wherein the abnormal index is an index corresponding to the abnormal data sequence.
According to a second aspect of the embodiments of the present invention, there is provided a data prediction apparatus including:
the first detection module is used for respectively detecting the index data sequences corresponding to the first indexes by using the abnormal detection model to obtain abnormal data sequences; each data in the index data sequence is arranged according to the time sequence of data generation;
the determining module is used for determining an abnormal index characteristic sequence according to the abnormal data sequence;
and the prediction module is used for inputting the abnormal data sequence and the abnormal index characteristic sequence into a trained time convolution network model to obtain a next predicted data value of the abnormal index output by the time convolution network model, wherein the abnormal index is an index corresponding to the abnormal data sequence.
According to a third aspect of embodiments of the present invention, there is provided an electronic apparatus, including:
a memory for storing executable instructions of the processor;
the processor is configured to execute the instructions to implement the method of any of the first aspect.
According to a fourth aspect of embodiments of the present invention, there is provided a computer-readable storage medium having stored thereon computer instructions which, when executed, implement the method of any one of the first aspect.
The technical scheme provided by the embodiment of the invention has the following beneficial effects:
according to the embodiment of the invention, the index data sequence corresponding to each first index is respectively detected by using the abnormal detection model to obtain the abnormal data sequence, the abnormal index characteristic sequence is determined according to the abnormal data sequence, the abnormal data sequence and the abnormal index characteristic sequence are input into the trained time convolution network model to obtain the next predicted data value of the abnormal index output by the time convolution network model, the predicted results of a plurality of indexes can be automatically obtained in real time, and better support is provided for auxiliary decision making.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the specification.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present specification and together with the description, serve to explain the principles of the specification.
Fig. 1 is a flowchart illustrating a data prediction method according to an embodiment of the present invention.
Fig. 2 is a functional block diagram of a data prediction apparatus according to an embodiment of the present invention.
Fig. 3 is a hardware structure diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatus and methods consistent with certain aspects of embodiments of the invention, as detailed in the following claims.
The terminology used in the embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of embodiments of the invention. As used in the examples of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.
It should be understood that although the terms first, second, third, etc. may be used to describe various information in embodiments of the present invention, the information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, the first information may also be referred to as second information, and similarly, the second information may also be referred to as first information, without departing from the scope of embodiments of the present invention. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.
The data prediction method provided by the embodiment of the invention can be used for carrying out prediction and early warning on multiple indexes of multiple subjects in multiple fields of the whole city so as to conveniently manage and monitor the city.
In the embodiment of the invention, the data prediction method can be operated on the server. The server may first access metric data for a plurality of metrics. The index data in each index are arranged according to the time sequence generated by the data.
The multiple indexes can include indexes in the fields of city-level economy, politics, culture, ecology, civil life and the like. The server can set corresponding index codes for each index, and the index codes corresponding to different indexes are different. For example, index codes corresponding to indexes in the economic field may be set to a, index codes corresponding to indexes in the political field may be set to B … …, and so on.
The generation period of the index data corresponding thereto may be different for different indices. For example, for index C, one index data may be generated every month, for index D, one index data may be generated every day, and for index E, one index data may be generated every week.
The server may read data corresponding to the index according to a generation cycle of the index data. For example, for the index C described above, the reading period may be set to one month, and for the index D described above, the reading period may be set to one day.
In this embodiment, all the indexes are divided into two types, one of which is referred to herein as a first index, and the other of which is referred to herein as a second index. The first index is an index predicted by a model, and the first index is processed by the method of the embodiment shown in fig. 1. The second index is an index predicted by a traffic threshold. In an application, corresponding tag information may be set for data of each index, where the tag information is used to indicate a tag type (the tag type is, for example, a first index and a second index), and whether the index is the first index or the second index is indicated by the tag information. For example, a label information of 1 indicates a first index, and a label information of 0 indicates a second index. When the system reads the index data, the prediction mode of the index can be determined through the label information corresponding to the index.
The data prediction method provided by the present invention will be described in detail below with reference to examples.
Fig. 1 is a flowchart illustrating a data prediction method according to an embodiment of the present invention. As shown in fig. 1, the data prediction method may include:
s101, respectively detecting the index data sequences corresponding to the first indexes by using an abnormal detection model to obtain abnormal data sequences; and each data in the index data sequence is arranged according to the time sequence of data generation.
And S102, determining an abnormal index characteristic sequence according to the abnormal data sequence.
S103, inputting the abnormal data sequence and the abnormal index feature sequence into a trained time convolution network TCN (temporal correlation networks) model to obtain a next predicted data value of an abnormal index output by the time convolution network model, wherein the abnormal index is an index corresponding to the abnormal data sequence.
In one example, before step S101, the method may further include:
reading label information used for indicating the index type from the current index data;
identifying the current index type according to the label information;
and if the current index type is the first index, respectively detecting the index data sequence corresponding to each first index by using an abnormal detection model to obtain an abnormal data sequence. It should be noted that each first index corresponds to one index data sequence. The lengths of the index data sequences corresponding to different first indexes may be different. When the data prediction method of the present embodiment is applied, the index data sequence corresponding to each first index is processed separately. For example, if there are 1000 first indexes, the index data sequence corresponding to each of the 1000 indexes is detected, and if 50 indexes are detected to be abnormal, the index data sequences corresponding to the 50 indexes are all abnormal data sequences. And for each abnormal data sequence, respectively processing according to the steps S102 and S103 to obtain a predicted data value corresponding to each index in the 50 indexes.
