Building air conditioner operation data cleaning method based on envelope analysis

Abstract

The invention provides a building air conditioner operation data cleaning method based on envelope analysis. The cleaning method comprises the following steps: acquiring a time sequence operation data group, wherein each data point of the time sequence operation data group represents operation values at different moments; screening out a plurality of upper extreme points and a plurality of lower extreme points from each data point of the time sequence operation data group, the plurality of upper extreme points forming an upper envelope line array, and the plurality of lower extreme points forming a lower envelope line array; first change slopes of two adjacent upper extreme points in the upper envelope line array are calculated in sequence, and abnormal upper extreme points in the upper envelope line array are determined and eliminated according to the first change slopes; and second change slopes of two adjacent lower extreme points in the lower envelope line array are calculated in sequence, and abnormal lower extreme points in the lower envelope line array are determined and eliminated according to the second change slopes. The method is suitable for the periodically fluctuating air conditioner operation data, and can identify and clean the light outlier data, thereby improving the identification accuracy and effectiveness of the abnormal data.

Description

Technical Field

[0001] This application belongs to the field of air conditioning system data analysis technology, specifically, it relates to a method for cleaning building air conditioning operation data based on envelope analysis. Background Technology

[0002] With the development of intelligent systems, the amount of operational data from building air conditioning systems is increasing significantly. Due to unavoidable interference factors, a large amount of abnormal data appears in the operational data. This abnormal data needs to be cleaned up; otherwise, it will affect the operation analysis and control process of the building air conditioning system. One existing technology uses a slope judgment method to identify abnormal data, that is, data is considered abnormal when the ratio of the change in data value to the change in time value exceeds a certain limit. Another existing technology uses aggregation degree analysis to identify abnormal data, that is, it analyzes the aggregation of data and identifies outliers as abnormal data.

[0003] Because building air conditioning systems are constantly adjusting, their operating data exhibits periodic fluctuations, and drastic changes between adjacent data points are normal. Therefore, the slope judgment method is not suitable for cleaning building air conditioning system operating data. The dispersion of normal operating data for building air conditioning systems is relatively high, so the aggregation degree analysis method is only applicable to severely outlier data and cannot be used for slightly or moderately outlier data. Summary of the Invention

[0004] The technical problem addressed in this application is: how to accurately and effectively identify abnormal data in building air conditioning operation data.

[0005] This application provides a method for cleaning building air conditioning operation data based on envelope analysis, the cleaning method comprising:

[0006] Obtain a time series running data set, wherein each data point in the time series running data set represents the running value at different times;

[0007] Several upper extreme points and several lower extreme points are selected from each data point of the time series running data group. The several upper extreme points form an upper envelope array, and the several lower extreme points form a lower envelope array.

[0008] Calculate the first change slope of two adjacent upper extreme points in the upper envelope array in turn, and determine and remove abnormal upper extreme points in the upper envelope array based on each first change slope;

[0009] The second slope of change of two adjacent lower extreme points in the lower envelope array is calculated sequentially, and abnormal lower extreme points in the lower envelope array are determined and removed based on each second slope of change.

[0010] Optionally, a method for selecting several upper extreme points from the various data points of the time series running data group includes:

[0011] Iterate through each data point in turn. If the running value of the data point at the current time is greater than the running value of the data point at the previous time and greater than the running value of the data point at the next time, then the data point at the current time is taken as the upper extreme point.

[0012] Optionally, a method for selecting several lower extreme points from the various data points of the time series running data group includes:

[0013] Iterate through each data point in turn. If the running value of the data point at the current time is less than the running value of the data point at the adjacent previous time and less than the running value of the data point at the adjacent next time, then the data point at the current time is taken as the next extreme point.

[0014] Optionally, the first slope of change for two adjacent upper extreme points is the ratio of the difference between the running values ​​of the two upper extreme points to the time difference. The method for determining and removing abnormal upper extreme points in the upper envelope array based on each of the first slopes includes:

[0015] Traverse each of the aforementioned upper extreme points and calculate the slope difference between the first slope of the upper extreme point at the current time and the upper extreme point at the adjacent previous time and the second slope of the upper extreme point at the current time and the upper extreme point at the adjacent next time.

[0016] If the slope difference is greater than the first limit, the upper extreme point at the current moment is determined to be an abnormal upper extreme point, and the upper extreme point at the current moment is removed.

[0017] Optionally, the first limit is n times the mean of each first change slope, where 2 ≤ n ≤ 3.

[0018] Optionally, the second slope of two adjacent lower extreme points is the ratio of the difference between the running values ​​of the two lower extreme points to the time difference. The method for determining and removing abnormal lower extreme points in the lower envelope array based on each of the second slopes includes:

[0019] Traverse each of the lower extreme points and calculate the slope difference between the second change slope of the lower extreme point at the current time and the lower extreme point at the adjacent previous time and the second change slope of the lower extreme point at the current time and the lower extreme point at the adjacent next time.

[0020] If the slope difference is less than the second limit, the lower extreme point at the current time is determined to be an abnormal lower extreme point, and the lower extreme point at the current time is removed.

[0021] Optionally, the second limit is n times the mean of the slopes of each second change, where 2 ≤ n ≤ 3.

[0022] The building air conditioning operation data cleaning method based on envelope analysis provided in this application has the following technical effects:

[0023] This method is suitable for air conditioning operation data with periodic fluctuations and can identify and clean slightly outlier data, thereby improving the accuracy and effectiveness of abnormal data identification. Attached Figure Description

[0024] Figure 1 This is a flowchart of a building air conditioning operation data cleaning method based on envelope analysis according to one or more embodiments;

[0025] Figure 2 A schematic diagram of the upper and lower envelopes formed by normal data points according to one or more embodiments;

[0026] Figure 3 This is a schematic diagram of the upper and lower envelopes formed by normal data points and abnormal data points according to one or more embodiments. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0028] Before describing the various embodiments of this application in detail, the technical concept of this application is first briefly described: Currently, when cleaning the operating data of central air conditioning systems, both slope judgment methods and aggregation degree analysis methods are difficult to use effectively and accurately. Therefore, this application provides a building air conditioning operating data cleaning method based on envelope analysis. First, upper and lower extreme points are selected from the data cluster to form upper and lower envelope arrays. Then, the slope of change between two adjacent upper extreme points in the upper and lower envelope arrays is calculated. Abnormal upper or lower extreme points are removed based on each slope. This method is suitable for periodically fluctuating air conditioning operating data and can identify and clean slightly outlier data. The specific principles of the building air conditioning operating data cleaning method based on envelope analysis of this application are described below with reference to more embodiments.

[0029] Specifically, such as Figure 1 As shown in the figure, the building air conditioning operation data cleaning method based on envelope analysis in this embodiment includes the following steps:

[0030] S10. Obtain the time series running data cluster. Each data point in the time series running data cluster represents the running value at different times.

[0031] S20. Select several upper extreme points and several lower extreme points from each data point of the time series running data group. The several upper extreme points form an upper envelope array, and the several lower extreme points form a lower envelope array.

[0032] S30. Calculate the first change slope of two adjacent upper extreme points in the upper envelope array in turn, and determine and remove abnormal upper extreme points in the upper envelope array based on each first change slope.

[0033] S40. Calculate the second slope of change of two adjacent lower extreme points in the lower envelope array in turn, and determine and remove abnormal lower extreme points in the lower envelope array based on each second slope.

[0034] In one or more embodiments, a method for selecting several upper extreme points from various data points in a time series running data set includes: sequentially traversing each data point; if the running value of the data point at the current time moment is greater than the running value of the data point at the adjacent previous time moment and greater than the running value of the data point at the adjacent next time moment, then the data point at the current time moment is taken as an upper extreme point, that is, the running value of the upper extreme point is greater than the running values ​​of its two adjacent data points. As shown in the figure, the upper extreme points are connected to form an upper envelope.

[0035] In one or more embodiments, a method for selecting several lower extreme points from various data points in a time series running data set includes: sequentially traversing each data point; if the running value of the data point at the current time is less than the running value of the data point at the adjacent previous time and less than the running value of the data point at the adjacent next time, then the data point at the current time is taken as the lower extreme point, that is, the running value of the lower extreme point is less than the running values ​​of its two adjacent data points. As shown in the figure, the lower extreme points are connected to form a lower envelope.

[0036] In one or more embodiments, the first slope of change between two adjacent upper extreme points is the ratio of the difference between the running values ​​of the two upper extreme points to the time difference. The method for determining and removing abnormal upper extreme points from the upper envelope array based on each first slope includes: traversing each upper extreme point, calculating the slope difference between the first slope of change between the upper extreme point at the current time and the upper extreme point at the adjacent previous time, and the slope difference between the upper extreme point at the current time and the upper extreme point at the adjacent next time; if the slope difference is greater than a first limit, then the upper extreme point at the current time is determined to be an abnormal upper extreme point, and the upper extreme point at the current time is removed. For example, point P in the figure is an abnormal upper extreme point.

[0037] For example, for ease of description, the slope between the i-th upper extreme point and the (i-1)-th upper extreme point is called the front slope of the i-th upper extreme point, and the slope between the (i+1)-th upper extreme point and the i-th upper extreme point is called the back slope of the i-th upper extreme point. If the difference W between the front slope and the back slope is greater than a first limit, the i-th upper extreme point is determined to be outlier data. For example, the first limit is n times the mean of all first slopes, where 2 ≤ n ≤ 3.

