CN117009750B - Methane concentration data complement method and device for machine learning - Google Patents
Methane concentration data complement method and device for machine learning Download PDFInfo
- Publication number
- CN117009750B CN117009750B CN202311266418.2A CN202311266418A CN117009750B CN 117009750 B CN117009750 B CN 117009750B CN 202311266418 A CN202311266418 A CN 202311266418A CN 117009750 B CN117009750 B CN 117009750B
- Authority
- CN
- China
- Prior art keywords
- methane concentration
- concentration data
- data
- methane
- module
- 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.)
- Active
Links
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 title claims abstract description 362
- 230000000295 complement effect Effects 0.000 title claims abstract description 52
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000010801 machine learning Methods 0.000 title claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 23
- 238000007781 pre-processing Methods 0.000 claims abstract description 16
- 230000001502 supplementing effect Effects 0.000 claims abstract description 6
- 238000009499 grossing Methods 0.000 claims description 16
- 238000013528 artificial neural network Methods 0.000 claims description 14
- 238000012795 verification Methods 0.000 claims description 13
- 238000010219 correlation analysis Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 9
- 230000002457 bidirectional effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 239000013589 supplement Substances 0.000 claims description 2
- 239000011159 matrix material Substances 0.000 claims 1
- 238000012544 monitoring process Methods 0.000 description 8
- 239000003345 natural gas Substances 0.000 description 8
- 238000010586 diagram Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The disclosure discloses a methane concentration data complement method and device for machine learning, wherein the method comprises the following steps: s100: collecting methane concentration data to be complemented; s200: preprocessing the collected methane concentration data; s300: constructing a methane concentration data complement model and training; s400: and inputting the preprocessed methane concentration data into a trained methane concentration data complement model to complement the methane concentration data to be complemented. According to the methane concentration data supplementing method and device, the residual error module, the self-attention mechanism module and the like are introduced to complete methane concentration data, different weights can be distributed to the methane data at different moments, and the reconstruction accuracy of the methane concentration can be effectively improved.
Description
Technical Field
The disclosure belongs to the technical field of methane data processing, and particularly relates to a methane concentration data complement method and device for machine learning.
Background
Natural gas is an important clean energy source, but natural gas leakage may occur for various reasons during the exploitation, transportation and use of natural gas, thereby causing environmental pollution and safety hazards. Methane is the main component of natural gas, and monitoring the concentration of methane can effectively detect the leakage condition of natural gas.
Traditional methane monitoring methods include hand-held methane detectors, methane analyzers and the like, which require manual operation, have low measurement efficiency and limited accuracy, and cannot locate leakage sources. In recent years, with the development and application of the internet of things technology, a natural gas leakage monitoring system based on a sensor network gradually becomes an emerging monitoring method, and the method utilizes a plurality of sensors to monitor the natural gas leakage condition in real time and transmits data through a network, so that the collection and processing process of the monitoring data is more automatic and efficient.
However, in actual monitoring, the sensor may fail or the data collection is incomplete for various reasons, which may affect the integrity and accuracy of the monitored data. Therefore, how to automatically complement the missing monitoring data becomes an important problem to be solved by the current natural gas leakage monitoring system. In recent years, many data-driven methods have been proposed and applied to methane data parameter estimation and reservoir description, such as Artificial Neural Networks (ANN), fuzzy Logic Models (FLM), decision Trees (DT), and Support Vector Machines (SVM). Because the methane data has obvious time or space sequence characteristics, long sequence dependency relationship exists between the data, the machine learning method cannot effectively extract the dependency relationship of the methane data, and the defects of low calculation efficiency, easiness in fitting and the like exist, so that the method cannot be fully applied to methane data completion to a certain extent.
Disclosure of Invention
In view of the shortcomings in the prior art, an object of the present disclosure is to provide a methane concentration data complement method for machine learning, which can effectively improve the reconstruction accuracy of methane concentration by constructing a methane data complement model.