In an example, before step S101, the method may further include:
and reading the index data sequence corresponding to each first index.
For example, after accessing the index data of a plurality of indexes, the aforementioned server may store the index data in a specified location, for example, a local storage area of the server or a database outside the server. The reading order may be set for all accessed metrics and embodied in the metric data sequence identification. For example, the index data sequence identification may be Xi, i representing the reading order. Assuming that the index data sequence identifications corresponding to the index a, the index B and the index C are X1, X2 and X3, respectively, the order of reading the index data sequences is: index A, index B, index C.
The anomaly detection model can detect all the first indexes one by one. The detection principle of the anomaly detection model can be that according to the distribution and variance fluctuation of the index data values, the anomaly condition that the numerical values in the index data sequence obviously deviate from the rest data values of the sequence is analyzed by adopting a 3 sigma principle. The 3 σ principle is that it is assumed that only random errors are contained in an index data sequence, the standard deviation needs to be calculated, an interval is determined according to a certain probability, and for errors exceeding the interval, the errors do not belong to the random errors but are coarse errors, and data containing the coarse errors need to be removed. The distribution intervals are as follows:
the probability of the numerical distribution in (. mu. - σ,. mu. + σ) is 0.6827
The probability of the numerical distribution in (μ -2 σ, μ +2 σ) is 0.9545
The probability of the numerical distribution in (mu-3 sigma, mu +3 sigma) is 0.9973
Where μ is the average of all data in the index data series and σ is the standard deviation.
Generally, the values of the data in an index data sequence are almost all concentrated in the (mu-3 sigma, mu +3 sigma) interval, the probability of exceeding the range is only less than 0.3%, and the data exceeding the range can be considered as abnormal values.
Therefore, in one example, the method for detecting the index data sequence by the anomaly detection model can comprise the following steps:
determining the average value and the standard deviation of all data in the index data sequence according to all data values in the index data sequence;
determining a target interval range corresponding to the index data sequence according to the average value and the standard deviation;
and judging whether all data values in the index data sequence are in the target interval range, if so, determining that the index data sequence is a normal data sequence, otherwise, determining that the index data sequence is an abnormal data sequence, and determining that the data values in the index data sequence which are not in the target interval range are abnormal values.
For example, for all accessed metrics, the metrics may be classified according to the metricsType (e.g. economy), index coding (e.g. index a), index time ascending order, reading each index data sequence Xi ═ x in a cyclei1，xi2，xi3，...xin]]Where i denotes the reading order and n denotes the length of the ith index sequence. Detecting whether the current index distribution is obeyed normal distribution or not, and calculating the average value mu i and standard deviation sigma of the current index data sequenceiComparing whether each data value in the index data sequence falls within the target interval range (mu)i-3σi,μi+3σi) If the standard deviation exceeds 3 times, the data value is determined to be an abnormal value, a time sequence containing the abnormal value, index coding information and the like are output, and a new abnormal data sequence set { X is formed1,X2,…,XpWhere p is the number of aberrant sequences.
The target interval range may be (μ - σ, μ + σ), (μ -2 σ, μ +2 σ), or (μ -3 σ, μ +3 σ), and the specific interval range may be determined according to the accuracy requirement of the actual application scenario. For example, (μ - σ, μ + σ) may be adopted as the target interval range if the accuracy requirement is high, and (μ -3 σ, μ +3 σ) may be adopted as the target interval range if the accuracy requirement is low.
In an example, before the index data sequences corresponding to the first indexes are respectively detected by using the abnormal detection model to obtain the abnormal data sequences, the method may further include:
and preprocessing each data value in the index data sequence to enable the preprocessed data value to meet the input data requirement of the anomaly detection model.
In one example, preprocessing the individual data values in the index data sequence may include:
and carrying out logarithmic transformation on each data value in the index data sequence to obtain a logarithmic value corresponding to the data value.
Among them, the logarithmic transformation may be a log logarithmic transformation. The log-log transformation has the following effects:
first, when the magnitudes of the index data values are not consistent, taking the logarithm can eliminate the situation that the magnitudes are greatly different.
Secondly, taking the logarithm can eliminate the variance.
Thirdly, the nonlinear variable relation can be converted into a linear relation by taking the logarithm, and parameter estimation is more convenient to carry out.
Fourthly, most index data distribution is in a biased state and can be corrected to be in a normal distribution.
For example, an index of the type of environmental monthly complaint volume gas-related class generates one index data per month, and the index data sequence of 12 months of the index is as follows: [28, 35, 29, 40, 48, 45, 52, 41, 37, 34, 119, 187] the sequence obtained by log-transforming each data value in the index data sequence is as follows:
[1.45，1.54，1.46，1.60，1.68，1.65，1.72，1.61，1.57，1.53，2.08，2.27]
in step S102, the abnormal data sequence refers to an index data sequence including an abnormal value. The abnormal index feature sequence is a sequence for indicating whether each data in the abnormal data sequence is an abnormal value. The data in the anomaly index feature sequence is called an anomaly feature value.
For example, if the 12 th data in the index data series [1.45, 1.54, 1.46, 1.60, 1.68, 1.65, 1.72, 1.61, 1.57, 1.53, 2.08, 2.27] of the above-mentioned environmental monthly complaint amount gas-related index at 12 months is an abnormal value, the index data series is an abnormal data series, that is, [1.45, 1.54, 1.46, 1.60, 1.68, 1.65, 1.72, 1.61, 1.57, 1.53, 2.08, 2.27] is an abnormal data series. Assuming that 0 represents non-abnormality and 1 represents abnormality, if a certain data in the abnormal data sequence is not an abnormal value, the corresponding abnormal characteristic value of the data in the abnormal index characteristic sequence is 0; if a certain data in the abnormal data sequence is an abnormal value, the corresponding abnormal characteristic value of the data in the abnormal index characteristic sequence is 1. For example, the abnormality index feature sequence corresponding to the abnormality data sequence [1.45, 1.54, 1.46, 1.60, 1.68, 1.65, 1.72, 1.61, 1.57, 1.53, 2.08, 2.27] is [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1 ].
In one example, the step S103 of inputting the abnormal data sequence and the abnormal index feature sequence into a trained time convolution network model to obtain a next predicted data value of the abnormal index output by the time convolution network model may include:
normalizing each data in the abnormal data sequence to obtain a normalized abnormal data sequence;
and inputting the normalized abnormal data sequence and the abnormal index characteristic sequence into a trained time convolution network model to obtain the next predicted data value of the abnormal index output by the time convolution network model.
Wherein, the normalization of the data can be performed according to the following formula (1):
wherein x isiFor any one data value, x, in the current abnormal data sequenceminIs the minimum value, x, in the current abnormal data sequencemaxIs the maximum value in the current abnormal data sequence.
When the time convolution network model is trained, the normalized abnormal data sequence is also used as training data, so that the influence of numerical dimension and volatility can be reduced, and the convergence speed of the model is improved.
Wherein, the time convolution network model can be trained by adopting a training mode in the related technology.
In one example, in training the time convolution network model, a 2 × 3 convolution kernel may be constructed, and the dilation convolution model dilation factor distorsion is [1, 2, 4, 8], and a skip-connected residual network may be added to speed up the training process and avoid the gradient vanishing problem of the deep learning model. The last layer of the time convolution network model is a full connection layer, and the activation function adopts ReLu. The output of the time convolution network model is the predicted next index data value, and the output dimension is 1 dimension.
After the time convolution network model is trained, a Mean-Square Error (MSE) can be used as a model evaluation index and a model training loss function loss, and the MSE can be calculated according to the following formula (2):
wherein x istIs the output value of the time convolution network model, i.e. the prediction data value (value after normalization of the prediction index data value), xiIs a value normalized by the actual value of the index data value. The smaller the value of MSE, the higher the accuracy of the time-convolutional network model.
In one example, the method may further comprise:
acquiring a next actual index data value of the abnormal index;
determining an abnormal characteristic value corresponding to the predicted data value according to the actual index data value and the predicted data value;
updating the abnormal index characteristic sequence according to the abnormal characteristic value, and updating the abnormal data sequence according to the actual index data value;
and optimizing the time convolution network model by using the updated abnormal index characteristic sequence and the updated abnormal data sequence.
The time convolution network model can be continuously optimized and updated, and the accuracy of the time convolution network model prediction result is improved.
For different first indicators, the time corresponding to the next actual indicator data value may be different. For example, for the index C, the time corresponding to the next actual index data value is a month from the generation time of the last data in the index data sequence of the index C at present; for the index D, the time corresponding to the next actual index data value is the time of one day from the generation time of the last data in the index data sequence at which the index D is currently generated.
In one example, determining the abnormal feature value corresponding to the predicted data value according to the actual index data value and the predicted data value may include:
determining a corresponding prediction index data value according to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
if the fluctuation function value is larger than a preset fluctuation threshold value, outputting a first value; the first value is used for indicating that the index is abnormal;
and if receiving the confirmed abnormal information returned aiming at the first value, setting the abnormal characteristic value corresponding to the predicted data value as the first value.
The prediction data value is a value obtained by normalizing the corresponding prediction index data value according to the formula (1), and therefore, the corresponding prediction index data value can be obtained by performing a process (inverse normalization) opposite to the formula (1) on the prediction data value.
The fluctuation function may be a logarithm of a difference between the actual index data value and the predicted index data value.
Wherein, the fluctuation threshold values corresponding to different indexes can be different.
Wherein the first value may be 1.
It should be noted that, if the fluctuation function value is less than or equal to the preset fluctuation threshold, a second value may be output, where the second value is used to indicate that the index is normal, and the abnormal characteristic value corresponding to the predicted data value is set as the second value. For example, the second value may be 0.
In one example, the method further comprises:
and determining whether alarm information indicating that the index is abnormal is output or not according to the predicted data value.
In one example, determining whether to output alarm information indicating an index anomaly based on the predicted data value includes:
acquiring a prediction index data value and an actual index data value corresponding to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
and if the fluctuation function value is larger than a preset fluctuation threshold value, determining to output alarm information indicating that the index is abnormal.
The alarm information indicating the index abnormality may be, for example, characters such as "data abnormality" and "index abnormality", or may be an icon of an alarm.
After seeing the alarm information indicating the index abnormality, the user can manually confirm whether the predicted data value is really abnormal, if the predicted data value is really abnormal, the user can input the confirmation abnormal information to confirm that the predicted data value is really abnormal, otherwise, the user can input the confirmation normal information to confirm that the predicted data value is not actually an abnormal value.
The index data series [1.45, 1.54, 1.46, 1.60, 1.68, 1.65, 1.72, 1.61, 1.57, 1.53, 2.08, 2.27] described above are also taken as examples. Assuming that a fluctuation function value corresponding to 13 th index data predicted by the time convolution network model is larger than a preset fluctuation threshold value, after the first value is output, if a user returns to confirm abnormal information, updating a corresponding abnormal index feature sequence to [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1 ]; and if the user returns the confirmed normal information, updating the corresponding abnormal index feature sequence to be [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0 ].
In one example, optimizing the time convolution network model using the updated anomaly index feature sequence and the updated anomaly data sequence may include:
updating the historical data of the abnormal indexes according to the updated abnormal index feature sequence and the updated abnormal data sequence;
constructing training data corresponding to the abnormal indexes according to the updated historical data of the abnormal indexes;
and training the time convolution network model based on the training data to obtain the time convolution network model with optimized parameters.
For example, as described above, assuming that the actual value of the next data in the index data sequence [1.45, 1.54, 1.46, 1.60, 1.68, 1.65, 1.72, 1.61, 1.57, 1.53, 2.08, 2.27] is 1.3, and the corresponding abnormal index feature sequence is [0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1], the index data sequence [1.45, 1.54, 1.46, 1.60, 1.68, 1.65, 1.72, 1.61, 1.57, 1.53, 2.08, 2.27, 1.3] and the abnormal index feature sequence [0, 0, 0, 0, 0, 0, 0, 0, 1, 1] are added to the historical data, the training data is reconstructed from all the historical data, the time-convolutional model is trained, and the time-convolutional model is optimized.
In one example, the method may further comprise:
for each second index, detecting the data of the second index according to the service threshold corresponding to the second index; the second indicator is different from the first indicator;
and if the detection result shows that the index data of the second index exceeds the service threshold corresponding to the second index, outputting alarm information indicating that the index is abnormal.
In practical application, if the current index type is identified as the second index through the tag information, the step of detecting the data of the second index according to the service threshold corresponding to the second index for each second index is executed.
The alarm information may include a data value of the index and a service threshold of the index, and may include information indicating that the index is abnormal, such as "index abnormal".
After the alarm information is output, the worker can confirm the alarm information.
In practical application, a worker can actively modify the service threshold corresponding to the second index according to the change condition of the service. For example, when the second index is the electricity consumption of a city, the threshold of the electricity consumption may be modified according to seasons, such as seasons in which the electricity consumption frequency is high for the air conditioners and the like in summer and winter, higher thresholds may be set for the electricity consumption in summer and winter, lower thresholds may be set for the electricity consumption in spring and autumn when the electricity consumption frequency is low for the air conditioners and the like in spring and autumn.
According to the data prediction method provided by the embodiment of the invention, the index data sequence corresponding to each first index is respectively detected by using the abnormal detection model to obtain the abnormal data sequence, the abnormal index characteristic sequence is determined according to the abnormal data sequence, the abnormal data sequence and the abnormal index characteristic sequence are input into the trained time convolution network model to obtain the next predicted data value of the abnormal index output by the time convolution network model, the predicted results of a plurality of indexes can be automatically obtained in real time, and better support is provided for auxiliary decision making.
According to the embodiment of the invention, through intelligent automatic prediction analysis, the input cost of human resources is reduced, the pressure of basic-level workers is reduced, and the management cost is reduced.
The embodiment of the invention can carry out accurate and efficient prediction and early warning on the abnormal index, improves the working efficiency, forms a feedback mechanism and continuously optimizes the model.
Based on the above method embodiment, the embodiment of the present invention further provides corresponding apparatus, device, and storage medium embodiments. For detailed implementation of the embodiments of the apparatus, device and storage medium of the embodiments of the present invention, please refer to the corresponding descriptions in the foregoing method embodiments.
Fig. 2 is a functional block diagram of a data prediction apparatus according to an embodiment of the present invention. As shown in fig. 2, in this embodiment, the data prediction apparatus may include:
the first detection module 210 is configured to respectively detect the index data sequences corresponding to the first indexes by using an anomaly detection model to obtain an anomaly data sequence; each data in the index data sequence is arranged according to the time sequence of data generation;
a determining module 220, configured to determine an abnormal index feature sequence according to the abnormal data sequence;
the prediction module 230 is configured to input the abnormal data sequence and the abnormal index feature sequence into a trained time convolution network model to obtain a next predicted data value of an abnormal index output by the time convolution network model, where the abnormal index is an index corresponding to the abnormal data sequence.
In one example, further comprising:
the acquisition module is used for acquiring the next actual index data value of the abnormal index;
an abnormal characteristic value determining module, configured to determine an abnormal characteristic value corresponding to the predicted data value according to the actual index data value and the predicted data value;
the updating module is used for updating the abnormal index characteristic sequence according to the abnormal characteristic value and updating the abnormal data sequence according to the actual index data value;
and the optimization module is used for optimizing the time convolution network model by using the updated abnormal index characteristic sequence and the updated abnormal data sequence.
In one example, the optimization module may be specifically configured to:
updating the historical data of the abnormal indexes according to the updated abnormal index feature sequence and the updated abnormal data sequence;
constructing training data corresponding to the abnormal indexes according to the updated historical data of the abnormal indexes;
and training the time convolution network model based on the training data to obtain the time convolution network model with optimized parameters.
In one example, the method for detecting the index data sequence by the anomaly detection model comprises the following steps:
determining the average value and the standard deviation of all data in the index data sequence according to all data values in the index data sequence;
determining a target interval range corresponding to the index data sequence according to the average value and the standard deviation;
and judging whether all data values in the index data sequence are in the target interval range, if so, determining that the index data sequence is a normal data sequence, otherwise, determining that the index data sequence is an abnormal data sequence, and determining that the data values in the index data sequence which are not in the target interval range are abnormal values.
In one example, the anomaly characteristic value determination module may be specifically configured to:
determining a corresponding prediction index data value according to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
if the fluctuation function value is larger than a preset fluctuation threshold value, outputting a first value; the first value is used for indicating index abnormity;
and if receiving the confirmed abnormal information returned aiming at the first value, setting the abnormal characteristic value corresponding to the predicted data value as the first value.
In one example, further comprising:
and the preprocessing module is used for preprocessing each data value in the index data sequence so as to enable the preprocessed data value to meet the input data requirement of the abnormality detection model.
In one example, the pre-processing module may be specifically configured to:
and carrying out logarithmic transformation on each data value in the index data sequence to obtain a logarithmic value corresponding to the data value.
In one example, the method may further include:
and the alarm module is used for determining whether alarm information indicating that the index is abnormal is output according to the predicted data value.
In one example, the alert module may be specifically configured to:
acquiring a prediction index data value and an actual index data value corresponding to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
and if the fluctuation function value is larger than a preset fluctuation threshold value, determining to output alarm information indicating that the index is abnormal.
In one example, the method may further include:
the second detection module is used for detecting the data of each second index according to the service threshold corresponding to the second index; the second indicator is different from the first indicator;
and the output module is used for outputting alarm information indicating that the index is abnormal if the detection result shows that the index data of the second index exceeds the service threshold corresponding to the second index.
In one example, the method may further include:
the reading module is used for reading label information used for indicating the index type from the current index data;
the identification module is used for identifying the type of the current index according to the label information;
and the execution module is used for executing the step of respectively detecting the index data sequences corresponding to the first indexes by using the abnormal detection model to obtain the abnormal data sequences if the current index type is the first index.
The embodiment of the invention also provides the electronic equipment. The electronic device may be the aforementioned server. Fig. 3 is a hardware structure diagram of an electronic device according to an embodiment of the present invention. As shown in fig. 3, the electronic apparatus includes: an internal bus 301, and a memory 302, a processor 303, and an external interface 304 connected through the internal bus.
The processor 303 is configured to read the machine-readable instructions in the memory 302 and execute the instructions to implement the following operations:
respectively detecting the index data sequences corresponding to the first indexes by using an abnormal detection model to obtain abnormal data sequences; each data in the index data sequence is arranged according to the time sequence of data generation;
determining an abnormal index characteristic sequence according to the abnormal data sequence;
and inputting the abnormal data sequence and the abnormal index characteristic sequence into a trained time convolution network model to obtain a next predicted data value of the abnormal index output by the time convolution network model, wherein the abnormal index is an index corresponding to the abnormal data sequence.
In one example, further comprising:
acquiring a next actual index data value of the abnormal index;
determining an abnormal characteristic value corresponding to the predicted data value according to the actual index data value and the predicted data value;
updating the abnormal index characteristic sequence according to the abnormal characteristic value, and updating the abnormal data sequence according to the actual index data value;
and optimizing the time convolution network model by using the updated abnormal index characteristic sequence and the updated abnormal data sequence.
In one example, optimizing the time convolution network model using the updated anomaly index feature sequence and the updated anomaly data sequence includes:
updating the historical data of the abnormal indexes according to the updated abnormal index feature sequence and the updated abnormal data sequence;
constructing training data corresponding to the abnormal indexes according to the updated historical data of the abnormal indexes;
and training the time convolution network model based on the training data to obtain the time convolution network model with optimized parameters.
In one example, the method for detecting the index data sequence by the anomaly detection model comprises the following steps:
determining the average value and the standard deviation of all data in the index data sequence according to all data values in the index data sequence;
determining a target interval range corresponding to the index data sequence according to the average value and the standard deviation;
and judging whether all data values in the index data sequence are in the target interval range, if so, determining that the index data sequence is a normal data sequence, otherwise, determining that the index data sequence is an abnormal data sequence, and determining that the data values in the index data sequence which are not in the target interval range are abnormal values.
In one example, determining an abnormal feature value corresponding to the predicted data value based on the actual metric data value and the predicted data value includes:
determining a corresponding prediction index data value according to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
if the fluctuation function value is larger than a preset fluctuation threshold value, outputting a first value; the first value is used for indicating index abnormity;
and if receiving the confirmed abnormal information returned aiming at the first value, setting the abnormal characteristic value corresponding to the predicted data value as the first value.
In one example, before the index data sequences corresponding to the first indexes are respectively detected by using an abnormal detection model to obtain abnormal data sequences, the method further includes:
and preprocessing each data value in the index data sequence to enable the preprocessed data value to meet the input data requirement of the anomaly detection model.
In one example, preprocessing individual data values in the index data sequence includes:
and carrying out logarithmic transformation on each data value in the index data sequence to obtain a logarithmic value corresponding to the data value.
In one example, further comprising:
and determining whether alarm information indicating that the index is abnormal is output or not according to the predicted data value.
In one example, determining whether to output alarm information indicating an index anomaly based on the predicted data value includes:
acquiring a prediction index data value and an actual index data value corresponding to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
and if the fluctuation function value is larger than a preset fluctuation threshold value, determining to output alarm information indicating that the index is abnormal.
