CN115935144A - Denoising and reconstructing method for operation and maintenance data - Google Patents
Denoising and reconstructing method for operation and maintenance data Download PDFInfo
- Publication number
- CN115935144A CN115935144A CN202211519663.5A CN202211519663A CN115935144A CN 115935144 A CN115935144 A CN 115935144A CN 202211519663 A CN202211519663 A CN 202211519663A CN 115935144 A CN115935144 A CN 115935144A
- Authority
- CN
- China
- Prior art keywords
- signal
- processing
- noise
- signals
- data
- 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.)
- Pending
Links
Images
Abstract
The invention belongs to the technical field of operation and maintenance systems of power distribution networks, and particularly relates to an operation and maintenance data denoising and reconstructing method, which comprises the following steps: the method for denoising and reconstructing the operation and maintenance data is simple to implement and high in adaptability, can realize integration and modification of local data characteristics and other data related to time-frequency attributes through noise reduction, segmentation smoothing and other means according to the time-frequency attributes of different devices or systems, retains time-frequency attribute elements while realizing the noise-eliminating smoothing and noise-reducing processing of the operation and maintenance data, and provides a processing scheme for partial targeted tasks with high requirements on the time-frequency data attributes.
Description
Technical Field
The invention belongs to the technical field of operation and maintenance systems of power distribution networks, and particularly relates to a denoising and reconstructing method for operation and maintenance data.
Background
The operation and detection data of the power distribution network has the characteristics of large data volume and multiple data types, and relates to the work of integrating and extracting a large amount of data in the daily maintenance and use process of an operation and detection system, wherein multiple attributes such as peak-valley values and the like on a curve are very important node data and are generally applied to the calculation of various parameters and the analysis of trends.
Disclosure of Invention
The invention aims to provide a method for denoising and reconstructing operation and maintenance data on the basis of the existing denoising and reconstructing scheme, in order to make up the problems of insufficient effectiveness of time-frequency characteristic attribute extraction and the like.
In order to achieve the purpose, the invention adopts the following technical scheme.
A denoising and reconstructing method for operation maintenance data comprises the following steps:
step 1, carrying out denoising smoothing processing on an original signal, specifically:
acquiring an original signal f (i) = x (i) + omega · r (i), wherein x (i) is an actual signal, r (i) is a noise signal, and omega is a standard deviation of the noise signal; performing wavelet decomposition on an original signal f (i) = x (i) + omega · r (i) to obtain a high-frequency component and a low-frequency component of the original signal, and reconstructing a signal f' (i) after performing threshold processing respectively;
step 2, time-frequency domain division processing, specifically:
for the reconstructed signal f' (i) obtained after threshold processing, the operation period T of the target object is taken as a reference, and the time period is taken as a basis
The sliding window intercepts the signal to obtain a plurality of segmented signals F (J), wherein a is more than 0 and less than or equal to 1, J =1,2.. J.. J; there may be overlapping portions of data in the segmented signal F (j);
assuming that the segmented signals F (m) and F (n) are two similar segments, the maximum distance S between the similar segmented signals max Can be used as the basis for the judgment of similar segments, S max Correlated with signal noise strength; for any noisy original signal f (m), it is assumedThe corresponding S-G filtering smoothing signal is G (m), and the signal difference value delta f is calculated m Variance σ of = f (m) -G (m) m The signal noise intensity is represented, the optimal signal-to-noise ratio is searched, and the corresponding Euclidean distance sigma is calculated m,r Determining the judgment basis of the noise signal under the original signal f (m);
step 3, threshold value retrieval, specifically:
respectively calculating Euclidean distances under different noise signal intensities for any two segmented signals F (m) and F (n) to obtain corresponding variances sigma m Obtaining a threshold function h (m) = k by using a polynomial fitting method for each group of data 1 σ 3 +k 2 σ 2 +k 3 σ+k 4 ；
The similarity of the segmented signals is expressed as
Judging based on the conditions, if d (m, n) is less than or equal to h (m); the corresponding segmented signals F (m) and F (n) are similar segmented signals, otherwise are differential segmented signals;
step 4, signal recovery processing, specifically:
and performing collaborative smoothing on the extracted similar segmented signals according to the positions of the similar segmented signals in the original signals, performing weighted average processing on a plurality of uniformly distributed segmented signal data belonging to the same position, and then performing smoothing processing on the different segmented signals by directly adopting a weighted average method.
In order to prevent signal oscillation or generate jumping points, a further improvement or preferred embodiment of the denoising and reconstructing method for operation maintenance data obtains threshold parameters of each component based on the following manner: for a signal e to be processed, sorting each element value in the signal from small to large after absolute value processing, and then squaring each element to obtain a new sequence e' (i) = (sort) 2 I =1,2.. K; in the sequence e' (i) = (sorte) 2 Square root of the kth element in (1)
A risk factor of being a threshold value of->
Determining the element e' (k) with the smallest risk factor min If the corresponding threshold value is->
In order to meet the requirement of smooth peak-preserving processing in the present application, in the process of performing wavelet processing, the threshold function of the method is a soft threshold function
The beneficial effects are that:
the denoising and reconstructing method for the operation and maintenance data is simple to implement and high in adaptability, local data features and other data related to time-frequency attributes can be integrated and corrected through noise reduction processing, segmentation smoothing processing and other means according to the time-frequency attributes of different devices or systems, time-frequency attribute elements are reserved while the noise reduction smoothing processing of the operation and maintenance data is achieved, and a processing scheme is provided for a part of targeted tasks with high requirements for the time-frequency data attributes.
Drawings
Fig. 1 is a schematic diagram of a route scheme of a denoising and reconstructing method of operation maintenance data.
Detailed Description
The present invention will be described in detail with reference to specific examples.
Noise signals are inevitably mixed in the operation detection signals of the power distribution network, and because of power supply planning of the power distribution network and periodic control of power supply equipment, the noise signals have complex time-frequency characteristics besides the attributes of traditional pulse signals and white noise signals; in order to realize data volume extraction, key information in the data volume needs to be reserved and extracted, peak extreme point data needs to be extracted and screened, necessary smoothing processing needs to be carried out in the process so that each peak point and extreme point data are smooth and stable, and corresponding time-frequency change characteristics can be reserved.
1. In the original signal processing, due to the existence of noise signals or error data, noise signals are inevitably mixed in the peak point data and the extreme point data, so that the original signals need to be denoised, and therefore, the denoising smoothing processing is performed on the original signals on the basis of the wavelet denoising principle, specifically:
acquiring an original signal f (i) = x (i) + omega · r (i), wherein x (i) is an actual signal, r (i) is a noise signal, and omega is a standard deviation of the noise signal; performing wavelet decomposition on an original signal f (i) = x (i) + omega · r (i) to obtain a high-frequency component and a low-frequency component of the original signal, and reconstructing a signal f' (i) after performing threshold processing respectively;
specifically, to prevent signal oscillation or generation of a jump point, the threshold parameters of the respective components are acquired based on the following manner: for a signal e to be processed, sorting each element value in the signal from small to large after absolute value processing, and then squaring each element to obtain a new sequence e' (i) = (sort) 2 I =1,2.. K.; in the sequence e' (i) = (sorte) 2 Square root of the kth element in (1)
A risk factor which is a threshold value of>
Determining the element e' (k) with the smallest risk factor min Then the corresponding threshold is->
In particular, due to the requirement of smooth peak-keeping processing in the present application, the threshold function thereof adopts a soft threshold function in the process of performing wavelet processing
The operation and inspection data of the power distribution network is huge in overall data volume, if a stable recording period is short, the data can be calculated in months or years after a day, meanwhile, the working periods of different devices and systems in the power distribution network are inconsistent, and meanwhile, the data of other periods with close time periods can be influenced by each operation period.
2. For massive operation and detection data, in order to conveniently acquire attribute data of a specific target such as a certain system or equipment and improve the data processing efficiency, the optimal scheme is to adopt a time-division segmentation processing mode according to the time-frequency characteristics of corresponding data types to ensure the data processing efficiency of each segment, and simultaneously, smooth processing is carried out by using the similar or same data characteristics of each segment, so that signals with similar or related data characteristics can be related to each other, specifically:
for the reconstructed signal f' (i) obtained after threshold processing, the operation period T of the target object is taken as a reference, and the time period is taken as a basis
The sliding window intercepts the signals to obtain a plurality of segmented signals F (J), wherein a is more than 0 and less than or equal to 1, J =1,2.. J.. J; there may be overlapping portions of data in the segmented signal F (j);
assuming that the segmented signals F (m) and F (n) are two similar segments, the maximum distance S between the similar segmented signals max Can be used as the basis for the judgment of similar segments, S max Correlated with signal noise strength; for any original signal f (m) containing noise, assuming that the corresponding S-G filtered smooth signal is G (m), calculating a signal difference value delta f m Variance σ of = f (m) -G (m) m Characterizing the signal noise intensity, searching the optimal signal-to-noise ratio and calculating the corresponding Euclidean distance sigma m,r Determining the judgment basis of the noise signal under the original signal f (m);
because the influence degrees of the noise signals in the whole signal are not consistent under different noise intensities, the noise signals are convenient to acquire at randomExtracting noise signals under the intensity, and respectively calculating Euclidean distances under different noise signal intensities for any two segmented signals F (m) and F (n) to obtain corresponding variance sigma m For each group of data, it is found through analysis that a very stable adaptive function can be obtained through multi-section fitting, so that a threshold function h (m) = k is obtained by using a polynomial fitting method 1 σ 3 +k 2 σ 2 +k 3 σ+k 4 (ii) a The similarity of the segmented signals can be expressed as
It should satisfy d (m, n) is less than or equal to h (m);
judging based on the conditions, if d (m, n) is less than or equal to h (m); the corresponding segmented signals F (m) and F (n) are similar segmented signals, otherwise are differential segmented signals;
and performing collaborative smoothing treatment on the extracted similar segmented signals according to the positions of the extracted similar segmented signals in the original signals, performing weighted average treatment on a plurality of uniformly distributed segmented signal data belonging to the same position, and performing smoothing treatment on the different segmented signals by directly adopting a weighted average method.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the protection scope of the present invention, although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions can be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims (3)
1. A denoising and reconstructing method for operation maintenance data is characterized by comprising the following steps:
step 1, carrying out denoising smoothing processing on an original signal, specifically:
acquiring an original signal f (i) = x (i) + omega · r (i), wherein x (i) is an actual signal, r (i) is a noise signal, and omega is a standard deviation of the noise signal; performing wavelet decomposition on an original signal f (i) = x (i) + omega · r (i) to obtain a high-frequency component and a low-frequency component of the original signal, and reconstructing a signal f' (i) after performing threshold processing respectively;
step 2, time-frequency domain division processing, specifically:
for the reconstructed signal f' (i) obtained after threshold processing, the operation period T of the target object is taken as a reference, and the time period is taken as a basis
The sliding window intercepts the signal to obtain a plurality of segmented signals F (J), wherein a is more than 0 and less than or equal to 1, J =1,2.. J.. J; there may be overlapping portions of data in the segmented signal F (j);
assuming that the segmented signals F (m) and F (n) are two similar segments, the maximum distance S between the similar segmented signals max Can be taken as the basis for the judgment of similar segments, S max Correlated with signal noise strength; for any original signal f (m) containing noise, assuming that the corresponding S-G filtering smooth signal is G (m), calculating a signal difference value delta f m Variance σ of = f (m) -G (m) m Characterizing the signal noise intensity, searching the optimal signal-to-noise ratio and calculating the corresponding Euclidean distance sigma m,r Determining the judgment basis of the noise signal under the original signal f (m);
step 3, threshold retrieval, specifically:
respectively calculating Euclidean distances under different noise signal intensities for any two segmented signals F (m) and F (n) to obtain corresponding variances sigma m Obtaining a threshold function h (m) = k by using a polynomial fitting method for each group of data 1 σ 3 +k 2 σ 2 +k 3 σ+k 4 ；
The similarity of the segmented signals is expressed as
Judging based on the conditions, if d (m, n) is less than or equal to h (m); the corresponding segmented signals F (m) and F (n) are similar segmented signals, otherwise are differential segmented signals;
step 4, signal recovery processing, specifically:
and performing collaborative smoothing on the extracted similar segmented signals according to the positions of the similar segmented signals in the original signals, performing weighted average processing on a plurality of uniformly distributed segmented signal data belonging to the same position, and then performing smoothing processing on the different segmented signals by directly adopting a weighted average method.
2. The method for denoising and reconstructing operation maintenance data according to claim 1, wherein to prevent signal oscillation or generate jumping points, the threshold parameters of each component are obtained based on the following manner: for a signal e to be processed, sorting each element value in the signal from small to large after absolute value processing, and squaring each element to obtain a new sequence e' (i) = (sort | e |) 2 I =1,2.. K; in the sequence e' (i) = (sort | e |) 2 Square root of the kth element in (1)
A risk factor of being a threshold value of->
Determining the element e' (k) with the smallest risk factor min If the corresponding threshold value is->
3. The method for denoising and reconstructing operation maintenance data according to claim 1, wherein a soft threshold function is adopted as the threshold function in the wavelet processing process according to the requirement of smooth peak-preserving processing in the application
/>
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211519663.5A CN115935144A (en) | 2022-11-30 | 2022-11-30 | Denoising and reconstructing method for operation and maintenance data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211519663.5A CN115935144A (en) | 2022-11-30 | 2022-11-30 | Denoising and reconstructing method for operation and maintenance data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115935144A true CN115935144A (en) | 2023-04-07 |
Family
ID=86648474
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211519663.5A Pending CN115935144A (en) | 2022-11-30 | 2022-11-30 | Denoising and reconstructing method for operation and maintenance data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115935144A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116976684A (en) * | 2023-09-25 | 2023-10-31 | 尚古智造(山东)智能装备有限公司 | Risk model predictive control method and system for logistics conveyor |
| CN117421937A (en) * | 2023-12-18 | 2024-01-19 | 山东利恩斯智能科技有限公司 | Method for inhibiting random vibration signal zero drift trend of sensor based on S-G algorithm |
-
2022
- 2022-11-30 CN CN202211519663.5A patent/CN115935144A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116976684A (en) * | 2023-09-25 | 2023-10-31 | 尚古智造(山东)智能装备有限公司 | Risk model predictive control method and system for logistics conveyor |
| CN116976684B (en) * | 2023-09-25 | 2024-01-02 | 尚古智造(山东)智能装备有限公司 | Risk model predictive control method and system for logistics conveyor |
| CN117421937A (en) * | 2023-12-18 | 2024-01-19 | 山东利恩斯智能科技有限公司 | Method for inhibiting random vibration signal zero drift trend of sensor based on S-G algorithm |
| CN117421937B (en) * | 2023-12-18 | 2024-03-29 | 山东利恩斯智能科技有限公司 | Method for inhibiting random vibration signal zero drift trend of sensor based on S-G algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115935144A (en) | Denoising and reconstructing method for operation and maintenance data | |
| Bai et al. | Automatic detection and removal of high‐density impulse noises | |
| CN109218223B (en) | Robust network traffic classification method and system based on active learning | |
| Meher et al. | An improved recursive and adaptive median filter for high density impulse noise | |
| Xu et al. | Edge detection algorithm of medical image based on Canny operator | |
| CN108491886A (en) | A kind of sorting technique of the polynary time series data based on convolutional neural networks | |
| CN110070548B (en) | Deep learning training sample optimization method | |
| CN111079559A (en) | Convolutional neural network-based mud pulse signal identification method | |
| CN113283331A (en) | Multi-class target identification method and system for unattended sensor system | |
| Fan et al. | Image denoising based on wavelet thresholding and Wiener filtering in the wavelet domain | |
| Wang et al. | An improved region growing method for medical image selection and evaluation based on Canny edge detection | |
| CN114595728A (en) | Signal denoising method based on self-supervision learning | |
| CN105550646B (en) | Generalized illumination invariant face feature description method based on logarithmic gradient histogram | |
| CN111091107A (en) | Face region edge detection method and device and storage medium | |
| CN110287853B (en) | Transient signal denoising method based on wavelet decomposition | |
| Kumar et al. | A hybrid method for the removal of RVIN using self organizing migration with adaptive dual threshold median filter | |
| Limin et al. | Low probability of intercept radar signal recognition based on the improved AlexNet model | |
| CN108241874A (en) | Video text area positioning method based on BP neural network and spectrum analysis | |
| CN109995690B (en) | Neural network self-optimization method for MFSK digital signal subclass modulation recognition | |
| CN110632191A (en) | Transformer chromatographic peak qualitative method and system based on decision tree algorithm | |
| CN113780301B (en) | Self-adaptive denoising machine learning application method for defending against attack | |
| CN113538275B (en) | Method and system for recovering fruit occlusion based on cycleGAN | |
| Kaur et al. | Analaysis of diseases in fruits using image proccessing techniques | |
| CN114626412A (en) | Multi-class target identification method and system for unattended sensor system | |
| Yi et al. | Adaptive threshold based video shot boundary detection framework |
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 |