CN115858630A - Abnormity detection method for energy storage data of energy storage power station - Google Patents
Abnormity detection method for energy storage data of energy storage power station Download PDFInfo
- Publication number
- CN115858630A CN115858630A CN202310139290.7A CN202310139290A CN115858630A CN 115858630 A CN115858630 A CN 115858630A CN 202310139290 A CN202310139290 A CN 202310139290A CN 115858630 A CN115858630 A CN 115858630A
- Authority
- CN
- China
- Prior art keywords
- data
- energy storage
- value
- abnormal data
- distribution
- 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.)
- Granted
Links
Images
Landscapes
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The invention relates to the technical field of data processing, in particular to an abnormity detection method for energy storage data of an energy storage power station. The method comprises the steps of constructing a key energy storage matrix by obtaining time sequence data of energy storage parameters, screening suspected abnormal data through the key energy storage matrix, analyzing distribution characteristics of the suspected abnormal data and non-suspected abnormal data, and obtaining a distribution deviation index; obtaining a local data distribution index according to the local distribution characteristics of the local data of the suspected abnormal data; obtaining a data change index according to the data value change difference of the suspected abnormal data neighborhood data, constructing an abnormal data detection model according to the distribution deviation index, the local data distribution index and the data change index, obtaining real abnormal data, and judging the abnormal condition of the energy storage power station according to the real abnormal data. According to the invention, through data processing, abnormal data are comprehensively and accurately detected, and the abnormal detection of the energy storage power station is realized.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to an abnormity detection method for energy storage data of an energy storage power station.
Background
In the installation of energy storage power station and the use of commissioning, the energy storage power station battery quality is good bad and ageing speed all is wayward, along with the continuous increase of live time, the security of energy storage power station just also can be more and more poor, and the risk of occurence of failure also can increase gradually. The energy storage power station has the main component of an energy storage battery, and has various different operating parameters such as voltage, current, resistance and the like, and the change of the operating parameters directly affects the comprehensive performance of the energy storage power station, so that the safety becomes one of the core problems concerned in the field of energy storage, and the method is an extremely important process for detecting the safety of the energy storage power station.
At present, there are different energy storage batteries in different energy storage power stations, and it is because use the scene with correspond the difference of time quantum, and the data change that corresponds the energy storage parameter is also different, consequently adopts traditional neural network training model to carry out unusual data detection, can not be fine detect out the abnormal value in the energy storage parameter under the different situation, and the cost volume of training according to particular case is great in addition, and the testing result is also not accurate enough. The abnormal data in different energy storage power stations are monotonously detected at the present stage, and the abnormal data in the energy storage data within a period of time cannot be comprehensively, quickly and accurately detected, so that the real-time abnormal condition of the energy storage power station is not accurately judged.
Disclosure of Invention
In order to solve the technical problems that in the prior art, abnormal data in different energy storage power stations are monotonously detected and cannot be comprehensively, quickly and accurately detected within a period of time, the invention aims to provide an abnormal detection method for energy storage data of an energy storage power station, and the adopted technical scheme is as follows:
the invention provides an abnormity detection method for energy storage data of an energy storage power station, which comprises the following steps:
acquiring time sequence data of at least two types of energy storage parameters in a preset time period through a sensor, and constructing a key energy storage data matrix according to the time sequence data; screening suspected abnormal data in the key energy storage data matrix according to the data fluctuation degree of each energy storage data in each type of energy storage parameters in the key energy storage data matrix;
acquiring distribution characteristic difference between actual distribution characteristics corresponding to the suspected abnormal data and standard distribution characteristics corresponding to the non-suspected abnormal data in the key energy storage data matrix, and acquiring distribution deviation indexes of each suspected abnormal data according to the distribution characteristic difference;
in the key energy storage data matrix, obtaining a local data distribution index of each suspected abnormal data according to the local distribution characteristics of the data values in different preset directions in a preset local window corresponding to each suspected abnormal data; acquiring a data change matrix according to the data value difference condition of the energy storage data in the preset neighborhood range corresponding to the suspected abnormal data, and acquiring a data change index of the suspected abnormal data according to the data value change matrix;
obtaining an abnormal data detection model according to the distribution deviation index, the local data distribution index and the data change index of the suspected abnormal data; and screening out real abnormal data in the suspected abnormal data according to the abnormal data detection model, obtaining an abnormal degree judgment index according to the data value of the real abnormal data, and detecting the abnormal condition of the energy storage power station according to the abnormal degree judgment index.
Further, screening out suspected abnormal data in the key energy storage data matrix according to the data fluctuation degree of each energy storage data in each type of energy storage parameter in the key energy storage data matrix comprises:
taking the absolute value of the difference between each energy storage data in the key energy storage data matrix and other energy storage data in the same type of energy storage parameters as a data difference, and taking the mean value of the data difference as a data fluctuation degree index of the energy storage data;
acquiring data fluctuation degree indexes of all energy storage data; when the data fluctuation degree index is larger than a preset fluctuation threshold value, taking the corresponding energy storage data as suspected abnormal data; and when the data fluctuation degree index is less than or equal to a preset fluctuation threshold value, taking the corresponding energy storage data as non-suspected abnormal data.
Further, the obtaining of the standard distribution characteristics and the actual distribution characteristics of the suspected abnormal data includes:
fitting non-suspected abnormal data in the key energy storage data by adopting a Gaussian model to obtain a standard Gaussian model, and taking a standard Gaussian model value corresponding to each energy storage data as a standard distribution characteristic of an energy storage parameter;
and adopting Gaussian model fitting to suspected abnormal data in the key energy storage data to obtain an abnormal data Gaussian model, and taking an abnormal data Gaussian model value corresponding to each suspected abnormal data as an actual distribution characteristic of the suspected abnormal data.
Further, the obtaining of the distribution deviation index includes:
taking the difference absolute value of the abnormal data Gaussian model value corresponding to the actual distribution characteristic of each suspected abnormal data and the standard Gaussian model value corresponding to the standard distribution characteristic as the distribution characteristic difference; and carrying out normalization processing on each distribution characteristic difference to obtain a distribution deviation index of each suspected abnormal data.
Further, the obtaining of the local distribution characteristics comprises:
the local distribution characteristics comprise data value distribution characteristic values and category distribution characteristic values;
dividing energy storage data with the same data value in a preset local window corresponding to each suspected abnormal data into the same data value type to obtain all data value types; taking the combination of data values corresponding to any two data value types in the data value types as a type binary group to obtain all types of binary groups;
traversing in the horizontal direction of a preset local window corresponding to suspected abnormal data by taking the size of each kind of binary group as the size of a sliding window, and counting the frequency which is the same as the data value in each kind of binary group as the horizontal frequency; similarly, the vertical frequency of traversal of each kind of binary group in the vertical direction is obtained; counting the frequency of all kinds of binary groups in a preset local window of the suspected abnormal data as a total frequency;
obtaining a horizontal entropy value according to the ratio of the horizontal frequency of each kind of binary group to the total frequency; obtaining a vertical entropy value according to the proportion of the vertical frequency of each kind of binary group in the total frequency; adding the horizontal entropy value and the vertical entropy value to obtain a category distribution characteristic value;
and taking the absolute value of the difference between the mean value of the data values of all the energy storage data in the preset local window and the mean value of the data values of the non-suspected abnormal data as a data value distribution characteristic value.