In order to achieve the above object, the present disclosure provides the following technical solutions:
a methane concentration data completion method for machine learning, comprising the steps of:
s100: collecting methane concentration data to be complemented;
s200: preprocessing the collected methane concentration data;
s300: constructing a methane concentration data complement model and training;
the methane concentration data complement model comprises an input layer, a residual error module, a self-attention mechanism module and a full-connection layer, wherein the residual error module comprises a plurality of bidirectional long-short-time memory neural networks,
s400: inputting the preprocessed methane concentration data into a trained methane concentration data complement model, wherein the methane concentration data complement model can allocate different weights for the methane data at different moments so as to complement the methane concentration data to be complemented;
in step S200, the preprocessing of the methane concentration data includes the following steps:
s201: normalizing the methane concentration data;
s202: smoothing the normalized methane concentration data;
s203: and carrying out correlation analysis on the methane concentration data after the smoothing treatment.
Preferably, in step S300, the methane concentration data complement model is trained by the following method:
s301: collecting an incomplete methane concentration data sample, and preprocessing the data sample to obtain a preprocessed methane concentration data sample;
s302: dividing the preprocessed methane concentration data sample into a training set and a verification set;
s303: training the methane concentration data complement model by using a training set, and completing model training when the training reaches the maximum iteration round number;
s304: and verifying the trained methane concentration data complement model by using a verification set, visually comparing the predicted value and the true value output by the model in the verification process, and if the error is smaller than 0.1, passing the model verification.
Preferably, step S400 includes the steps of:
s401: inputting the pretreated methane concentration data into a trained methane concentration data complement model to obtain the initial concentration of the methane concentration data;
s402: the length before the initial concentration isAs initial data, is input into a two-way long-short-time memory neural network through an input layer to extract the characteristics of methane concentration data;
s403: different weights are distributed for the extracted features at different moments, and the features with the distributed weights are input into a full-connection layer for prediction so as to output a prediction result;
s404: step S403 is repeatedly executed, and each prediction is based on the prediction result of the last step in the sequence, so as to complete the completion of the methane concentration data to be completed.
The present disclosure also provides a methane concentration data complement apparatus for machine learning, comprising:
the acquisition module is used for acquiring methane concentration data to be complemented;
the pretreatment module is used for preprocessing the collected methane concentration data;
the model construction module is used for constructing a methane concentration data complement model and training;
the methane concentration data complement model comprises an input layer, a residual error module, a self-attention mechanism module and a full-connection layer, wherein the residual error module comprises a plurality of bidirectional long-short-time memory neural networks,
the methane concentration data after pretreatment is input into a trained methane concentration data supplementing model, and the methane concentration data supplementing model can allocate different weights for the methane data at different moments so as to supplement the methane concentration data to be supplemented;
wherein, the preprocessing module includes:
the normalization sub-module is used for normalizing the methane concentration data;
the smoothing processing submodule is used for carrying out smoothing processing on the normalized methane concentration data;
and the correlation analysis submodule is used for carrying out correlation analysis on the methane concentration data after the smoothing treatment.
Preferably, the complement module includes:
the acquisition submodule is used for acquiring the preprocessed methane concentration data;
the characteristic extraction submodule is used for extracting the characteristics of the methane concentration data;
the weight distribution sub-module is used for distributing different weights for the extracted characteristics at different moments and predicting the characteristics;
and the completion sub-module is used for completing the completion of the methane concentration data to be completed based on the prediction result.
The present disclosure also provides an electronic device, characterized by comprising:
a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein,
the processor, when executing the program, implements a method as described in any of the preceding.
The present disclosure also provides a computer storage medium storing computer executable instructions for performing a method as described in any one of the preceding claims.
