CN-121980436-A - Data analysis device and data analysis method

CN121980436ACN 121980436 ACN121980436 ACN 121980436ACN-121980436-A

Abstract

The disclosure relates to a data analysis device and a data analysis method, wherein the data analysis device comprises the steps of detecting time sequence data in a target time interval based on an abnormal point detection algorithm to obtain an abnormal point information sequence, traversing identifications in each information set in the abnormal point information sequence based on a stack structure and target logic, determining an abnormal interval in the target time interval, discarding a current identification when the identification at the top of the stack indicates an ascending abnormal point in the stack structure and the current identification indicates the ascending abnormal point, pushing the current identification into the stack when the identification at the top of the stack indicates the ascending abnormal point in the stack structure and the current identification indicates a descending abnormal point in the stack structure, and pushing the identification at the top of the stack and discarding the current identification when the identification at the top of the stack indicates the descending abnormal point in the stack structure and the current identification indicates the descending abnormal point.

Inventors

XIN QUANQI
ZHENG XIANG
ZHANG JIKUAN
Ji Songxi
LIU XIANPAN

Assignees

青岛聚看云科技有限公司

Dates

Publication Date: 20260505
Application Date: 20251208

Claims (10)

1. A data analysis apparatus, comprising: a controller configured to: detecting time sequence data in a target time interval based on an abnormal point detection algorithm to obtain an abnormal point information sequence, wherein the abnormal point information sequence comprises at least one information set, one information set comprises an abnormal point identifier and attribute information, and the attribute information is used for indicating that the corresponding abnormal point is an ascending abnormal point or a descending abnormal point in the time sequence data; traversing the mark in each information set in the abnormal point information sequence based on a stack structure and target logic, and determining an abnormal interval in the target time interval; wherein the target logic comprises: Discarding the current identifier under the condition that the identifier at the top of the stack in the stack structure indicates an ascending abnormal point and the current identifier indicates the ascending abnormal point; under the condition that an identifier positioned at the top of a stack in the stack structure indicates an ascending abnormal point and a current identifier indicates a descending abnormal point, the current identifier is pushed into the stack; Under the condition that the mark at the top of the stack indicates a descending abnormal point in the stack structure and the current mark indicates the descending abnormal point, the mark at the top of the stack is popped off and discarded, and the current mark is pushed into the stack; Under the condition that the mark at the top of the stack in the stack structure indicates a descending abnormal point and the current mark indicates an ascending abnormal point, the current mark is pushed into the stack; the starting time of the abnormal section is the time corresponding to the ascending abnormal point indicated by the first mark, the ending time of the abnormal section is the time corresponding to the descending abnormal point indicated by the second mark, and the first mark and the second mark are two adjacent marks in the stack structure.
2. The data analysis device of claim 1, wherein the controller is specifically configured to: traversing the identifiers in each information set in the outlier information sequence, and determining whether the stack structure is empty or not under the condition that the current identifier indicates an ascending outlier; Under the condition that the stack structure is empty, pushing the current mark into a stack; Determining whether an identifier at the top of a stack in the stack structure indicates an ascending abnormal point or not under the condition that the stack structure is not empty; discarding the current identifier under the condition that the identifier positioned at the top of the stack indicates an abnormal point rising; Under the condition that the mark at the top of the stack indicates an abnormal point of descent, the existing mark in the stack structure is popped off; Determining an abnormal interval based on the time corresponding to the existing identifier; And pushing the current identifier to a stack.
3. The data analysis device of claim 2, wherein the controller is specifically configured to: Determining whether two marks exist in the stack structure under the condition that marks at the top of the stack indicate abnormal points of descent; Under the condition that two identifiers exist in the stack structure, the two identifiers are popped off; Determining the abnormal interval based on the time corresponding to the two identifiers respectively; Under the condition that one identifier exists in the stack structure, the one identifier is popped off the stack; and determining the abnormal interval based on the time corresponding to the one identifier and the starting time of the target time interval.
4. The data analysis device of claim 1, wherein the controller is specifically configured to: traversing the identifiers in each information set in the outlier information sequence, and determining whether the stack structure is empty or not under the condition that the current identifier indicates a descending outlier; Under the condition that the stack structure is empty, pushing the current mark into a stack; Determining whether an identifier at the top of a stack in the stack structure indicates an ascending abnormal point or not under the condition that the stack structure is not empty; under the condition that the mark at the top of the stack indicates an ascending abnormal point, pushing the current mark into the stack; under the condition that the mark at the top of the stack indicates an abnormal point of descent, the mark at the top of the stack is popped off and discarded; And pushing the current identifier to a stack.
5. The data analysis device of claim 1, wherein the controller is specifically configured to: after traversing the identifiers in each information set in the outlier information sequence, determining whether two identifiers exist in the stack structure; Under the condition that two identifiers exist in the stack structure, the two identifiers are popped off; Determining the abnormal interval based on the time corresponding to the two identifiers respectively; Under the condition that one identifier exists in the stack structure, the one identifier is popped off the stack; determining an abnormal section based on the time corresponding to the one mark and the ending time of the target time section under the condition that the one mark indicates the abnormal point of rising; and determining an abnormal section based on the time corresponding to the one identifier and the starting time of the target time section under the condition that the one identifier indicates the abnormal point of descent.
6. The data analysis device of claim 1, wherein the controller is specifically configured to: After traversing the identification in each information set in the abnormal point information sequence, determining that the target time interval is an abnormal interval under the condition that the identification in any information set in the abnormal point information sequence indicates a constant abnormal point.
7. The data analysis device of any one of claims 1-6, wherein the controller is specifically configured to: When traversing the mark in each information set in the abnormal point information sequence and determining a plurality of abnormal intervals meeting a target condition, correcting the plurality of abnormal intervals into target abnormal intervals, wherein the starting time of the target abnormal intervals is the minimum value in the starting time of the plurality of abnormal intervals, and the ending time of the target abnormal intervals is the maximum value in the ending time of the plurality of abnormal intervals; wherein the target condition is: and arranging the plurality of abnormal intervals according to a time sequence, wherein the difference value between the larger starting time and the smaller ending time in the starting time and the ending time of any two adjacent abnormal intervals is smaller than or equal to a difference value threshold.
8. A method of data analysis, comprising: detecting time sequence data in a target time interval based on an abnormal point detection algorithm to obtain an abnormal point information sequence, wherein the abnormal point information sequence comprises at least one information set, one information set comprises an abnormal point identifier and attribute information, and the attribute information is used for indicating that the corresponding abnormal point is an ascending abnormal point or a descending abnormal point in the time sequence data; traversing the mark in each information set in the abnormal point information sequence based on a stack structure and target logic, and determining an abnormal interval in the target time interval; wherein the target logic comprises: Discarding the current identifier under the condition that the identifier at the top of the stack in the stack structure indicates an ascending abnormal point and the current identifier indicates the ascending abnormal point; under the condition that an identifier positioned at the top of a stack in the stack structure indicates an ascending abnormal point and a current identifier indicates a descending abnormal point, the current identifier is pushed into the stack; Under the condition that the mark at the top of the stack indicates a descending abnormal point in the stack structure and the current mark indicates the descending abnormal point, the mark at the top of the stack is popped off and discarded, and the current mark is pushed into the stack; Under the condition that the mark at the top of the stack in the stack structure indicates a descending abnormal point and the current mark indicates an ascending abnormal point, the current mark is pushed into the stack; the starting time of the abnormal section is the time corresponding to the ascending abnormal point indicated by the first mark, the ending time of the abnormal section is the time corresponding to the descending abnormal point indicated by the second mark, and the first mark and the second mark are two adjacent marks in the stack structure.
9. The method of claim 8, wherein traversing the identity in each information set in the sequence of outlier information based on stack structure, target logic, determines an outlier interval within the target time interval comprises: traversing the identifiers in each information set in the outlier information sequence, and determining whether the stack structure is empty or not under the condition that the current identifier indicates an ascending outlier; Under the condition that the stack structure is empty, pushing the current mark into a stack; Determining whether an identifier at the top of a stack in the stack structure indicates an ascending abnormal point or not under the condition that the stack structure is not empty; discarding the current identifier under the condition that the identifier positioned at the top of the stack indicates an abnormal point rising; Under the condition that the mark at the top of the stack indicates an abnormal point of descent, the existing mark in the stack structure is popped off; Determining an abnormal interval based on the time corresponding to the existing identifier; And pushing the current identifier to a stack.
10. The method of claim 8 or 9, wherein the traversing the identity in each information set in the outlier information sequence based on stack structure, target logic, determining an outlier region within the target time region comprises: traversing the identifiers in each information set in the outlier information sequence, and determining whether the stack structure is empty or not under the condition that the current identifier indicates a descending outlier; Under the condition that the stack structure is empty, pushing the current mark into a stack; Determining whether an identifier at the top of a stack in the stack structure indicates an ascending abnormal point or not under the condition that the stack structure is not empty; under the condition that the mark at the top of the stack indicates an ascending abnormal point, pushing the current mark into the stack; under the condition that the mark at the top of the stack indicates an abnormal point of descent, the mark at the top of the stack is popped off and discarded; And pushing the current identifier to a stack.

Description

Data analysis device and data analysis method Technical Field Embodiments of the present disclosure relate to data analysis techniques. And more particularly, to a data analysis apparatus and a data analysis method. Background With the development of the Internet of things, big data and intelligent operation and maintenance technology, time sequence data are widely applied in the fields of industrial monitoring, financial wind control, equipment health diagnosis, system performance analysis and the like. The automatic detection of anomalies in the time sequence data becomes a key link for guaranteeing the stability of the system and improving the operation and maintenance efficiency. The current time series data abnormality detection algorithm is usually focused on identifying a single abnormal point, and can effectively find a mutation point or an outlier, but in the actual time series data analysis and operation and maintenance monitoring scene, the abnormality is always presented as a continuous interval, namely, the system is continuously in an abnormal state for a period of time, so that only returning to the isolated abnormal point can cause fragmentation of abnormal information, and the subsequent abnormal mode identification and decision optimization are not facilitated. Disclosure of Invention In order to solve the above technical problems or at least partially solve the above technical problems, embodiments of the present disclosure provide a data analysis device and a data analysis method. In a first aspect, an embodiment of the present disclosure provides a data analysis apparatus, including a controller configured to detect time-series data in a target time interval based on an outlier detection algorithm, to obtain an outlier information sequence including at least one information set, one information set including an identification of an outlier and attribute information indicating whether the corresponding outlier is an ascending outlier or a descending outlier in the time-series data; the method comprises the steps of traversing an identifier in each information set in an abnormal point information sequence based on a stack structure and target logic, determining an abnormal interval in the target time interval, wherein the target logic comprises discarding the current identifier when the identifier at the top of the stack indicates an ascending abnormal point in the stack structure and the current identifier indicates an ascending abnormal point, pushing the current identifier to a stack when the identifier at the top of the stack indicates an ascending abnormal point in the stack structure and the current identifier indicates a descending abnormal point in the stack structure, pushing the identifier at the top of the stack to the stack when the identifier at the top of the stack indicates a descending abnormal point in the stack structure and the current identifier indicates a descending abnormal point in the stack top and discarding the identifier at the current identifier to the stack, pushing the current identifier to the stack when the identifier at the top of the stack indicates a descending abnormal point in the stack structure and the current identifier indicates a ascending abnormal point in the stack structure, pushing the starting time of the abnormal interval to the first identifier to the second identifier to the first identifier to the second identifier. In the embodiment of the disclosure, on the basis of detecting time sequence data in a target time interval based on an abnormal point detection algorithm to obtain an abnormal point information sequence, based on target logic, traversing marks in each information set in the abnormal point information sequence, controlling the marks to be stacked and to be stacked out, and then determining an abnormal interval in the target time interval based on two adjacent marks in a stack structure. The algorithm not only reserves the sensitivity of point level detection, but also remarkably improves the timeliness and practicality of time sequence anomaly detection through structured interval aggregation, and solves the problem that the related technology cannot directly obtain an anomaly interval. In some embodiments of the present disclosure, the controller is specifically configured to traverse the identifier in each information set in the outlier information sequence, determine if the stack structure is empty if the current identifier indicates an elevated outlier, push the current identifier onto the stack if the stack structure is empty, determine if the identifier at the top of the stack indicates an elevated outlier if the stack structure is not empty, discard the current identifier if the identifier at the top of the stack indicates an elevated outlier, push the existing identifier onto the stack if the identifier at the top of the stack indicates a lowered outlier, determine an anomaly interval based on a time corresponding to