Further, the obtaining of the local data distribution index includes:
and taking the product of the data value distribution characteristic value and the class distribution characteristic value as a local data distribution index.
Further, the obtaining of the data change index of the suspected abnormal data according to the data value change matrix includes:
calculating the sum of the difference values of the suspected abnormal data and each energy storage data in each preset direction in the corresponding preset neighborhood range to obtain the integral data value difference in the corresponding preset direction; taking a square matrix constructed by arranging the integral data value differences in all preset directions as a data value change matrix;
and obtaining the eigenvalue of the data value change matrix, multiplying the absolute values of all eigenvalues, and taking the product of the obtained eigenvalues as a data change index.
Further, the obtaining of the abnormal data detection model comprises:
comparing the product of the distribution deviation index of the suspected abnormal data and the local data distribution index with the data change index to obtain a first ratio; and carrying out normalization processing on the first ratio to obtain an abnormal data detection model.
Further, the obtaining of the abnormality degree determination index includes:
and acquiring the accumulated values of the data values of all the real abnormal data, and carrying out normalization processing on the accumulated values to acquire an abnormal degree judgment index.
Further, the obtaining of the key energy storage data matrix comprises:
normalizing the energy storage data in each type of energy storage parameter according to the data values in the same type of energy storage parameters to obtain normalized time sequence data; taking the time information and the energy storage parameters in all the normalized time sequence data as rows and columns of the matrix to obtain an energy storage data matrix;
and reducing the dimension of the energy storage data matrix by adopting a principal component analysis method, and obtaining a matrix reconstructed by the time information and the key energy storage parameter data as a key energy storage data matrix.
The invention has the following beneficial effects:
1. the invention comprehensively analyzes the time sequence data of various energy storage parameters in a matrix mode, can better reflect the abnormal conditions of the energy storage power station, and is applicable to various scenes. The suspected abnormal data are obtained by preliminarily judging the fluctuation degree of the data in the key energy storage data matrix, and then the most accurate real abnormal data are screened out by comprehensively analyzing the distribution deviation characteristics of the suspected abnormal data and the distribution change characteristics of the local data, so that the abnormal data are more accurately and quickly detected, and the abnormal condition of the energy storage power station is better pre-warned.
2. For the screening of the suspected abnormal data, in consideration of certain fluctuation condition of the normal energy storage data, the deviation degree of the suspected abnormal data to the non-suspected abnormal data is obtained through the distribution deviation condition of the suspected abnormal data, and the distribution deviation index reflecting the deviation degree is calculated and used for screening the real abnormal data subsequently. Meanwhile, the influence of noise data is also considered in the process of accurately acquiring the real abnormal data, and the influence of the noise data in the suspected abnormal data can be eliminated by acquiring a local data distribution index and a data change index according to the distribution uniformity condition and the data value change condition of the local data around the suspected abnormal data. And screening the comprehensive distribution deviation index, the local data distribution index and the data change index to obtain real abnormal data, so that the obtained real abnormal data is more accurate.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of an abnormality detection method for energy storage data of an energy storage power station according to an embodiment of the present invention.
Detailed Description
To further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description, structures, features and effects of the method for detecting the abnormality of the energy storage data of the energy storage power station according to the present invention are provided with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following describes a specific scheme of the method for detecting the abnormality of the energy storage data of the energy storage power station in detail with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of an anomaly detection method for energy storage data of an energy storage power station according to an embodiment of the present invention is shown, where the method includes the following steps:
s1: acquiring time sequence data of at least two types of energy storage parameters in a preset time period through a sensor, and constructing a key energy storage data matrix according to the time sequence data; and screening suspected abnormal data in the key energy storage data matrix according to the data fluctuation degree of each energy storage data in each type of energy storage parameter in the key energy storage data matrix.
The invention mainly aims at the energy storage data in the energy storage power station to detect the abnormal data, and then judges the abnormal condition in time according to the abnormal data so as to prevent the occurrence of the risk accident. Because the most main part of the energy storage power station is the energy storage battery, the data parameter change of the energy storage battery can directly influence the operation efficiency and the comprehensive performance of the energy storage system in the whole energy storage power station, and therefore when the energy storage data in the energy storage power station is analyzed, the mainly analyzed energy storage data are the data in the energy storage battery parameters, such as the temperature, the current, the voltage, the energy storage SOC value and the like of the energy storage battery. In order to comprehensively analyze the abnormal condition of the energy storage data in the energy storage power station, the time sequence data of at least two types of energy storage parameters in a preset time period is obtained through a sensor, and a key energy storage data matrix is constructed according to the time sequence data, and the method specifically comprises the following steps:
different data sequences may be generated by various energy storage parameters at different moments due to external influences, and in order to more accurately detect abnormal data in the current time period, the data sequences with moment information need to be acquired during data acquisition. Therefore, the invention adopts the corresponding data acquisition sensor to acquire the real-time data of the energy storage parameters, and obtains the time sequence data of at least two types of energy storage parameters in the preset time period.
In the embodiment of the present invention, the data of the energy storage parameter is acquired once every T time period, the length of the time sequence data set for each acquisition is m, the specific set interval time T is 0.5s, and the length m of the time sequence data is 800, and the data may be specifically set according to the actual situation, which is not limited herein. It should be noted that there may be requirements for acquiring different energy storage parameters for different energy storage power stations, and therefore, for selection of the type and model of the sensor for data acquisition, an implementer may select the type and model according to actual conditions, and the type of the acquired energy storage parameters is not limited herein.
On the basis of all the acquired time sequence data, all the energy storage parameters are comprehensively analyzed in a matrix construction mode, in order to avoid the influence of different dimensions of various energy storage parameters, the energy storage data in each type of energy storage parameters are normalized according to the data values in the same type of energy storage parameters, and the normalized time sequence data is obtained, so that the subsequent analysis is facilitated. The specific method for constructing the energy storage data matrix according to the time sequence data comprises the following steps:
and taking the time information and the energy storage parameters in all the normalized time sequence data as the rows and the columns of the matrix to obtain an energy storage data matrix. In the embodiment of the present invention, each row of the energy storage data matrix represents the same time, and each column represents the same energy storage parameter, and for the accuracy of the subsequent calculation, the energy storage data matrix specifically includes:
in the formula (I), the compound is shown in the specification,
expressed as having a size>
Is selected based on the stored energy data matrix, < > is selected>
Expressed as a total time duration of the moment corresponding to the energy storage data, is/are>
Expressed as a total number of classes of the energy storage parameter corresponding to the energy storage data, <' >>
Expressed as a number one>
The instant at which the first->
And normalizing the energy storage data by the class energy storage parameters.