In one example, the method may further include:
for each second index, detecting the data of the second index according to the service threshold corresponding to the second index; the second indicator is different from the first indicator;
and if the detection result shows that the index data of the second index exceeds the service threshold corresponding to the second index, outputting alarm information indicating that the index is abnormal.
In one example, before the index data sequences corresponding to the first indexes are respectively detected by using an abnormal detection model to obtain abnormal data sequences, the method further includes:
reading label information used for indicating the index type from the current index data;
identifying the current index type according to the label information;
and if the current index type is the first index, respectively detecting the index data sequence corresponding to each first index by using an abnormal detection model to obtain an abnormal data sequence.
An embodiment of the present invention further provides a computer-readable storage medium, where a plurality of computer instructions are stored on the computer-readable storage medium, and when executed, the computer instructions perform the following processing:
respectively detecting the index data sequences corresponding to the second indexes by using an abnormal detection model to obtain abnormal data sequences; each data in the index data sequence is arranged according to the time sequence of data generation;
determining an abnormal index characteristic sequence according to the abnormal data sequence;
and inputting the abnormal data sequence and the abnormal index characteristic sequence into a trained time convolution network model to obtain a next predicted data value of the abnormal index output by the time convolution network model, wherein the abnormal index is an index corresponding to the abnormal data sequence.
In one example, further comprising:
acquiring a next actual index data value of the abnormal index;
determining an abnormal characteristic value corresponding to the predicted data value according to the actual index data value and the predicted data value;
updating the abnormal index characteristic sequence according to the abnormal characteristic value, and updating the abnormal data sequence according to the actual index data value;
and optimizing the time convolution network model by using the updated abnormal index characteristic sequence and the updated abnormal data sequence.
In one example, optimizing the time convolution network model using the updated anomaly index feature sequence and the updated anomaly data sequence includes:
updating the historical data of the abnormal indexes according to the updated abnormal index feature sequence and the updated abnormal data sequence;
constructing training data corresponding to the abnormal indexes according to the updated historical data of the abnormal indexes;
and training the time convolution network model based on the training data to obtain the time convolution network model with optimized parameters.
In one example, the method for detecting the index data sequence by the anomaly detection model comprises the following steps:
determining the average value and the standard deviation of all data in the index data sequence according to all data values in the index data sequence;
determining a target interval range corresponding to the index data sequence according to the average value and the standard deviation;
and judging whether all data values in the index data sequence are in the target interval range, if so, determining that the index data sequence is a normal data sequence, otherwise, determining that the index data sequence is an abnormal data sequence, and determining that the data values in the index data sequence which are not in the target interval range are abnormal values.
In one example, determining an abnormal feature value corresponding to the predicted data value based on the actual metric data value and the predicted data value includes:
determining a corresponding prediction index data value according to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
if the fluctuation function value is larger than a preset fluctuation threshold value, outputting a first value; the first value is used for indicating index abnormity;
and if receiving the confirmed abnormal information returned aiming at the first value, setting the abnormal characteristic value corresponding to the predicted data value as the first value.
In one example, before the index data sequences corresponding to the first indexes are respectively detected by using an abnormal detection model to obtain abnormal data sequences, the method further includes:
and preprocessing each data value in the index data sequence to enable the preprocessed data value to meet the input data requirement of the anomaly detection model.
In one example, preprocessing individual data values in the index data sequence includes:
and carrying out logarithmic transformation on each data value in the index data sequence to obtain a logarithmic value corresponding to the data value.
In one example, further comprising:
and determining whether alarm information indicating that the index is abnormal is output or not according to the predicted data value.
In one example, determining whether to output alarm information indicating an index anomaly based on the predicted data value includes:
acquiring a prediction index data value and an actual index data value corresponding to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
and if the fluctuation function value is larger than a preset fluctuation threshold value, determining to output alarm information indicating that the index is abnormal.
In one example, the method may further include:
for each second index, detecting the data of the second index according to the service threshold corresponding to the second index; the second indicator is different from the first indicator;
and if the detection result shows that the index data of the second index exceeds the service threshold corresponding to the second index, outputting alarm information indicating that the index is abnormal.
In one example, before the index data sequences corresponding to the first indexes are respectively detected by using an abnormal detection model to obtain abnormal data sequences, the method further includes:
reading label information used for indicating the index type from the current index data;
identifying the current index type according to the label information;
and if the current index type is the first index, respectively detecting the index data sequence corresponding to each first index by using an abnormal detection model to obtain an abnormal data sequence.
For the device and apparatus embodiments, as they correspond substantially to the method embodiments, reference may be made to the partial description of the method embodiments for relevant points. The above-described embodiments of the apparatus are merely illustrative, wherein the modules described as separate parts may or may not be physically separate, and the parts displayed as modules may or may not be physical modules, may be located in one place, or may be distributed on a plurality of network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution in the specification. One of ordinary skill in the art can understand and implement it without inventive effort.
The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
Other embodiments of the present description will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This specification is intended to cover any variations, uses, or adaptations of the specification following, in general, the principles of the specification and including such departures from the present disclosure as come within known or customary practice within the art to which the specification pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the specification being indicated by the following claims.
It will be understood that the present description is not limited to the precise arrangements described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present description is limited only by the appended claims.
The above description is only a preferred embodiment of the present disclosure, and should not be taken as limiting the present disclosure, and any modifications, equivalents, improvements, etc. made within the spirit and principle of the present disclosure should be included in the scope of the present disclosure.
Claims (14)
1. A method of data prediction, comprising:
respectively detecting the index data sequences corresponding to the first indexes by using an abnormal detection model to obtain abnormal data sequences; each data in the index data sequence is arranged according to the time sequence of data generation;
determining an abnormal index characteristic sequence according to the abnormal data sequence;
and inputting the abnormal data sequence and the abnormal index characteristic sequence into a trained time convolution network model to obtain a next predicted data value of the abnormal index output by the time convolution network model, wherein the abnormal index is an index corresponding to the abnormal data sequence.
2. The method of claim 1, further comprising:
acquiring a next actual index data value of the abnormal index;
determining an abnormal characteristic value corresponding to the predicted data value according to the actual index data value and the predicted data value;
updating the abnormal index characteristic sequence according to the abnormal characteristic value, and updating the abnormal data sequence according to the actual index data value;
and optimizing the time convolution network model by using the updated abnormal index characteristic sequence and the updated abnormal data sequence.
3. The method of claim 2, wherein optimizing the time-convolutional network model using the updated anomaly index feature sequence and the updated anomaly data sequence comprises:
updating the historical data of the abnormal indexes according to the updated abnormal index feature sequence and the updated abnormal data sequence;
constructing training data corresponding to the abnormal indexes according to the updated historical data of the abnormal indexes;
and training the time convolution network model based on the training data to obtain the time convolution network model with optimized parameters.
4. The method of claim 1, wherein the anomaly detection model detects a sequence of metric data by:
determining the average value and the standard deviation of all data in the index data sequence according to all data values in the index data sequence;
determining a target interval range corresponding to the index data sequence according to the average value and the standard deviation;
and judging whether all data values in the index data sequence are in the target interval range, if so, determining that the index data sequence is a normal data sequence, otherwise, determining that the index data sequence is an abnormal data sequence, and determining that the data values in the index data sequence which are not in the target interval range are abnormal values.
5. The method of claim 2, wherein determining the abnormal feature value corresponding to the predicted data value based on the actual metric data value and the predicted data value comprises:
determining a corresponding prediction index data value according to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
if the fluctuation function value is larger than a preset fluctuation threshold value, outputting a first value; the first value is used for indicating index abnormity;
and if receiving the confirmed abnormal information returned aiming at the first value, setting the abnormal characteristic value corresponding to the predicted data value as the first value.
6. The method according to claim 1, wherein before the step of detecting the index data sequence corresponding to each first index by using the anomaly detection model to obtain the anomaly data sequence, the method further comprises:
and preprocessing each data value in the index data sequence to enable the preprocessed data value to meet the input data requirement of the anomaly detection model.
7. The method of claim 6, wherein preprocessing each data value in the index data sequence comprises:
and carrying out logarithmic transformation on each data value in the index data sequence to obtain a logarithmic value corresponding to the data value.
8. The method of claim 1, further comprising:
and determining whether alarm information indicating that the index is abnormal is output or not according to the predicted data value.
9. The method of claim 1, wherein determining whether to output alarm information indicative of an index anomaly based on the predicted data value comprises:
acquiring a prediction index data value and an actual index data value corresponding to the prediction data value;
determining a fluctuation function value according to the actual index data value and the predicted index data value;
and if the fluctuation function value is larger than a preset fluctuation threshold value, determining to output alarm information indicating that the index is abnormal.
10. The method of claim 1, further comprising:
for each second index, detecting the index data of the second index according to the service threshold corresponding to the second index; the second indicator is different from the first indicator;
and if the detection result shows that the index data of the second index exceeds the service threshold corresponding to the second index, outputting alarm information indicating that the index is abnormal.
11. The method according to claim 1, wherein before the step of detecting the index data sequence corresponding to each first index by using the anomaly detection model to obtain the anomaly data sequence, the method further comprises:
reading label information used for indicating the index type from the current index data;
identifying the current index type according to the label information;
and if the current index type is the first index, respectively detecting the index data sequence corresponding to each first index by using an abnormal detection model to obtain an abnormal data sequence.
12. A data prediction apparatus, comprising:
the first detection module is used for respectively detecting the index data sequences corresponding to the first indexes by using the abnormal detection model to obtain abnormal data sequences; each data in the index data sequence is arranged according to the time sequence of data generation;
the determining module is used for determining an abnormal index characteristic sequence according to the abnormal data sequence;
and the prediction module is used for inputting the abnormal data sequence and the abnormal index characteristic sequence into a trained time convolution network model to obtain a next predicted data value of the abnormal index output by the time convolution network model, wherein the abnormal index is an index corresponding to the abnormal data sequence.
13. An electronic device, comprising:
a memory for storing executable instructions of the processor;
the processor is used for executing the instructions to realize the method of any one of claims 1 to 11.