[0038] In one or more embodiments, the second change slope of two adjacent lower extreme points is the ratio of the difference between the running values ​​of the two lower extreme points to the time difference. The method for determining and removing abnormal lower extreme points in the lower envelope array based on each second change slope includes: traversing each lower extreme point, calculating the slope difference between the second change slope of the lower extreme point at the current time and the lower extreme point at the adjacent previous time, and the slope difference between the second change slope of the lower extreme point at the current time and the lower extreme point at the adjacent next time; if the slope difference is less than a second limit, then the lower extreme point at the current time is determined to be an abnormal lower extreme point, and the lower extreme point at the current time is removed.

[0039] For example, for ease of description, the slope between the i-th lower extreme point and the (i-1)-th lower extreme point is called the front slope of the i-th lower extreme point, and the slope between the (i+1)-th lower extreme point and the i-th lower extreme point is called the back slope of the i-th lower extreme point. If the difference W between the front slope and the back slope is less than a second limit, the i-th lower extreme point is determined to be outlier data. For example, the second limit is n times the mean of all second slopes, where 2 ≤ n ≤ 3.

[0040] The specific embodiments of this application have been described in detail above. Although some embodiments have been shown and described, those skilled in the art should understand that modifications and improvements can be made to these embodiments without departing from the principles and spirit of this application as defined by the claims and their equivalents, and such modifications and improvements should also be within the protection scope of this application.

Claims

1. A method for cleaning building air conditioning operation data based on envelope analysis, characterized in that, The cleaning method includes: Obtain a time series running data set, wherein each data point in the time series running data set represents the running value at different times; Several upper extreme points and several lower extreme points are selected from each data point of the time series running data group. The several upper extreme points form an upper envelope array, and the several lower extreme points form a lower envelope array. Calculate the first change slope of two adjacent upper extreme points in the upper envelope array in turn, and determine and remove abnormal upper extreme points in the upper envelope array based on each first change slope; The second slope of change of two adjacent lower extreme points in the lower envelope array is calculated sequentially, and abnormal lower extreme points in the lower envelope array are determined and removed based on each second slope of change.

2. The method for cleaning building air conditioning operation data based on envelope analysis according to claim 1, characterized in that, The method for selecting several upper extreme points from each data point of the time series running data group includes: Iterate through each data point in turn. If the running value of the data point at the current time is greater than the running value of the data point at the previous time and greater than the running value of the data point at the next time, then the data point at the current time is taken as the upper extreme point.

3. The method for cleaning building air conditioning operation data based on envelope analysis according to claim 1, characterized in that, The method for selecting several lower extreme points from each data point of the time series running data group includes: Iterate through each data point in turn. If the running value of the data point at the current time is less than the running value of the data point at the adjacent previous time and less than the running value of the data point at the adjacent next time, then the data point at the current time is taken as the next extreme point.

4. The method for cleaning building air conditioning operation data based on envelope analysis according to claim 1, characterized in that, The first slope of change for two adjacent upper extreme points is the ratio of the difference in the running values ​​of the two upper extreme points to the time difference. The method for determining and removing abnormal upper extreme points in the upper envelope array based on each of the first slopes includes: Traverse each of the aforementioned upper extreme points and calculate the slope difference between the first slope of the upper extreme point at the current time and the upper extreme point at the adjacent previous time and the second slope of the upper extreme point at the current time and the upper extreme point at the adjacent next time. If the slope difference is greater than the first limit, the upper extreme point at the current moment is determined to be an abnormal upper extreme point, and the upper extreme point at the current moment is removed.

5. The method for cleaning building air conditioning operation data based on envelope analysis according to claim 4, characterized in that, The first limit is n times the mean of the slopes of each first change, where 2 ≤ n ≤ 3.

6. The method for cleaning building air conditioning operation data based on envelope analysis according to claim 1, characterized in that, The second slope of two adjacent lower extreme points is the ratio of the difference in the running values ​​of the two lower extreme points to the time difference. The method for determining and removing abnormal lower extreme points in the lower envelope array based on each of the second slopes includes: Traverse each of the lower extreme points and calculate the slope difference between the second change slope of the lower extreme point at the current time and the lower extreme point at the adjacent previous time and the second change slope of the lower extreme point at the current time and the lower extreme point at the adjacent next time. If the slope difference is less than the second limit, the lower extreme point at the current time is determined to be an abnormal lower extreme point, and the lower extreme point at the current time is removed.

7. The method for cleaning building air conditioning operation data based on envelope analysis according to claim 6, characterized in that, The second limit is n times the mean of the slopes of each second change, where 2 ≤ n ≤ 3.

Images