In order to further improve the accuracy of abnormal data analysis, reduce the data volume and improve the detection rate, the dimension reduction processing of the data is realized by reconstructing the energy storage data matrix, preferably, the dimension reduction is performed on the energy storage data matrix by adopting a principal component analysis method, and the principal component analysis method can recombine a plurality of indexes with certain correlation to obtain a group of mutually unrelated comprehensive indexes to replace the original indexes.
Taking a matrix reconstructed by the time information and the key energy storage parameter data as a key energy storage data matrix, wherein in the embodiment of the invention, the key energy storage data matrix is
The size of the reconstruction matrix is ≥>
Wherein->
，/>
Expressed as a total time duration of the moment corresponding to the energy storage data, is/are>
Expressed as a total number of classes of the energy storage parameter corresponding to the energy storage data, <' >>
The total number of the key energy storage parameters after dimension reduction is expressed, and it should be noted that the principal component analysis method is a technical means well known to those skilled in the art, and is not described herein again.
And after the key energy storage data matrix is obtained, abnormal data analysis of the energy storage power station is carried out through the key energy storage data matrix. Under normal conditions, in a key energy storage data matrix corresponding to the energy storage power station, corresponding energy storage data in the energy storage parameters have consistency, namely, the change degree of the data value is small. When an abnormal condition occurs, the abnormal data has a larger difference compared with other energy storage data, so that the data fluctuation degree of each energy storage data is preliminarily analyzed in the abnormal data detection, suspected abnormal data in the key energy storage data matrix is screened out according to the data fluctuation degree of each energy storage data in each type of energy storage parameters in the key energy storage data matrix, and the method specifically comprises the following steps:
the difference absolute value between each energy storage data in the key energy storage data matrix and other energy storage data in the corresponding energy storage parameter is used as a data difference, the data difference can reflect the difference degree of the data values between the energy storage data and the other energy storage data in the same energy storage parameter, and when the data difference is larger, the larger difference exists between the two energy storage data.
Taking the mean value of all data differences corresponding to each energy storage data as the data fluctuation degree index of the energy storage data, wherein the data fluctuation degree index can reflect the difference situation of the corresponding energy storage data compared with other energy storage data, the larger the difference is, the more likely the energy storage data is abnormal data, and the expression of the specific data fluctuation degree index is as follows:
in the formula (I), the compound is shown in the specification,
expressed as energy storage data ≥ in the key energy storage data matrix>
Corresponding data fluctuation degree indexes;
expressed as the ^ th or greater in the key energy storage data matrix>
Line is on the fifth or fifth side>
Energy storage data for the column; />
And the total duration of the corresponding moment of the energy storage data, namely the total row number of the key energy storage data matrix, is represented.
In the embodiment of the invention, in order to better screen out the energy storage data of the suspected abnormal data through the data fluctuation degree index, the data fluctuation degree index is normalized, and the data fluctuation degree indexes of the subsequent analysis are the normalized data fluctuation degree indexes.
Acquiring data fluctuation degree indexes of all energy storage data, and when the data fluctuation degree indexes are larger than a preset fluctuation threshold, indicating that the corresponding energy storage data has larger data value difference compared with other energy storage data in the same energy storage parameter, and recording the corresponding energy storage data as suspected abnormal data; and when the data fluctuation degree index is smaller than or equal to the preset fluctuation threshold value, the difference of the data value of the corresponding energy storage data is smaller than that of other energy storage data in the same energy storage parameter, and the corresponding energy storage data is used as non-suspected abnormal data. In the embodiment of the present invention, the preset fluctuation threshold is 0.5.
Therefore, all suspected abnormal data in the key energy storage data matrix can be obtained, a suspected abnormal data set is formed, and the suspected abnormal data can be further analyzed.
S2: and obtaining the distribution characteristic difference between the actual distribution characteristic corresponding to the suspected abnormal data and the standard distribution characteristic corresponding to the non-suspected abnormal data in the key energy storage data matrix, and obtaining the distribution deviation index of each suspected abnormal data according to the distribution characteristic difference.
According to the S1, a suspected abnormal data set can be obtained, so that real abnormal data can be identified more accurately, the abnormal condition of the energy storage power station can be detected, the energy storage data in the suspected abnormal data set can be further analyzed, and more accurate abnormal data can be screened. In the process of acquiring the suspected abnormal data, there may be normal energy storage data, that is, non-suspected abnormal data, which has a large fluctuation degree corresponding to the non-suspected abnormal data, so that the distribution characteristics of the suspected abnormal data are analyzed first to obtain a distribution characteristic difference between an actual distribution characteristic corresponding to the suspected abnormal data and a standard distribution characteristic corresponding to the non-suspected abnormal data in the key energy storage data matrix, and a distribution deviation index of each suspected abnormal data is obtained according to the distribution characteristic difference, which specifically includes:
in order to better reflect the distribution condition of each suspected abnormal data, considering that the gaussian model can represent the distribution condition of the data, preferably, the suspected abnormal data in the key energy storage data is fitted by the gaussian model to obtain an abnormal data gaussian model, and the abnormal data gaussian model value corresponding to each suspected abnormal data is used as the actual distribution characteristic of the suspected abnormal data to reflect the actual distribution condition of the suspected abnormal data.
Meanwhile, in order to better compare the distribution situation of the suspected abnormal data, a normal distribution situation is required to be obtained for other non-suspected abnormal data in the key energy storage data matrix, namely normal data, to be compared and analyzed, preferably, a gaussian model is adopted for fitting the non-suspected abnormal data in the key energy storage data to obtain a standard gaussian model, and a standard gaussian model value corresponding to each energy storage data is used as a standard distribution characteristic of the energy storage parameter to reflect the distribution situation corresponding to each energy storage data in the non-suspected abnormal data. It should be noted that the gaussian model fitting is a well-known technical means for those skilled in the art, and is not described herein.
Taking the absolute value of the difference between the abnormal data Gaussian model value corresponding to the actual distribution characteristic of each suspected abnormal data and the standard Gaussian model value corresponding to the standard distribution characteristic as the distribution characteristic difference, wherein the distribution characteristic difference reflects the distribution deviation degree of each suspected abnormal data, and when the distribution characteristic difference is larger, the distribution deviation degree of the data value distribution of the suspected abnormal data and the corresponding ideal normal energy storage data is larger, so that the distribution characteristic difference of each suspected abnormal data is normalized to obtain the distribution deviation index of each suspected abnormal data, and the expression of the distribution deviation index is as follows:
in the formula (I), the compound is shown in the specification,
indicates as suspected abnormal data->
A distribution deviation index of (a); />
Expressed as the ^ th/th in the key energy storage data matrix>
Line is on the fifth or fifth side>
The listed energy storage data are marked as suspected abnormal data; />
Indicates as suspected abnormal data->
Corresponding abnormal data Gaussian model values, namely actual distribution characteristics; />
Indicates as suspected abnormal data->
The corresponding standard Gaussian model value, namely the standard distribution characteristic; />
Expressed as natural constants. />
In the expression, the expression is shown,
expressed as distribution characteristic difference, when the distribution characteristic difference is larger, the distribution deviation index is larger, and the corresponding suspected abnormality is indicatedThe higher the deviation degree of the distribution of the constant data is, the more probable the suspected abnormal data is to be the real abnormal data, so that the distribution characteristic difference and the distribution deviation index have positive correlation and are/is judged>
Expressed as a difference in a bisecting profile>
And carrying out normalization processing.
And obtaining distribution deviation indexes of all suspected abnormal data, and taking the distribution deviation index of each suspected abnormal data as one of indexes for abnormal data detection.
S3: in the key energy storage data matrix, obtaining a local data distribution index of each suspected abnormal data according to the local distribution characteristics of data values in different preset directions in a preset local window corresponding to each suspected abnormal data; and obtaining a data change matrix according to the data value difference condition of the energy storage data in the preset neighborhood range corresponding to the suspected abnormal data, and obtaining the data change index of the suspected abnormal data according to the data value change matrix.
In the suspected abnormal data set, certain noise data may be mistaken as suspected abnormal data, and the noise data affects the accuracy of abnormal data detection, so that a large error exists in the abnormal condition analysis of the final energy storage power station. Considering that the noise data has the characteristic of isolation, the data distribution of the noise data corresponding to the periphery is uniform, and the data value of the noise data is mostly a sudden change value, so that the local energy storage data around each suspected abnormal data is analyzed, and the corresponding index is obtained according to the distribution uniformity degree of the local data and the data value change degree, so that the subsequent judgment is facilitated.
S3.1: firstly, analyzing the local data distribution uniformity degree of suspected abnormal data, and in a key energy storage data matrix, obtaining a local data distribution index of each suspected abnormal data according to the local distribution characteristics of data values in different preset directions in a preset local window corresponding to each suspected abnormal data, wherein the method specifically comprises the following steps:
according to the isolation of the noise data, the distribution of the energy storage data in the local range of the noise data is regular and uniform, namely the consistency of the energy storage data is high, so that the distribution condition of the local data of each suspected abnormal data is analyzed to obtain the degree of uniformity of the distribution of the local data reflected by the local distribution characteristics. According to the embodiment of the invention, the local distribution characteristics are reflected through two characteristic values, namely the data distribution characteristic value and the category distribution characteristic value, wherein the category distribution characteristic value mainly reflects the data value distribution condition of the local data of the suspected abnormal data, and in order to better represent the distribution regularity of the data value values of the local data, the category of the data value of the local data is firstly counted.
The specific data value categories are divided into: the method comprises the steps of obtaining a preset local window taking each suspected abnormal data as a center, dividing energy storage data with the same data value in the preset local window corresponding to each suspected abnormal data into the same data value type, wherein different data values are different data value types.
The total number of data value types is counted, in order to reflect the distribution rule condition of local data more clearly, data value types are combined into type duplets, the distribution condition of local data is analyzed through the type duplets, the combination of data values corresponding to any two data value types in the data value types is used as the type duplets, specifically, for example, when there are four data values of 1,2,3,4 in a preset local window, the corresponding data value types are also four, when the data values of any two data types are combined, each data type can be combined with itself, namely, the obtained type duplets are (1, 1), (1, 2), (1, 3), (1, 4), (2, 1), (2, 2) \\ 8230; (4, 4), and 16 type duplets can be obtained in total.
Taking the size range corresponding to each kind of binary group as the size of the sliding window, that is, the size of the sliding window is 1 × 2, analyzing the data in the preset local window in different preset directions through the sliding window, and counting the frequency of each kind of binary group in different preset directions in the local window, preferably, selecting different preset directions as the horizontal direction and the vertical direction, specifically: traversing each kind of binary group in the horizontal direction of a preset local window corresponding to the suspected abnormal data, and counting the frequency which is the same as the data value in each kind of binary group as the horizontal frequency; similarly, the vertical frequency of each kind of binary group traversed in the vertical direction is obtained, and the frequency of all kinds of binary groups in the preset local window of the suspected abnormal data is counted as the total frequency. It should be noted that, if the position of the suspected abnormal data in the key energy storage data matrix is located at the edge of the matrix, only a part of the energy storage data exists in the size of the preset local window, and in the traversal process, only the energy storage data existing in the preset local window is traversed.
The frequency of all kinds of binary groups in different preset directions is obtained, and the chaos degree of data value distribution in a preset local window of suspected abnormal data can be reflected by combining the operation of entropy, and the method specifically comprises the following steps: obtaining a horizontal entropy value according to the ratio of the horizontal frequency of each kind of binary group to the total frequency; and obtaining the vertical entropy value according to the proportion of the vertical frequency of each kind of binary group in the total frequency.
Through entropy values of the category binary group in the horizontal direction and the vertical direction, category distribution characteristic values can be obtained, wherein the category distribution characteristic values mainly represent the local data distribution regularity of the suspected abnormal data, and specifically include: and adding the horizontal entropy value and the vertical entropy value to obtain a class distribution characteristic value. When the class characteristic value is larger, it is indicated that the local data corresponding to the suspected abnormal data is distributed more irregularly, the distribution condition is more disordered, and the corresponding suspected abnormal data is more likely to be real abnormal data.
After the species distribution characteristic value of the suspected abnormal data local data is obtained, the data value distribution characteristic of the local data is further analyzed, because the species distribution characteristic is only analyzed for the distribution of the data value, the size of the specific data value is not judged, when the suspected abnormal data presets the data value mean value in the local window, and the difference of the data value mean value of the non-suspected abnormal data is large, it is indicated that the local data corresponding to the suspected abnormal data is more uneven compared with the non-suspected abnormal data, the suspected abnormal data is more likely to be real abnormal data, and therefore the absolute value of the difference between the mean value of the data values of all the energy storage data in the preset local window and the mean value of the data values of the non-suspected abnormal data is taken as the data value distribution characteristic value. When the larger the data value distribution characteristic value is, the larger the difference between the data value of the local data of the suspected abnormal data and the data value of the non-suspected abnormal data is, the more likely the suspected abnormal data is to be the real abnormal data.
According to the type distribution characteristic value and the data value distribution characteristic value of the local distribution characteristic, obtaining a local data distribution index, taking the product of the data value distribution characteristic value and the type distribution characteristic value as the local data distribution index, and reflecting the distribution uniformity degree of the local data of the suspected abnormal data through the local data distribution index, wherein in the embodiment of the invention, for the accuracy of subsequent calculation, the expression of the specific local data distribution index is as follows:
in the formula (I), the compound is shown in the specification,
indicates as suspected abnormal data->
Is based on the local data distribution indicator of->
Expressed as a first ÷ in a key energy storage data matrix>
Line is on the fifth or fifth side>
The energy storage data marked as suspected abnormal data of the column is/are>
Is expressed as suspected abnormalityConstant data>
A mean value of the data values in the preset local window, <' > or>
Mean, or |, of data values that are represented as non-suspected abnormal data in the key energy storage data matrix>
Is expressed as a fifth->
The frequency of the seed kind doublet in the horizontal direction in the preset local window is judged>
Is expressed as a fifth->
The frequency of the seed type doublet in the vertical direction in the preset local window is/are selected>
Expressed as the total frequency of all kinds of tuples in the preset partial window, <' >>
Indicates as suspected abnormal data->
A total number of classes of doublets of classes in a predetermined partial window->
Expressed as a logarithmic function based on natural constants.