14. A computer-readable storage medium having stored thereon computer instructions which, when executed, implement the method of any one of claims 1 to 11.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110476640.XA CN112949951A (en) | 2021-04-29 | 2021-04-29 | Data prediction method, data prediction device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110476640.XA CN112949951A (en) | 2021-04-29 | 2021-04-29 | Data prediction method, data prediction device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112949951A true CN112949951A (en) | 2021-06-11 |
Family
ID=76233659
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110476640.XA Pending CN112949951A (en) | 2021-04-29 | 2021-04-29 | Data prediction method, data prediction device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112949951A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113362200A (en) * | 2021-07-02 | 2021-09-07 | 上海天麦能源科技有限公司 | Anomaly detection method and system of space-time dimension combination |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070055477A1 (en) * | 2005-09-02 | 2007-03-08 | Microsoft Corporation | Web data outlier detection and mitigation |
| CN110008079A (en) * | 2018-12-25 | 2019-07-12 | 阿里巴巴集团控股有限公司 | Monitor control index method for detecting abnormality, model training method, device and equipment |
| CN111126680A (en) * | 2019-12-11 | 2020-05-08 | 浙江大学 | Road section traffic flow prediction method based on time convolution neural network |
| US20200210393A1 (en) * | 2018-09-14 | 2020-07-02 | Verint Americas Inc. | Framework and method for the automated determination of classes and anomaly detection methods for time series |
| CN111552609A (en) * | 2020-04-12 | 2020-08-18 | 西安电子科技大学 | Abnormal state detection method, system, storage medium, program and server |
| CN112148768A (en) * | 2020-09-14 | 2020-12-29 | 北京基调网络股份有限公司 | Index time series abnormity detection method, system and storage medium |
| CN112527788A (en) * | 2020-12-17 | 2021-03-19 | 北京中恒博瑞数字电力科技有限公司 | Method and device for detecting and cleaning abnormal value of transformer monitoring data |
-
2021
- 2021-04-29 CN CN202110476640.XA patent/CN112949951A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070055477A1 (en) * | 2005-09-02 | 2007-03-08 | Microsoft Corporation | Web data outlier detection and mitigation |
| US20200210393A1 (en) * | 2018-09-14 | 2020-07-02 | Verint Americas Inc. | Framework and method for the automated determination of classes and anomaly detection methods for time series |
| CN110008079A (en) * | 2018-12-25 | 2019-07-12 | 阿里巴巴集团控股有限公司 | Monitor control index method for detecting abnormality, model training method, device and equipment |
| CN111126680A (en) * | 2019-12-11 | 2020-05-08 | 浙江大学 | Road section traffic flow prediction method based on time convolution neural network |
| CN111552609A (en) * | 2020-04-12 | 2020-08-18 | 西安电子科技大学 | Abnormal state detection method, system, storage medium, program and server |
| CN112148768A (en) * | 2020-09-14 | 2020-12-29 | 北京基调网络股份有限公司 | Index time series abnormity detection method, system and storage medium |
| CN112527788A (en) * | 2020-12-17 | 2021-03-19 | 北京中恒博瑞数字电力科技有限公司 | Method and device for detecting and cleaning abnormal value of transformer monitoring data |
Non-Patent Citations (1)
| Title |
|---|
| 邬伦等: ""从数字城管到智慧城管:平台实现与关键技术"", 《城市发展研究》, vol. 24, no. 6, 30 June 2017 (2017-06-30), pages 99 - 106 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113362200A (en) * | 2021-07-02 | 2021-09-07 | 上海天麦能源科技有限公司 | Anomaly detection method and system of space-time dimension combination |
| CN113362200B (en) * | 2021-07-02 | 2024-02-23 | 上海天麦能源科技有限公司 | Abnormality detection method and system for space-time dimension combination |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111459778B (en) | Operation and maintenance system abnormal index detection model optimization method, device and storage medium | |
| CN106951984B (en) | Dynamic analysis and prediction method and device for system health degree | |
| CA2471013A1 (en) | Method and system for analyzing and predicting the behavior of systems | |
| CN110059894B (en) | Equipment state evaluation method, device, system and storage medium | |
| CN111639798A (en) | Intelligent prediction model selection method and device | |
| CN116739829B (en) | Big data-based power data analysis method, system and medium | |
| CN113988723A (en) | User behavior locking method and system based on power consumption data anomaly analysis | |
| CN113868953A (en) | Multi-unit operation optimization method, device and system in industrial system and storage medium | |
| CN116882804A (en) | Intelligent power monitoring method and system | |
| CN111444075A (en) | Method for automatically discovering key influence indexes | |
| CN112949951A (en) | Data prediction method, data prediction device, electronic equipment and storage medium | |
| CN112380073B (en) | Fault position detection method and device and readable storage medium | |
| CN116362607B (en) | Material reserve efficiency evaluation method and device | |
| CN111783883A (en) | Abnormal data detection method and device | |
| CN117827593A (en) | Method and system for determining abnormality cause based on abnormality recognition result | |
| CN115934490A (en) | Server performance prediction model training method, device, equipment and storage medium | |
| CN114169590B (en) | Reservoir-in-storage runoff forecast correction method and system based on multi-scenario division | |
| CN115659271A (en) | Sensor abnormality detection method, model training method, system, device, and medium | |
| CN112418534B (en) | Method and device for predicting quantity of collected parts, electronic equipment and computer readable storage medium | |
| CN114037285A (en) | Distribution network automation application success analysis method and related system | |
| CN115278757A (en) | Method and device for detecting abnormal data and electronic equipment | |
| CN116401604B (en) | Method for classifying and detecting cold head state and predicting service life | |
| CN112926801A (en) | Load curve combined prediction method and device based on quantile regression | |
| CN112669093A (en) | Ocean economy prediction method, system, electronic device and storage medium | |
| CN115730811A (en) | Alarm method, alarm device, storage medium and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |