Continuous data discrete point rapid identification and analysis method

Abstract

The invention relates to a rapid identification and analysis method for discrete points of continuous data. The method comprises the following steps: step 1, defining basic information; 2, calibrating the data level of the normal data value in the initial operation period through the data dispersion; 3, identifying, tracking and locking discrete points; the method has the advantages that the requirement for data processing computing power is lowered, and the effectiveness and timeliness of discrete point grabbing are improved.

Description

Technical Field

[0001] The invention belongs to the technical field of abnormal data detection, and in particular relates to a method for rapid identification and analysis of discrete points of continuous data. Background Art

[0002] With the advent of the big data era of the Internet of Everything, the collection, monitoring and analysis of continuously changing data of specific project indicators have gradually become the focus of industrial informatization. The scale and complexity of data are growing continuously. These continuously changing large-scale data contain rich information, but there will also be abnormal data. These abnormal data may represent potential risks, errors, and even bring financial losses, risks to personnel life safety, etc. Therefore, the detection, tracking and locking of abnormal data in continuously changing data are particularly important in the field of big data analysis and processing; effective anomaly detection methods can help enterprises or organizations to promptly discover potential problems and threats, take corresponding measures, and avoid economic losses. At present, the construction of discrete point analysis models for big data analysis is relatively complex, the computing power requirements are relatively high, and the data processing has obvious lags; therefore, it is very necessary to provide a method for rapid identification and analysis of discrete points of continuous data that reduces the computing power requirements for data processing and improves the effectiveness and timeliness of discrete point capture. Summary of the invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a method for rapidly identifying and analyzing discrete points of continuous data, which reduces the requirements for data processing computing power and improves the effectiveness and timeliness of discrete point capture.

[0004] The object of the present invention is achieved by: a method for rapid identification and analysis of discrete points of continuous data, the method comprising the following steps:

[0005] Step 1: Basic information definition;

[0006] Step 2: calibrate the data level of normal data values ​​in the initial operation cycle through data dispersion;

[0007] Step 3: Discrete point identification, tracking and locking.

[0008] The basic information definition in step 1 includes: t,n : represents the data value corresponding to the nth position at time t; ②Cycle: represents the continuously changing data completing an operation cycle; ③mCycle: represents the mth complete operation cycle, m and n are both natural numbers greater than or equal to 1, and the value of m is defined according to the specific application scenario; ④ΔU t,n : represents the relative deviation of the data corresponding to the nth position relative to the adjacent data at time t recorded during the operation; ⑤ΔU T,n:It is the standard starting reference value, which serves as the basic reference for subsequent data fluctuations.

[0009] The relative deviation value ΔU t,n The calculation formula is: ΔU t,n =U t,n-1 +U t,n+1 -U t,n ×2.

[0010] The standard starting reference value ΔU T,n The calculation formula is: ΔU T,n =(ΔU t1,n +ΔU t2,n +...+ΔU tm,n ) / m.

[0011] The data level of the normal data value in the initial operation cycle is calibrated by the data dispersion in step 2 as follows: the data dispersion H of the nth position at time t t,n Defined as: H t,n =(ΔU t,n / ΔU T,n -1)×100%, that is, the dispersion of the data is characterized by comparing the deviation of the relative values.

[0012] The discrete point identification, tracking and locking in step 3 include the following steps:

[0013] Step 3.1: Identify suspicious point data;

[0014] Step 3.2: Tracking of suspicious point data;

[0015] Step 3.3: Lock the data of suspicious points.

[0016] The identification of suspicious point data in step 3.1 is specifically: ΔU t,n The data signs are consistent, that is, the data values ​​are both positive or negative, that is, U t,n *U t,n+1 >0, the data of the corresponding position point n is marked as a suspicious point.

[0017] The data tracking of the suspected point in step 3.2 is specifically as follows: tracking ΔU t,n The point data with consistent symbols and the data tracking at time scale t can clarify the change trend of the nth string of suspicious point data fluctuations over time.

[0018] The suspect point data locking in step 3.3 is specifically as follows: t,n The data value of H locks the suspected data point. t,n The larger the value is, the more discrete the corresponding point data is, which means that the discrete point locking is completed.

[0019] Beneficial effects of the present invention: The present invention is a method for quickly identifying and analyzing discrete points of continuous data. In use, the present invention first defines basic information, and characterizes the discreteness of the data by comparing the deviation of relative values, and then identifies, tracks and locks the discrete points. t,n The absolute value of the data identifies the suspicious cell, ΔU t,n The larger the absolute value, the greater the difference between the corresponding point data and the adjacent point data; finally, through H t,n The data value is used to lock the suspected data point, that is, the discrete battery cell point is locked; the present invention effectively simplifies the big data analysis and processing procedures, reduces the requirements for computer computing power for data processing, and at the same time improves the effectiveness and timeliness of discrete point (abnormal point) capture; the present invention has the advantages of reducing the requirements for data processing computing power and improving the effectiveness and timeliness of discrete point capture. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is the ΔU of the present invention t,n When U is positive t,n Relative position diagram.

[0021] Figure 2 is the nth string of abnormal data ΔU of the present invention t,n Graph of changes over time.

[0022] Figure 3 It is the voltage distribution diagram at different times of the present invention.

[0023] Figure 4 is the different time ΔU of the present invention t,n Change trend chart.

[0024] Figure 5 For different moments H of the present invention t Variation graph. DETAILED DESCRIPTION

[0025] The present invention will be further described below in conjunction with the accompanying drawings.

[0026] Example 1

[0027] like Figure 1-5 As shown, a method for rapid identification and analysis of discrete points of continuous data comprises the following steps:

[0028] Step 1: Basic information definition;

[0029] Step 2: calibrate the data level of normal data values ​​in the initial operation cycle through data dispersion;

[0030] Step 3: Discrete point identification, tracking and locking.

[0031] The present invention is a method for quickly identifying and analyzing discrete points of continuous data. In use, the present invention first defines basic information, and characterizes the discreteness of the data by comparing the deviation of relative values, and then identifies, tracks and locks the discrete points. t,n The absolute value of the data identifies the suspicious cell, ΔU t,n The larger the absolute value, the greater the difference between the corresponding point data and the adjacent point data; finally, through H t,n The data value is used to lock the suspected data point, that is, the discrete battery cell point is locked; the present invention effectively simplifies the big data analysis and processing procedures, reduces the requirements for computer computing power for data processing, and at the same time improves the effectiveness and timeliness of discrete point (abnormal point) capture; the present invention has the advantages of reducing the requirements for data processing computing power and improving the effectiveness and timeliness of discrete point capture.

[0032] Example 2

[0033] like Figure 1-5 As shown, a method for rapid identification and analysis of discrete points of continuous data comprises the following steps:

[0034] Step 1: Basic information definition;

[0035] In this embodiment, the basic information definition includes: ①U t,n : represents the data value corresponding to the nth position at time t; ②Cycle: represents the continuously changing data completing an operation cycle; ③mCycle: represents the mth complete operation cycle, m and n are both natural numbers greater than or equal to 1, and the value of m is defined according to the specific application scenario; ④ΔU t,n : represents the relative deviation of the data corresponding to the nth position at time t recorded during the operation relative to the adjacent two sides of the data, ΔU t,n =U t,n-1 +U t,n+1 -U t,n ×2; ⑤ΔU T,n : is the standard starting reference value, which serves as the basic reference for subsequent data fluctuations, ΔU T,n =(ΔU t1,n +ΔU t2,n +...+ΔU tm,n ) / m.

[0036] Step 2: calibrate the data level of normal data values ​​in the initial operation cycle through data dispersion;

[0037] In this embodiment, specifically: the data discreteness H of the nth position at time t t,n Defined as: H t,n =(ΔU t,n / ΔU T,n-1)×100%, that is, the dispersion of the data is characterized by comparing the deviation of the relative values.

[0038] Step 3: Discrete point identification, tracking and locking.

[0039] In this embodiment, the following steps are included:

[0040] Step 3.1: Identify suspicious point data;

[0041] Specifically: ΔU t,n The data signs are consistent, that is, the data values ​​are both positive or negative, that is, U t,n *U t,n+1 >0, the data of the corresponding position point n is marked as a suspicious point; through ΔU t,n The absolute value of the data identifies the suspicious cell, ΔU t,n The larger the absolute value, the greater the difference between the corresponding point data and the adjacent point data; ΔU t,n The typical form of positive value is Figure 1 shown.

[0042] Step 3.2: Tracking of suspicious point data;

[0043] Specifically: Tracking ΔU t,n The point data with the same symbol, and through the data tracking at the time scale t, the change trend of the n-th string of suspicious point data fluctuations over time can be clearly determined, such as Figure 2 shown.

[0044] Step 3.3: Lock the data of suspicious points.

[0045] Specifically: Through H t,n The data value of H locks the suspected data point. t,n The larger the value is, the more discrete the corresponding point data is, which means that the discrete cell point locking is completed.

[0046] In this embodiment, example analysis: According to the discrete point identification, tracking and locking process, ΔU t,n The data of a new energy battery pack is processed as follows:

[0047] Figure 3 The voltage distribution of different numbers of strings at different times t (due to the large amount of data, only part of the data is shown below); through ΔU t,n The data values ​​used to identify the suspected points are 2#, 4#, 9#, 11#, 15#, 20#, 23#, 25#, 27#, 30#, 32#, 34#, 37#, 38#, 40#, 42#, 44#, 45#, 48#, 53#, 57#, 59#, 63#, 65#, 67#, 69#, 71#, 73#, 75#, 79#, 83#, and 84#.

[0048] Figure 4 By ΔU t,n The data value of gradually increases over time. 4# is a normal value (for comparison). Through time scale tracking, ΔU t,n The points that became larger are 9#, 11#, 20#, 30#, 32#, 34#, and 37#, and the number of suspicious data points decreased.

[0049] Figure 5 To pass H t,n As the data values ​​change over time, the discreteness of voltage data differs significantly, and the locked abnormal points are 20# and 37#; the problem points can be locked according to the tracking data and handled in time.

[0050] The present invention is a method for quickly identifying and analyzing discrete points of continuous data. In use, the present invention first defines basic information, and characterizes the discreteness of the data by comparing the deviation of relative values, and then identifies, tracks and locks the discrete points. t,n The absolute value of the data identifies the suspicious cell, ΔU t,n The larger the absolute value, the greater the difference between the corresponding point data and the adjacent point data; finally, through H t,n The data value is used to lock the suspected data point, that is, the discrete battery cell point is locked; the present invention effectively simplifies the big data analysis and processing procedures, reduces the requirements for computer computing power for data processing, and at the same time improves the effectiveness and timeliness of discrete point (abnormal point) capture; the present invention has the advantages of reducing the requirements for data processing computing power and improving the effectiveness and timeliness of discrete point capture.

Claims

1. A method for rapid identification and analysis of discrete points of continuous data, characterized by: The method comprises the following steps: Step 1: Basic information definition; Step 2: calibrate the data level of normal data values ​​in the initial operation cycle through data dispersion; Step 3: Discrete point identification, tracking and locking.

2. A method for rapid identification and analysis of discrete points of continuous data as claimed in claim 1, characterized in that: The basic information definition in step 1 includes: t,n : represents the data value corresponding to the nth position at time t; ②Cycle: represents the continuously changing data completing an operation cycle; ③mCycle: represents the mth complete operation cycle, m and n are both natural numbers greater than or equal to 1, and the value of m is defined according to the specific application scenario; ④ΔU t,n : represents the relative deviation of the data corresponding to the nth position relative to the adjacent data at time t recorded during the operation; ⑤ΔU T,n :It is the standard starting reference value, which serves as the basic reference for subsequent data fluctuations.

3. A method for rapid identification and analysis of discrete points of continuous data as claimed in claim 2, characterized in that: The relative deviation value ΔU t,n The calculation formula is: ΔU t,n =U t,n-1 +U t,n+1 -U t,n ×2.

4. A method for rapid identification and analysis of discrete points of continuous data as claimed in claim 3, characterized in that: The standard starting reference value ΔU T,n The calculation formula is: ΔU T,n =(ΔU t1,n +ΔU t2,n +...+ΔU tm,n ) / m.

5. The method for rapid identification and analysis of discrete points of continuous data as claimed in claim 1, characterized in that: The data level of the normal data value in the initial operation cycle is calibrated by the data dispersion in step 2 as follows: the data dispersion H of the nth position at time t t,n Defined as: H t,n =(ΔU t,n / ΔU T,n -1)×100%, that is, the dispersion of the data is characterized by comparing the deviation of the relative values.

6. A method for rapid identification and analysis of discrete points of continuous data as claimed in claim 1, characterized in that: The discrete point identification, tracking and locking in step 3 include the following steps: Step 3.1: Identify suspicious point data; Step 3.2: Tracking of suspicious point data; Step 3.3: Lock the data of suspicious points.

7. A method for rapid identification and analysis of discrete points of continuous data as claimed in claim 6, characterized in that: The identification of suspicious point data in step 3.1 is specifically: ΔU t,n The data signs are consistent, that is, the data values ​​are both positive or negative, that is, U t,n *U t,n+1 >0, the data of the corresponding position point n is marked as a suspicious point.

8. A method for rapid identification and analysis of discrete points of continuous data as claimed in claim 7, characterized in that: The data tracking of the suspected point in step 3.2 is specifically as follows: tracking ΔU t,n The point data with consistent symbols and the data tracking at time scale t can clarify the change trend of the nth string of suspicious point data fluctuations over time.

9. A method for rapid identification and analysis of discrete points of continuous data as claimed in claim 8, characterized in that: The suspect point data locking in step 3.3 is specifically as follows: t,n The data value of H locks the suspected data point. t,n The larger the value is, the more discrete the corresponding point data is, which means that the discrete point locking is completed.

Images