Combining the data value distribution characteristic value and the class distribution characteristic value by adopting a multiplication mode, wherein
The data value distribution characteristic value is expressed as suspected abnormal data, and when the data value distribution characteristic value is larger, the corresponding suspected abnormal data is indicatedThe larger the difference between the data value of the local data and the data value of the non-suspected abnormal data is, the more uneven the data value of the local data corresponding to the suspected abnormal data is, and the larger the local data distribution index is.
In the formula (I), the compound is shown in the specification,
representing the level entropy value obtained according to the proportion of the level frequency of each kind of binary group in the total frequency in a preset local window for the suspected abnormal data, and then judging whether the value is greater than or equal to the preset value>
The vertical entropy values obtained from the vertical frequency of each kind of binary group in the total frequency ratio in the local window are preset for the suspected abnormal data,
the data value distribution characteristic value is larger when the entropy value is larger, which indicates that the distribution situation of the data value in the local data of the suspected abnormal data is more disordered, and the local data distribution is more irregular, so that the local data distribution index is larger, and the suspected abnormal data is more likely to be real abnormal data. It should be noted that the application of the formula of entropy value is a technical means well known to those skilled in the art, and therefore the meaning of the specific formula is not described in detail.
And analyzing the local data distribution uniformity of the suspected abnormal data.
S3.2: further, analyzing the data value change degree of the local data of the suspected abnormal data, obtaining a data change matrix according to the data value difference condition of the energy storage data in the preset neighborhood range corresponding to the suspected abnormal data, and obtaining the data change index of the suspected abnormal data according to the data value change matrix specifically includes:
for the abnormal condition of the energy storage power station, when abnormal data is generated, the energy storage data adjacent to the abnormal data also has a certain degree of change, and the isolation of noise point data is also reflected in the form that the change of the data value is a sudden change, so that the data value change of the local data of each suspected abnormal data is analyzed to obtain the data value change degree of the suspected abnormal data compared with the local data. According to the method, the data value change characteristics of the local data of the suspected abnormal data are represented in a matrix form, and a data value change matrix is established for the suspected abnormal data.
Obtaining data change values of the suspected abnormal data and the local data, preferably, calculating the sum of difference values of the suspected abnormal data and each energy storage data in each preset direction in the corresponding preset neighborhood range, obtaining the overall data value difference in the corresponding preset direction, wherein each overall data value difference reflects the gradient change condition of the data values of the suspected abnormal data in the corresponding preset direction, and taking a square matrix constructed by arranging the overall data value differences in all preset directions as a data value change matrix.
In the embodiment of the present invention, the preset neighborhood range is an eight neighborhood range adjacent to the suspected abnormal data, and the preset directions are a horizontal direction, a vertical direction, a 45-degree linear direction, and a 135-degree linear direction. Therefore, the suspected abnormal data
For example, the overall data value difference in the horizontal direction is ≥>
The global data value difference in the vertical direction is ≥>
With a variance of ≦ based on the overall data value in the 45 degree linear direction>
The overall data value difference in the 135 degree linear direction is
. The obtained integral data value differences in four preset directions are arranged to obtain suspected abnormal data->
The corresponding numerical variation matrix is:
in the formula (I), the compound is shown in the specification,
indicates as suspected abnormal data->
Is changed by the data value change matrix, < > is greater or less>
Indicates as suspected abnormal data->
Overall data value difference in the horizontal direction>
Data expressed as suspect anomaly>
Overall data value difference in vertical direction>
Indicates as suspected abnormal data->
The overall data value difference in the 45 degree linear direction, device for selecting or keeping>
Indicates as suspected abnormal data->
The overall data value difference at 135 degree line direction.
Further, calculating the eigenvalue of the data value change matrix, wherein the magnitude of the eigenvalue is used for representing the data value change degree of the eigenvalue in the direction corresponding to the eigenvector, and when the absolute value of the eigenvalue is larger, the data value change degree of the suspected abnormal data in the direction corresponding to the eigenvector is considered to be larger, so that the obtained number is largerIn the embodiment of the present invention, since the data value change matrix is a square matrix of 2 × 2, two eigenvalues, that is, suspected abnormal data, can be obtained
The corresponding data change index expression is as follows:
in the formula (I), the compound is shown in the specification,
indicates as suspected abnormal data->
In the data change indicator of (1), based on the number of the preceding frames in the frame>
Indicates as suspected abnormal data->
A characteristic value, corresponding to the data-value change matrix>
Indicates as suspected abnormal data->
Another eigenvalue of the corresponding data value change matrix.
And analyzing the data value change degree in a product form, wherein the larger the absolute value of all the characteristic values is, the larger the data value change degree of the suspected abnormal data in each direction is, the more likely the suspected abnormal data is to be noise data, the larger the corresponding data change index is, and the positive correlation relationship between the absolute value of the characteristic value and the data change index is formed.
Therefore, comprehensive analysis of the data distribution uniformity degree and the data value change degree of each suspected abnormal data local data is completed, the local data distribution index and the data change index of each suspected abnormal data are obtained, the influence of noise data on abnormal data detection can be greatly reduced, and the abnormal data detection is more accurate.
S4: obtaining an abnormal data detection model according to the distribution deviation index, the local data distribution index and the data change index of the suspected abnormal data; and screening out real abnormal data in the suspected abnormal data according to the abnormal data detection model, obtaining an abnormal degree judgment index according to the data value of the real abnormal data, and detecting the abnormal condition of the energy storage power station according to the abnormal degree judgment index.
According to the S2 and the S3, the distribution deviation index, the local data analysis index and the data change index of each suspected abnormal data can be obtained, an abnormal data detection model is built according to the three indexes, and comprehensive analysis of the abnormal condition of the energy storage data in the energy storage power station is achieved. And identifying abnormal data in the energy storage data through the abnormal data detection model, screening out real abnormal data, and judging the abnormal state of the energy storage power station through the real abnormal data.
The construction of the abnormal data detection model specifically comprises the following steps: and comparing the product of the distribution deviation index of the suspected abnormal data and the local data distribution index with the data change index to obtain a first ratio, and normalizing the first ratio to obtain the abnormal data detection model. In the embodiment of the invention, for the accuracy of the subsequent analysis, the suspected abnormal data is used
The specific way for constructing the abnormal data detection model is as follows:
in the formula (I), the compound is shown in the specification,
indicates as suspected abnormal data->
Is greater than or equal to>
Indicates as suspected abnormal data->
Deviates from the criterion, is>
Indicates as suspected abnormal data->
Is based on the local data distribution indicator of->
Indicates as suspected abnormal data->
Based on the data change indicator of (4), is greater than or equal to>
Expressed as a constant coefficient, is present>
Expressed as a normalization function. The purpose of the constant coefficients in the formula is to avoid the meaningless case of the formula in which the denominator is zero, in embodiments of the present invention, when the formula is not significant>
The constant coefficient is set to 0.01, and it should be noted that the normalization function may select various normalization methods such as linear normalization or zero-mean normalization, which is not limited herein.
Comprehensively analyzing all indexes of the suspected abnormal data in a product-sum ratio mode, wherein when the distribution deviation index of the suspected abnormal data is larger, the larger the distribution difference of the data value distribution of the suspected abnormal data to the non-suspected abnormal data is, the more probable the suspected abnormal data is to be real abnormal data; when the local data distribution index of the suspected abnormal data is larger, the more disordered the distribution of the local data of the suspected abnormal data is, the more probable the suspected abnormal data is to be real abnormal data; therefore, the larger the distribution deviation index and the local data distribution index, the larger the first ratio, which indicates that the suspected abnormal data is more likely to be real abnormal data, and both the distribution deviation index and the local data distribution index have positive correlation with the first ratio. The larger the data change index of the suspected abnormal data is, the larger the data value change degree of the suspected abnormal data in each direction is, the more likely the suspected abnormal data is to be noise data, so that the smaller the data change index is, the larger the first ratio is, the more likely the corresponding suspected abnormal data is to be real abnormal data, and the data change index and the first ratio are in a negative correlation relationship.
And normalizing the obtained first ratio to complete the construction of an abnormal data detection model, wherein an abnormal detection index of each suspected abnormal data can be obtained according to the abnormal data detection model, in the embodiment of the invention, an abnormal threshold value is set to be 0.5, and when the abnormal detection index is greater than the abnormal threshold value, the corresponding suspected abnormal data is considered as real abnormal data, and all real abnormal data are screened out.
Through the abnormal data detection model, accomplish the extraction to real abnormal data, obtain real abnormal data set, further can judge the abnormal situation of energy storage power station according to real abnormal data, obtain abnormal degree judgement index according to the data value of real abnormal data, judge the abnormal situation of index detection energy storage power station through abnormal degree, specifically include:
and acquiring an accumulated value of data values of all real abnormal data, wherein the accumulated value can reflect the abnormal degree of the energy storage data, when the accumulated value is larger, the abnormal degree of the energy storage data in the period of time is large, and the abnormal condition is more serious.
In the embodiment of the invention, the preset early warning threshold value is 0.4, and when the abnormal degree judgment index is less than or equal to the early warning threshold value, the abnormal degree of the energy storage data in the energy storage power station in the period of time is low, and early warning prompt is not performed temporarily; when the abnormal degree judgment index is larger than the early warning threshold value, the abnormal degree of the energy storage data in the energy storage power station in the period of time is higher, and at the moment, early warning prompt is needed to prompt relevant management personnel to check and maintain relevant problems as soon as possible, so that serious accidents are avoided.
In summary, the invention obtains the time series data of the energy storage parameters in the preset time period through the sensor, constructs the key energy storage data matrix according to the time series data, obtains the suspected abnormal data according to the fluctuation degree of each energy storage data in each type of energy storage parameters in the key energy storage data matrix, obtains the distribution deviation index according to the distribution deviation degree of the suspected abnormal data, obtains the local data distribution index according to the uniform distribution condition of the local data around the suspected abnormal data, obtains the data change index according to the data value change condition of the local data of the suspected abnormal data, screens out the real abnormal data according to the distribution deviation index, the local data distribution index and the data change index, obtains the abnormal detection index according to the data value of the real abnormal data, and judges the abnormal condition of the energy storage power station through the abnormal detection index.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. 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.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments.
Claims (10)
1. An abnormality detection method for energy storage data of an energy storage power station is characterized by comprising the following steps:
acquiring time sequence data of at least two types of energy storage parameters in a preset time period through a sensor, and constructing a key energy storage data matrix according to the time sequence data; screening suspected abnormal data in the key energy storage data matrix according to the data fluctuation degree of each energy storage data in each type of energy storage parameters in the key energy storage data matrix;
acquiring distribution characteristic difference between actual distribution characteristics corresponding to the suspected abnormal data and standard distribution characteristics corresponding to the non-suspected abnormal data in the key energy storage data matrix, and acquiring distribution deviation indexes of each suspected abnormal data according to the distribution characteristic difference;
in the key energy storage data matrix, obtaining a local data distribution index of each suspected abnormal data according to the local distribution characteristics of the data values in different preset directions in a preset local window corresponding to each suspected abnormal data; acquiring a data change matrix according to the data value difference condition of the energy storage data in the preset neighborhood range corresponding to the suspected abnormal data, and acquiring a data change index of the suspected abnormal data according to the data value change matrix;
obtaining an abnormal data detection model according to the distribution deviation index, the local data distribution index and the data change index of the suspected abnormal data; and screening out real abnormal data in the suspected abnormal data according to the abnormal data detection model, obtaining an abnormal degree judgment index according to the data value of the real abnormal data, and detecting the abnormal condition of the energy storage power station according to the abnormal degree judgment index.
2. The method for detecting the abnormality of the energy storage data of the energy storage power station as claimed in claim 1, wherein the step of screening out the suspected abnormal data in the key energy storage data matrix according to the data fluctuation degree of each type of energy storage data in each type of energy storage parameter in the key energy storage data matrix comprises:
taking the absolute value of the difference between each energy storage data in the key energy storage data matrix and other energy storage data in the same type of energy storage parameters as a data difference, and taking the mean value of the data difference as a data fluctuation degree index of the energy storage data;
acquiring data fluctuation degree indexes of all energy storage data; when the data fluctuation degree index is larger than a preset fluctuation threshold value, taking the corresponding energy storage data as suspected abnormal data; and when the data fluctuation degree index is less than or equal to a preset fluctuation threshold value, taking the corresponding energy storage data as non-suspected abnormal data.
3. The method for detecting the abnormality of the energy storage data of the energy storage power station as claimed in claim 1, wherein the obtaining of the standard distribution characteristics and the actual distribution characteristics of the suspected abnormal data comprises:
fitting non-suspected abnormal data in the key energy storage data by adopting a Gaussian model to obtain a standard Gaussian model, and taking a standard Gaussian model value corresponding to each energy storage data as a standard distribution characteristic of an energy storage parameter;
and fitting suspected abnormal data in the key energy storage data by adopting a Gaussian model to obtain an abnormal data Gaussian model, and taking an abnormal data Gaussian model value corresponding to each suspected abnormal data as the actual distribution characteristic of the suspected abnormal data.
4. The method for detecting the abnormality of the energy storage data of the energy storage power station as claimed in claim 3, wherein the obtaining of the distribution deviation index comprises:
taking the difference absolute value of the abnormal data Gaussian model value corresponding to the actual distribution characteristic of each suspected abnormal data and the standard Gaussian model value corresponding to the standard distribution characteristic as the distribution characteristic difference; and carrying out normalization processing on each distribution characteristic difference to obtain a distribution deviation index of each suspected abnormal data.
5. The method for detecting the abnormality of the energy storage data of the energy storage power station as claimed in claim 1, wherein the obtaining of the local distribution characteristics comprises:
the local distribution characteristics comprise data value distribution characteristic values and category distribution characteristic values;
dividing energy storage data with the same data value in a preset local window corresponding to each suspected abnormal data into the same data value type to obtain all data value types; taking the combination of data values corresponding to any two data value types in the data value types as a type binary group to obtain all types of binary groups;
traversing in the horizontal direction of a preset local window corresponding to suspected abnormal data by taking the size of each kind of binary group as the size of a sliding window, and counting the frequency which is the same as the data value in each kind of binary group as the horizontal frequency; similarly, the vertical frequency of traversal of each kind of binary group in the vertical direction is obtained; counting the frequency of all kinds of binary groups in a preset local window of the suspected abnormal data as a total frequency;
obtaining a horizontal entropy value according to the ratio of the horizontal frequency of each kind of binary group to the total frequency; obtaining a vertical entropy value according to the proportion of the vertical frequency of each kind of binary group in the total frequency; adding the horizontal entropy value and the vertical entropy value to obtain a category distribution characteristic value;
and taking the absolute value of the difference between the mean value of the data values of all the energy storage data in the preset local window and the mean value of the data values of the non-suspected abnormal data as a data value distribution characteristic value.
6. The method for detecting the abnormality of the energy storage data of the energy storage power station as claimed in claim 5, wherein the obtaining of the local data distribution index comprises:
and taking the product of the data value distribution characteristic value and the class distribution characteristic value as a local data distribution index.
7. The method for detecting the abnormality of the energy storage data of the energy storage power station as claimed in claim 1, wherein the obtaining the data change index of the suspected abnormal data according to the data value change matrix comprises:
calculating the sum of the difference values of the suspected abnormal data and each energy storage data in each preset direction in the corresponding preset neighborhood range to obtain the integral data value difference in the corresponding preset direction; taking a square matrix constructed by arranging the integral data value differences in all preset directions as a data value change matrix;
and obtaining the eigenvalue of the data value change matrix, multiplying the absolute values of all the eigenvalues, and taking the obtained eigenvalue product as the data change index.
8. The method for detecting the abnormality of the energy storage data of the energy storage power station as claimed in claim 1, wherein the obtaining of the abnormal data detection model comprises:
comparing the product of the distribution deviation index of the suspected abnormal data and the local data distribution index with the data change index to obtain a first ratio; and carrying out normalization processing on the first ratio to obtain an abnormal data detection model.
9. The method for detecting the abnormality of the energy storage data of the energy storage power station as claimed in claim 1, wherein the obtaining of the abnormality degree judgment index comprises:
and acquiring the accumulated values of the data values of all the real abnormal data, and carrying out normalization processing on the accumulated values to acquire an abnormal degree judgment index.
10. The method for detecting the abnormality of the energy storage data of the energy storage power station as claimed in claim 1, wherein the obtaining of the key energy storage data matrix comprises:
normalizing the energy storage data in each type of energy storage parameter according to the data values in the same type of energy storage parameters to obtain normalized time sequence data; taking the time information and the energy storage parameters in all the normalized time sequence data as rows and columns of the matrix to obtain an energy storage data matrix;
and reducing the dimension of the energy storage data matrix by adopting a principal component analysis method, and obtaining a matrix reconstructed by the time information and the key energy storage parameter data as a key energy storage data matrix.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310139290.7A CN115858630B (en) | 2023-02-21 | 2023-02-21 | Abnormality detection method for energy storage data of energy storage power station |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310139290.7A CN115858630B (en) | 2023-02-21 | 2023-02-21 | Abnormality detection method for energy storage data of energy storage power station |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115858630A true CN115858630A (en) | 2023-03-28 |
| CN115858630B CN115858630B (en) | 2023-05-09 |
Family
ID=85658500
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310139290.7A Active CN115858630B (en) | 2023-02-21 | 2023-02-21 | Abnormality detection method for energy storage data of energy storage power station |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115858630B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115600119A (en) * | 2022-12-13 | 2023-01-13 | 青岛左岸数据科技有限公司（Cn） | Data processing method and system suitable for wind power generation |
| CN116087759A (en) * | 2023-04-12 | 2023-05-09 | 广东翰唐智控有限公司 | Method for inspecting conductive path of circuit board and circuit system |
| CN116230193A (en) * | 2023-05-11 | 2023-06-06 | 聊城市第二人民医院 | Intelligent hospital file management method and system |
| CN116340795A (en) * | 2023-05-29 | 2023-06-27 | 山东一然环保科技有限公司 | Operation data processing method for pure oxygen combustion heating furnace |
| CN116610682A (en) * | 2023-07-21 | 2023-08-18 | 山东盈动智能科技有限公司 | Temperature tester data classification method based on data storage |
| CN116930595A (en) * | 2023-09-18 | 2023-10-24 | 法拉迪电气有限公司 | Accurate data metering method for new energy grid-connected voltage regulation |
| CN116992391A (en) * | 2023-09-27 | 2023-11-03 | 青岛冠宝林活性炭有限公司 | Hard carbon process environment-friendly monitoring data acquisition and processing method |
| CN117113107A (en) * | 2023-10-24 | 2023-11-24 | 广东顺德爱顺机电科技有限公司 | Rotor die casting abnormal error data processing method |
| CN117195008A (en) * | 2023-11-08 | 2023-12-08 | 山东神光航天科技有限公司 | Abnormal data processing method for air quality monitoring |
| CN117421689A (en) * | 2023-12-18 | 2024-01-19 | 杭州湘亭科技有限公司 | Uranium radioactivity pollution measurement transmission system based on pipeline robot |
| CN117874445A (en) * | 2024-03-11 | 2024-04-12 | 北京盛拓达生物技术有限公司 | Enzyme preparation production monitoring method for real-time online monitoring data analysis |
| CN117907845A (en) * | 2024-03-20 | 2024-04-19 | 山东泰开电力电子有限公司 | Electrochemical energy storage system insulation detection method based on electrical parameter analysis |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200211325A1 (en) * | 2018-12-31 | 2020-07-02 | Playtika Ltd. | System and method for outlier detection in gaming |
| US20220058174A1 (en) * | 2020-08-24 | 2022-02-24 | Microsoft Technology Licensing, Llc | System and method for removing exception periods from time series data |
| CN115082744A (en) * | 2022-08-22 | 2022-09-20 | 启东市罗源光伏设备有限公司 | Solar energy heat collection efficiency analysis method and system based on artificial intelligence |
| CN115392408A (en) * | 2022-10-31 | 2022-11-25 | 山东鲁抗医药集团赛特有限责任公司 | Method and system for detecting abnormal operation of electronic particle counter |
| CN115600837A (en) * | 2022-11-02 | 2023-01-13 | 国网山东省电力公司营销服务中心（计量中心）（Cn） | Energy-saving diagnosis method for load abnormity of electric appliance |
-
2023
- 2023-02-21 CN CN202310139290.7A patent/CN115858630B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200211325A1 (en) * | 2018-12-31 | 2020-07-02 | Playtika Ltd. | System and method for outlier detection in gaming |
| US20220058174A1 (en) * | 2020-08-24 | 2022-02-24 | Microsoft Technology Licensing, Llc | System and method for removing exception periods from time series data |
| CN115082744A (en) * | 2022-08-22 | 2022-09-20 | 启东市罗源光伏设备有限公司 | Solar energy heat collection efficiency analysis method and system based on artificial intelligence |
| CN115392408A (en) * | 2022-10-31 | 2022-11-25 | 山东鲁抗医药集团赛特有限责任公司 | Method and system for detecting abnormal operation of electronic particle counter |
| CN115600837A (en) * | 2022-11-02 | 2023-01-13 | 国网山东省电力公司营销服务中心（计量中心）（Cn） | Energy-saving diagnosis method for load abnormity of electric appliance |
Non-Patent Citations (1)
| Title |
|---|
| 董天文;叶勇超;俞周瑜;杨嘉欣;王松静;: "ATM交易状态特征分析与异常检测", 数学建模及其应用 * |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115600119B (en) * | 2022-12-13 | 2023-06-16 | 青岛左岸数据科技有限公司 | Data processing method and system suitable for wind power generation |
| CN115600119A (en) * | 2022-12-13 | 2023-01-13 | 青岛左岸数据科技有限公司（Cn） | Data processing method and system suitable for wind power generation |
| CN116087759A (en) * | 2023-04-12 | 2023-05-09 | 广东翰唐智控有限公司 | Method for inspecting conductive path of circuit board and circuit system |
| CN116087759B (en) * | 2023-04-12 | 2023-05-30 | 广东翰唐智控有限公司 | Method for inspecting conductive path of circuit board and circuit system |
| CN116230193A (en) * | 2023-05-11 | 2023-06-06 | 聊城市第二人民医院 | Intelligent hospital file management method and system |
| CN116230193B (en) * | 2023-05-11 | 2023-07-21 | 聊城市第二人民医院 | Intelligent hospital file management method and system |
| CN116340795A (en) * | 2023-05-29 | 2023-06-27 | 山东一然环保科技有限公司 | Operation data processing method for pure oxygen combustion heating furnace |
| CN116610682A (en) * | 2023-07-21 | 2023-08-18 | 山东盈动智能科技有限公司 | Temperature tester data classification method based on data storage |
| CN116610682B (en) * | 2023-07-21 | 2023-09-29 | 山东盈动智能科技有限公司 | Temperature tester data classification method based on data storage |
| CN116930595B (en) * | 2023-09-18 | 2023-11-24 | 法拉迪电气有限公司 | Accurate data metering method for new energy grid-connected voltage regulation |
| CN116930595A (en) * | 2023-09-18 | 2023-10-24 | 法拉迪电气有限公司 | Accurate data metering method for new energy grid-connected voltage regulation |
| CN116992391A (en) * | 2023-09-27 | 2023-11-03 | 青岛冠宝林活性炭有限公司 | Hard carbon process environment-friendly monitoring data acquisition and processing method |
| CN116992391B (en) * | 2023-09-27 | 2023-12-15 | 青岛冠宝林活性炭有限公司 | Hard carbon process environment-friendly monitoring data acquisition and processing method |
| CN117113107A (en) * | 2023-10-24 | 2023-11-24 | 广东顺德爱顺机电科技有限公司 | Rotor die casting abnormal error data processing method |
| CN117113107B (en) * | 2023-10-24 | 2023-12-29 | 广东顺德爱顺机电科技有限公司 | Rotor die casting abnormal error data processing method |
| CN117195008A (en) * | 2023-11-08 | 2023-12-08 | 山东神光航天科技有限公司 | Abnormal data processing method for air quality monitoring |
| CN117195008B (en) * | 2023-11-08 | 2024-03-12 | 山东神光航天科技有限公司 | Abnormal data processing method for air quality monitoring |
| CN117421689A (en) * | 2023-12-18 | 2024-01-19 | 杭州湘亭科技有限公司 | Uranium radioactivity pollution measurement transmission system based on pipeline robot |
| CN117421689B (en) * | 2023-12-18 | 2024-03-12 | 杭州湘亭科技有限公司 | Uranium radioactivity pollution measurement transmission system based on pipeline robot |
| CN117874445A (en) * | 2024-03-11 | 2024-04-12 | 北京盛拓达生物技术有限公司 | Enzyme preparation production monitoring method for real-time online monitoring data analysis |
| CN117907845A (en) * | 2024-03-20 | 2024-04-19 | 山东泰开电力电子有限公司 | Electrochemical energy storage system insulation detection method based on electrical parameter analysis |
| CN117907845B (en) * | 2024-03-20 | 2024-05-17 | 山东泰开电力电子有限公司 | Electrochemical energy storage system insulation detection method based on electrical parameter analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115858630B (en) | 2023-05-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115858630A (en) | Abnormity detection method for energy storage data of energy storage power station | |
| CN107742127B (en) | Improved electricity stealing prevention intelligent early warning system and method | |
| CN112731159B (en) | Method for pre-judging and positioning battery faults of battery compartment of energy storage power station | |
| CN112505549B (en) | New energy automobile battery abnormity detection method based on isolated forest algorithm | |
| CN109583680B (en) | Power stealing identification method based on support vector machine | |
| CN110222991B (en) | Metering device fault diagnosis method based on RF-GBDT | |
| CN109165242B (en) | Fault diagnosis and early warning method based on entropy sorting and space-time analysis | |
| CN115409131B (en) | Production line abnormity detection method based on SPC process control system | |
| CN113791350B (en) | Battery fault prediction method | |
| CN113962259A (en) | Multi-mode double-layer fault diagnosis method for fuel cell system | |
| CN114559819A (en) | Electric vehicle battery safety early warning method based on signal processing | |
| CN109947815B (en) | Power theft identification method based on outlier algorithm | |
| CN110580492A (en) | Track circuit fault precursor discovery method based on small fluctuation detection | |
| CN112327192A (en) | Battery capacity diving phenomenon identification method based on curve form | |
| CN112257784A (en) | Electricity stealing detection method based on gradient boosting decision tree | |
| CN117148045A (en) | Fault studying and judging management system for running state of power distribution network | |
| CN113533995B (en) | Consistency detection method for power battery | |
| CN115219903A (en) | Battery self-discharge rate abnormity judgment method and device based on Internet of vehicles data analysis | |
| CN110852906A (en) | Method and system for identifying electricity stealing suspicion based on high-dimensional random matrix | |
| CN114325433A (en) | Lithium ion battery fault detection method and system based on electrochemical impedance spectrum test | |
| CN112036087A (en) | Multi-strategy fused nuclear power key equipment fault diagnosis method and system | |
| CN116977807A (en) | Multi-sensor fusion-based intelligent monitoring system and method for refrigerator | |
| CN113581015B (en) | Safety early warning method and device for fuel cell system | |
| Arbon et al. | Anomaly detection in satellite communications networks using support vector machines | |
| CN115511106B (en) | Method, device and readable storage medium for generating training data based on time sequence data |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |