Real-time abnormity-diagnosing and interpolation method for water regime-monitoring data

Patent metadata

Field	Value
Publication number	US-12158341-B2
Application number	US-202117625639-A
Country	US
Kind code	B2
Filing date	Mar 10, 2021
Priority date	Apr 2, 2020
Publication date	Dec 3, 2024
Grant date	Dec 3, 2024

Abstract

A real-time abnormity-diagnosing and interpolation method for water regime-monitoring data relates to the technical field of monitoring water regime. This method includes the following steps: acquiring water regime-monitoring data, drawing a box plot, recognizing and diagnosing abnormal data in real time based on the box plot, performing grey correlation analysis on other variables related to a predictor variable, building a BP neural network model and making training, applying the BP neural network model to predict water regime-monitoring data in real time, and performing abnormity diagnosis and data interpolation. Adopting this method, we can effectively enhance predicting and monitoring the water regime-monitoring data in real time, and diagnose abnormal data and make interpolations in time, thereby improving the reliability of data, objectively reflecting water regime changes, and effectively guiding engineering scheduling.

First claim

What is claimed is:

1. A real-time abnormity-diagnosing and interpolation method, executed by a processor, for water regime-monitoring data, comprising the following steps:
S1: acquiring water regime-monitoring data, drawing a box plot, recognizing and diagnosing abnormal data in real time based on the box plot;
S2: performing grey correlation analysis on other variables related to a predictor variable;
S3: building a BP neural network model in a neural processing unit (NPU) of a device and making training, with the high-correlation variable compared and selected by the grey correlation analysis as an input to the model, and the predictor variable as an output from the model;
S4: applying the BP neural network model in the neural processing unit (NPU) of the device to predict water regime-monitoring data in real time, and performing abnormity diagnosis and data interpolation, thereby improving reliability of water regime-monitoring data, objectively reflecting changes in water regimes, and effectively guiding engineering scheduling, and predicting and controlling water level to solve problem of water-regime monitoring in open canal engineering to rapidly and effectively identify the abnormal data and interpolate the abnormal data with normal data in real time,
wherein S2 specifically includes:
S21, according to the nature and characteristics of the predictor variable, selecting other variables that can influence and reflect the predictor variable, and performing the gray correlation analysis;
S22: nondimensionalizing the selected variables for grey relational analysis, making the variables with different physical meanings and different data dimensions easy to be compared with each other;
S23: calculating gray correlation coefficients ξ representing the difference between each relevant variable and the predictor variable at a certain moment, of each relevant variable, and simplifying them into Formula (1):

ξ

0
⁢
i

=

Δ
⁡
(
min
)

+

ρΔ
⁡
(
max
)

Δ

0
⁢
i

(
k
)

+

ρΔ
⁡
(
max
)

(
1
)

wherein ξ 0i is a correlation coefficient; ρ is a resolution coefficient, generally between 0 and 1, usually 0.5; Δ(min) is a second-level minimum difference, Δ(max) is a second-level maximum difference; Δ 0i (k) is an absolute difference between each point on each comparison sequence and each point on a reference sequence curve;
S24: calculating the correlation degree of each relevant variable, and concentrating the correlation coefficients at each moment into one value as Formula (2),

r
i

=

1
N

⁢

∑

k
=
1

N

ξ
i

(
k
)

(
2
)

wherein r i is a gray correlation degree of the comparison sequence to the reference sequence, and ξ i is a correlation coefficient calculated in S3;
S25: sequencing the correlation degree of each relevant variable to the predictor variable, reflecting the correlation size between each relevant variable and the predictor variable.

2. The real-time abnormity-diagnosing and interpolation method for water regime-monitoring data according to claim 1 , wherein S1 specifically includes: selecting the water-regime monitoring data continued for two days as 2-hour water-regime monitoring data to draw a box plot, depicting the discrete distribution of said data, adopting the quartile and interquartile range of discrete data as a criteria to judge an abnormal value, so as to identify the abnormal value in the monitoring data.

3. The real-time abnormity-diagnosing and interpolation method for water regime-monitoring data according to claim 1 , wherein said criteria to judge an abnormal value is specifically: taking data less than Q 1 −1.5QR or greater than Q 3 +1.5QR as the criteria to judge abnormal data, where Q 1 is the first quartile, Q 3 is the third quartile, and QR is the interquartile range, with QR=Q 3 −Q.

4. The real-time abnormity-diagnosing and interpolation method for water regime-monitoring data according to claim 1 , wherein said predictor variables refer to real-time water regime-monitoring data, including the water level and flow of check-gates.

5. The real-time abnormity-diagnosing and interpolation method for water regime-monitoring data according to claim 1 , wherein the nondimensionalizing process in S22 specifically includes:
selecting a standardization method, performing a linear transformation on the original data of selected variables, defining minA and maxA as the minimum and maximum values of a variable A, respectively, and mapping an original value x of A into a value x′ which is the dimensionless result of the original value x, within the interval [0,1] through min-max standardization.

6. The real-time abnormity-diagnosing and interpolation method for water regime-monitoring data according to claim 1 , wherein in S3, the high correlation variable in the BP neural network model in the neural processing unit (NPU) of the device is the variable with the largest correlation coefficient.

7. The real-time abnormity-diagnosing a

Assignees

• China Three Gorges Corp
• China Inst Water Resources & Hydropower Res

Inventors

• Wang Hao
• Lei Xiaohui
• Dai Huichao
• Zhang Zhao
• Wang Chao
• Yang Heng
• Zhu Yongnan
• Yang Zhaohui

Classifications

• G06N3/084

  Backpropagation, e.g. using gradient descent · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N3/04

  Architecture, e.g. interconnection topology · CPC title

• G01C25/00

  Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass (testing, calibrating or compensating compasses G01C17/38) · CPC title

• G01C13/00 Primary

  Surveying specially adapted to open water, e.g. sea, lake, river or canal (liquid level metering G01F) · CPC title