Compared with the prior art, the beneficial effects that this disclosure brought are:
1. the model is a time sequence model, so that the methane time sequence data characteristics can be effectively extracted;
2. because the model introduces a residual error module, a self-attention mechanism module and the like to realize methane concentration data completion, different weights can be distributed for methane data at different moments, and the reconstruction accuracy of methane concentration can be effectively improved.
Drawings
FIG. 1 is a flow chart of a methane concentration data completion method for machine learning provided in one embodiment of the present disclosure;
FIG. 2 is a schematic diagram of a network structure of a methane concentration data completion model provided in one embodiment of the present disclosure;
FIG. 3 is a schematic diagram of a residual module in the network structure shown in FIG. 2;
FIG. 4 is a schematic diagram of the attention mechanism module in the network architecture of FIG. 2;
fig. 5 is a schematic diagram of a methane concentration data completion process provided by another embodiment of the present disclosure.
Detailed Description
Specific embodiments of the present disclosure will be described in detail below with reference to fig. 1 to 5. While specific embodiments of the disclosure are shown in the drawings, it should be understood that the disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
It should be noted that certain terms are used throughout the description and claims to refer to particular components. Those of skill in the art will understand that a person may refer to the same component by different names. The specification and claims do not identify differences in terms of components, but rather differences in terms of the functionality of the components. As used throughout the specification and claims, the terms "include" and "comprise" are used in an open-ended fashion, and thus should be interpreted to mean "include, but not limited to. The description hereinafter sets forth the preferred embodiments for carrying out the present disclosure, but is not intended to limit the scope of the disclosure in general, as the description proceeds. The scope of the present disclosure is defined by the appended claims.
For the purposes of promoting an understanding of the embodiments of the disclosure, reference will now be made to the embodiments illustrated in the drawings and specific examples, without the intention of being limiting the embodiments of the disclosure.
In one embodiment, as shown in fig. 1, the present disclosure provides a methane concentration data supplementing method, comprising the steps of:
s100: collecting methane concentration data to be complemented;
s200: preprocessing the collected methane concentration data;
s300: constructing a methane concentration data complement model and training;
the methane concentration data complement model comprises an input layer, a residual error module, a self-attention mechanism module and a full-connection layer, wherein the residual error module comprises a plurality of bidirectional long-short-time memory neural networks,
s400: and inputting the preprocessed methane concentration data into a trained methane concentration data complement model, wherein the methane concentration data complement model can allocate different weights for the methane data at different moments so as to complement the methane concentration data to be complemented.
In another embodiment, in step S200, since the collected raw data of methane concentration have abnormal values and different dimensions, a pretreatment is required, and the specific pretreatment process includes the following steps:
s201: normalizing the methane concentration data;
in this step, methane concentration data was normalized using the MaxMin algorithm to unify dimensions.
S202: smoothing the normalized methane concentration data;
in the step, the normalized methane concentration data is smoothed by Kalman filtering to eliminate the influence of abnormal values on the complement result.
S203: and carrying out correlation analysis on the methane concentration data after the smoothing treatment.
In the step, the data after the smoothing processing is subjected to correlation analysis by using a Pearson algorithm, and main factors influencing the loss of the methane concentration data, such as pressure, temperature, wind speed and the like, are found out.
In another embodiment, in step S300, as shown in fig. 2, the methane concentration data complement model includes: an input layer, a residual module, a self-attention mechanism module and a full connection layer.
In this embodiment, in the methane concentration data, the data affected by each sampling point may be up to several hundred, which includes not only a portion before the current data but also a portion subsequent to the current data. This means that methane concentration data reconstruction and generation is typically a long-sequence data analysis problem with bi-directional spatial correlation. In order to solve the problem, the embodiment introduces a residual error module and a self-attention mechanism module in the methane concentration data complement model. Firstly, the residual error module is utilized to properly increase the network depth (for example, the network depth is increased to 4 layers), so that the capability of extracting data semantic information of the model can be improved, the problem of gradient disappearance caused by the increase of the network depth is solved, and a network model with deeper hierarchy is constructed to obtain more accurate data characteristic expression. It should be noted that, as shown in fig. 2, the residual module in this embodiment includes a plurality of Bi-directional long-short-time memory neural networks (Bi-LSTM) to extract data features as a base network. Wherein, as shown in fig. 3, each Bi-LSTM is composed of a forward LSTM network and a backward LSTM network. As two LSTM networks with opposite time sequences, the forward LSTM can acquire the information of the reservoir section before the missing sequence, the backward LSTM can acquire the information of the reservoir section after the missing sequence, namely the model can fully acquire the information of the upper reservoir section and the lower reservoir section from the front aspect and the rear aspect, and the complementing effect of the model on methane data is improved.
As shown in fig. 4, the self-attention mechanism module includes a feature similarity calculation layer andweight importance assignment layer. The characteristic similarity calculation layer mainly outputs residual error modules at the times of t-1, t and t+1 respectively、Andmining correlations between known different formation depth sampling data and the layer to be completed as inputs and outputting correlation coefficientsThe correlation coefficient is specifically calculated by the following formula:
wherein,in order to activate the function,andis a parameter to be trained in the self-attention mechanism module.
Self-attention mechanism module adoptionThe function implements an attention scoring operation,the function calculates the correlation coefficient of the output of the layer of characteristic similarityConversion into weights, i.e. state signal weight matricesAnd thus the importance of the known methane curve to the missing curve.
Wherein,expressed in terms ofIs a soleT represents different time steps.
And then, correspondingly weighting the input methane concentration data, adopting different weights for data points of reservoirs with different depths, and predicting the weighted methane data as the input of the fully connected layer.
In another embodiment, in step S300, the training process of the methane concentration data complement model includes the following steps:
s301: collecting an incomplete methane concentration data sample, and preprocessing the data sample to obtain a preprocessed methane concentration data sample;
in this step, the embodiment selects methane concentration data collected by 35 sensors for 3 days, wherein each sensor performs data collection every 0.5 seconds, and 18144000 pieces of data are collected in total. Next, the acquired data is preprocessed according to the preprocessing step.
S302: dividing the preprocessed methane concentration data sample into a training set and a verification set according to the ratio of 7:3;
s303: training the methane concentration data complement model by using a training set, and completing model training when the training reaches the maximum iteration round number (for example, the maximum iteration round number is set to be 500 times);
s304: and verifying the trained methane concentration data complement model by using a verification set, visually comparing a predicted value output by the model with a true value acquired by a sensor in the verification process, and if the error is smaller than 0.1, passing the model verification.
In another embodiment, as shown in fig. 5, step S400 includes the steps of:
s401: inputting the pretreated methane concentration data into a trained methane concentration data complement model to obtain the initial concentration of the methane concentration data;
s402: the length before the initial concentration isAs initial data, is input into a two-way long-short-time memory neural network through an input layer to extract the characteristics of methane concentration data;
s403: different weights are distributed for the extracted features at different moments, and the features with the distributed weights are input into a full-connection layer for prediction so as to output a prediction result;
s404: step S403 is repeatedly executed, and each prediction is based on the prediction result of the last step in the sequence, so as to complete the complement of the missing methane concentration data.
In another embodiment, the present disclosure also provides a methane concentration data completion apparatus for machine learning, comprising:
the acquisition module is used for acquiring methane concentration data to be complemented;
the pretreatment module is used for preprocessing the collected methane concentration data;
the model construction module is used for constructing a methane concentration data complement model and training;
the methane concentration data complement model comprises an input layer, a residual error module, a self-attention mechanism module and a full-connection layer, wherein the residual error module comprises a plurality of bidirectional long-short-time memory neural networks,
and the complementing module is used for inputting the preprocessed methane concentration data into a trained methane concentration data complementing model, wherein the methane concentration data complementing model can allocate different weights for the methane data at different moments so as to complement the methane concentration data to be complemented.
In another embodiment, the preprocessing module includes:
the normalization sub-module is used for normalizing the methane concentration data;
the smoothing processing submodule is used for carrying out smoothing processing on the normalized methane concentration data;
and the correlation analysis submodule is used for carrying out correlation analysis on the methane concentration data after the smoothing treatment.
In another embodiment, the complement module includes:
the acquisition submodule is used for acquiring the preprocessed methane concentration data;
the characteristic extraction submodule is used for extracting the characteristics of the methane concentration data;
the weight distribution sub-module is used for distributing different weights for the extracted characteristics at different moments and predicting the characteristics;
and the completion sub-module is used for completing the completion of the methane concentration data to be completed based on the prediction result.
In another embodiment, the present disclosure further provides an electronic device, including:
a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein,
the processor, when executing the program, implements a method as described in any of the preceding.
In another embodiment, the present disclosure also provides a computer storage medium storing computer-executable instructions for performing the method of any one of the preceding claims.
Although embodiments of the present invention have been described above with reference to the accompanying drawings, the present invention is not limited to the specific embodiments and application fields described above, wherein the verification object is not limited to a specific sensor arrangement angle or a split leaf disk structure, and the specific embodiments described above are merely illustrative, instructive, and not restrictive. Those skilled in the art, having the benefit of this disclosure, may effect numerous forms of the invention without departing from the scope of the invention as claimed.
Claims (2)
1. A methane concentration data completion method for machine learning, the method comprising the steps of:
s100: collecting methane concentration data to be complemented;
s200: preprocessing the collected methane concentration data;
s300: constructing a methane concentration data complement model and training;
the methane concentration data complement model comprises an input layer, a residual error module, a self-attention mechanism module and a full-connection layer, wherein the residual error module comprises a plurality of bidirectional long-short-time memory neural networks, and is specifically:
the residual module increases the network depth to 4 layers and comprises 4 layers of two-way long short-time memory neural networks, wherein each two-way long-time memory neural network consists of a forward LSTM network and a backward LSTM network, the two LSTM networks are used as two LSTM networks with opposite time sequences, the forward LSTM acquires the information of a reservoir section before a missing sequence, the backward LSTM acquires the information of the reservoir section after the missing sequence, and the information of an upper reservoir section and a lower reservoir section are fully acquired from the front aspect and the rear aspect;
the self-attention mechanism module comprises a feature similarity calculation layerThe weight importance distribution layer, wherein the characteristic similarity calculation layer calculates the output h of the residual error module at the time t-1, t and t+1 respectively t-1 、h t And h t+1 As input, the correlation between the known different depth of layer sample data and the layer to be complemented is mined and the correlation coefficient +.>The correlation coefficient is specifically calculated by the following formula:
；
wherein,to activate the function +.>And->Parameters to be trained in the self-attention mechanism module;
self-attention mechanism module adoptionFunction implementation attention scoring operations，/>Correlation coefficient of function output of feature similarity calculation layer>Converted into weight, i.e. state signal weight matrix +.>And thereby obtaining the importance of the known methane curve to the missing curve;
；
wherein exp (r) denotes e as a baseT represents different time steps;
then, correspondingly weighting the input methane concentration data, adopting different weights for data points of reservoirs with different depths, and predicting the weighted methane data as the input of the full-connection layer;
s400: inputting the preprocessed methane concentration data into a trained methane concentration data complement model, wherein the methane concentration data complement model can allocate different weights for the methane data at different moments so as to complement the methane concentration data to be complemented;
wherein,
in step S200, preprocessing the collected methane concentration data includes the steps of:
s201: normalizing the methane concentration data;
s202: smoothing the normalized methane concentration data;
s203: performing correlation analysis on the methane concentration data after the smoothing treatment;
step S400 includes the steps of:
s401: inputting the pretreated methane concentration data into a trained methane concentration data complement model to obtain the initial concentration of the methane concentration data;
s402: the length before the initial concentration isAs initial data, is input into a two-way long-short-time memory neural network through an input layer to extract the characteristics of methane concentration data;
s403: different weights are distributed for the extracted features at different moments, and the features with the distributed weights are input into a full-connection layer for prediction so as to output a prediction result;
s404: step S403 is repeatedly executed, and each prediction is based on the prediction result of the last step in the sequence so as to complete the completion of the methane concentration data to be completed;
wherein the methane concentration data complement model is trained by:
s301: the method comprises the steps of preprocessing a collected methane concentration missing data sample to obtain a preprocessed methane concentration missing data sample;
s302: dividing the pretreated methane concentration missing data sample into a training set and a verification set;
s303: training the methane concentration data complement model by using a training set, and completing model training when the training reaches the maximum iteration round number;
s304: and verifying the trained methane concentration data complement model by using a verification set, visually comparing the predicted value and the true value output by the model in the verification process, and if the error is smaller than 0.1, passing the model verification.
2. A methane concentration data replenishing apparatus for machine learning that implements the methane concentration data replenishing method of claim 1, characterized in that the apparatus comprises:
the acquisition module is used for acquiring methane concentration data to be complemented;
the pretreatment module is used for preprocessing the collected methane concentration data;
the model construction module is used for constructing a methane concentration data complement model and training;
the methane concentration data complement model comprises an input layer, a residual error module, a self-attention mechanism module and a full-connection layer, wherein the residual error module comprises a plurality of bidirectional long-short-time memory neural networks,
the methane concentration data after pretreatment is input into a trained methane concentration data supplementing model, and the methane concentration data supplementing model can allocate different weights for the methane data at different moments so as to supplement the methane concentration data to be supplemented;
wherein,
the preprocessing module comprises:
the normalization sub-module is used for normalizing the methane concentration data;
the smoothing processing submodule is used for carrying out smoothing processing on the normalized methane concentration data;
the correlation analysis submodule is used for carrying out correlation analysis on the methane concentration data after the smoothing treatment;
the complement module includes:
the acquisition submodule is used for acquiring the preprocessed methane concentration data;
the characteristic extraction submodule is used for extracting the characteristics of the methane concentration data;
the weight distribution sub-module is used for distributing different weights for the extracted characteristics at different moments and predicting the characteristics;
and the completion sub-module is used for completing the completion of the methane concentration data to be completed based on the prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311266418.2A CN117009750B (en) | 2023-09-28 | 2023-09-28 | Methane concentration data complement method and device for machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311266418.2A CN117009750B (en) | 2023-09-28 | 2023-09-28 | Methane concentration data complement method and device for machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117009750A CN117009750A (en) | 2023-11-07 |
| CN117009750B true CN117009750B (en) | 2024-01-02 |
Family
ID=88569419
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311266418.2A Active CN117009750B (en) | 2023-09-28 | 2023-09-28 | Methane concentration data complement method and device for machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117009750B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113094357A (en) * | 2021-04-23 | 2021-07-09 | 大连理工大学 | Traffic missing data completion method based on space-time attention mechanism |
| CN113821760A (en) * | 2021-11-23 | 2021-12-21 | 湖南工商大学 | Air data completion method, device, equipment and storage medium |
| CN115935139A (en) * | 2023-01-09 | 2023-04-07 | 吉林大学 | Space field interpolation method for ocean observation data |
| CN116051936A (en) * | 2023-03-23 | 2023-05-02 | 中国海洋大学 | Chlorophyll concentration ordered complement method based on space-time separation external attention |
| CN116312861A (en) * | 2023-05-09 | 2023-06-23 | 济南作为科技有限公司 | Denitration system gas concentration prediction method, device, equipment and storage medium |
| CN116402874A (en) * | 2023-04-13 | 2023-07-07 | 哈尔滨工业大学 | Spacecraft depth complementing method based on time sequence optical image and laser radar data |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111562358B (en) * | 2020-05-06 | 2021-03-16 | 武汉大学 | Transformer oil gas content prediction method and system based on combined model |
-
2023
- 2023-09-28 CN CN202311266418.2A patent/CN117009750B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113094357A (en) * | 2021-04-23 | 2021-07-09 | 大连理工大学 | Traffic missing data completion method based on space-time attention mechanism |
| CN113821760A (en) * | 2021-11-23 | 2021-12-21 | 湖南工商大学 | Air data completion method, device, equipment and storage medium |
| CN115935139A (en) * | 2023-01-09 | 2023-04-07 | 吉林大学 | Space field interpolation method for ocean observation data |
| CN116051936A (en) * | 2023-03-23 | 2023-05-02 | 中国海洋大学 | Chlorophyll concentration ordered complement method based on space-time separation external attention |
| CN116402874A (en) * | 2023-04-13 | 2023-07-07 | 哈尔滨工业大学 | Spacecraft depth complementing method based on time sequence optical image and laser radar data |
| CN116312861A (en) * | 2023-05-09 | 2023-06-23 | 济南作为科技有限公司 | Denitration system gas concentration prediction method, device, equipment and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| methane concentration prediction method based on deep learning and classical time series analysis;Xiangrui Meng等;energies;全文 * |
| 基于PSO-Adam-GRU的煤矿瓦斯浓度预测模型;马莉;西安科技大学学报(02);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117009750A (en) | 2023-11-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110221225B (en) | Spacecraft lithium ion battery cycle life prediction method | |
| CN110232280B (en) | Software security vulnerability detection method based on tree structure convolutional neural network | |
| CN109685314B (en) | Non-intrusive load decomposition method and system based on long-term and short-term memory network | |
| CN113255733A (en) | Unsupervised anomaly detection method under multi-modal data loss | |
| CN109886021A (en) | A kind of malicious code detecting method based on API overall situation term vector and layered circulation neural network | |
| CN110780878A (en) | Method for carrying out JavaScript type inference based on deep learning | |
| CN113434685A (en) | Information classification processing method and system | |
| CN115017791A (en) | Tunnel surrounding rock grade identification method and device | |
| CN112763967A (en) | BiGRU-based intelligent electric meter metering module fault prediction and diagnosis method | |
| CN114166509A (en) | Motor bearing fault prediction method | |
| CN116361191A (en) | Software compatibility processing method based on artificial intelligence | |
| CN117521512A (en) | Bearing residual service life prediction method based on multi-scale Bayesian convolution transducer model | |
| CN114974306A (en) | Transformer abnormal voiceprint detection and identification method and device based on deep learning | |
| CN117009750B (en) | Methane concentration data complement method and device for machine learning | |
| CN116992757A (en) | Wellhead pressure prediction method and device based on deep learning and rolling optimization | |
| CN116680639A (en) | Deep-learning-based anomaly detection method for sensor data of deep-sea submersible | |
| CN115184734A (en) | Power grid line fault detection method and system | |
| CN113297356A (en) | Information classification method and system based on BERT model | |
| CN114462127A (en) | Structural damage identification method and system based on deep extreme learning machine | |
| CN118195179B (en) | Ecological system restoration method based on data analysis | |
| Sugave et al. | Multi-Objective Optimization Model and Hierarchical Attention Networks for Mutation Testing | |
| CN113837440B (en) | Blasting effect prediction method and device, electronic equipment and medium | |
| Deutsch | Development of deep learning based prognostics for rotating component | |
| Qin et al. | Macroscopic-Microscopic Attention in LSTM Networks based on fusion Features for prediction of bearing remaining life | |
| CN117093867A (en) | TBM stable segment performance parameter prediction method based on ascending segment data |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |