CN111765449B - Superheater explosion and leakage early warning method based on long-term and short-term memory network - Google Patents
Superheater explosion and leakage early warning method based on long-term and short-term memory network Download PDFInfo
- Publication number
- CN111765449B CN111765449B CN202010912762.4A CN202010912762A CN111765449B CN 111765449 B CN111765449 B CN 111765449B CN 202010912762 A CN202010912762 A CN 202010912762A CN 111765449 B CN111765449 B CN 111765449B
- Authority
- CN
- China
- Prior art keywords
- superheater
- data
- label
- explosion
- sample
- 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
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F22—STEAM GENERATION
- F22B—METHODS OF STEAM GENERATION; STEAM BOILERS
- F22B37/00—Component parts or details of steam boilers
- F22B37/02—Component parts or details of steam boilers applicable to more than one kind or type of steam boiler
- F22B37/42—Applications, arrangements, or dispositions of alarm or automatic safety devices
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Thermal Sciences (AREA)
- Mechanical Engineering (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a superheater explosion and leakage early warning method based on a long-term and short-term memory network. According to the invention, the measured point data of the temperature and the pressure of the boiler superheater are regarded as the matrixed time series data, the rule of the input data on the time dimension is learned by utilizing the good learning and predicting capability of the long-short term memory network on the time series data, and the analysis of the superheater pipe explosion problem based on data driving is realized, so that the purpose of early warning the superheater pipe explosion is achieved. The method has high accuracy, the superheater pipe explosion problem is early warned in a data driving mode, and the model accurately warns the superheater pipe explosion in a data test stage and after the model is put into use. The timeliness and the comprehensiveness are high, and compared with manual equipment troubleshooting, the model can monitor the operation of the whole superheater in real time and send out early warning to pipe explosion in advance; the cost is low and the operability is strong.
Description
Technical Field
The invention belongs to the technical field of thermal power plant boilers, and particularly relates to a superheater explosion and leakage early warning method based on a long-term and short-term memory network.
Background
The problem of tube explosion of four tubes in a boiler is a common problem in the operation of a power plant, the operation safety and the economical efficiency of the boiler are seriously influenced, and a major accident is also caused in the serious case. The four pipes refer to a water wall pipe, a superheater pipe, a reheater pipe and an economizer pipe, and are places which are easy to explode at high temperature and high pressure in a power plant. Wherein the superheater pipe takes place that the booster is more than other water pipes, and the influence factor is many, and the problem is complicated, introduces machine learning technique and solves this problem, has important value and realistic meaning to thermal power plant's safe high-efficient operation.
At present, the prevention and control of the pipe explosion problem are mainly started from the mechanism aspect, for example, chemical analysis is carried out on the pipe explosion position, chemical components of the pipe wall corrosion position are detected, and then the influence of water on the pipe explosion problem is analyzed; or the microscopic metallographic structure at the pipe bursting position is checked, and the phenomenon that the pipe bursting position has local overtemperature is inferred; and analyzing the occurrence position of the pipe burst by a person to obtain that the pipe burst can be caused by overlarge local stress. According to the analyzed conclusion, the overhaul strength and quality are enhanced, and corresponding technical specifications are formulated.
Although starting from the mechanism aspect, a solution which is consistent with business logic and is clear and easy to understand can be obtained. However, the following problems also exist.
1 inspection and related regulations are ultimately performed by field workers, the role played by which depends largely on the performance of the field workers.
2 because the working condition of the boiler is bad, the equipment environment is complex, and the inspection work can be only carried out on partial equipment.
And 3, after the inspection and replacement work is finished, the real-time monitoring on the working condition of the equipment cannot be carried out. When the water pipe of the superheater has a deterioration trend, early warning cannot be given.
4 the protocol developed based on empirical summary lacks rigorous support on the data and is difficult to quantify accurately.
Disclosure of Invention
The invention refers to the measuring point data of the temperature and pressure of the boiler superheater as the superheater data for short, the continuous superheater data forms time sequence data, the number of the measuring points is 35, and the long-short term memory network is referred to as LSTM for short.
The technical scheme adopted by the invention for solving the technical problem is as follows.
And 3, constructing an LSTM model of superheater pipe explosion, wherein the long-term and short-term memory network is called LSTM for short.
Step 4, inputting the training set data processed in the step 2 into the LSTM model set in the step 3, and inputting a single training sample into a superheater data matrix; outputting a label value of the training sample; and obtaining the trained LSTM model.
Step 5, inputting the test set data processed in the step 2 into the LSTM model trained in the step 4, and inputting a single test sample into a superheater data matrix; outputting a label value of the training sample; and obtaining the well-trained LSTM model.
Step 6, taking superheater data with specified real-time length, processing the superheater data according to the step 2, and inputting the trained LSTM model to obtain a predicted label value, wherein the predicted label value is between 0 and 1, the closer the predicted label value is to 0, the smaller the probability of pipe explosion is, and the closer to 1, the greater the probability of pipe explosion is; and when the output prediction label value is larger than the set early warning threshold value, performing pipe explosion early warning on the superheater.
The specified real-time length is 30 minutes.
The T1 time period is greater than or equal to 60 minutes.
Further, the pretreatment in step 2 is specifically realized as follows.
2-1, carrying out normalization processing on the obtained original data matrix to obtain a normalization matrix of each label, wherein the normalization calculation formula is as follows;
wherein:
respectively the indices of the rows and columns in the original data matrix,
after expressing normalization
The matrix elements of the position are,
before expressing normalization
The matrix elements of the position are,
to normalize the smallest matrix element in the first j columns,
is normalized to the largest matrix element in the first j columns.
And 2-2, performing sliding sampling on the normalized matrix of the single label obtained in the step 2-1 by taking 30 minutes as a time period and taking 1 minute as a step length in a time dimension to form 30 sample data, namely each label can generate 30 samples, and the single sample is a superheater data matrix with the size of 30 x 35.
2-3, dividing all samples generated by all labels into a training set and a testing set, wherein the training set comprises 240 pipe bursting samples and 240 non-pipe bursting samples; the test set explodes 60 samples, and the test set explodes 60 samples.
Further, the sample of step 2-2 is as follows.
For a sample generated by a certain tube bursting label, the corresponding label value is marked as n/30 in the generation process, wherein
。
For a non-detonator label, the generated sample data label value is directly marked as 0.
Further, the LSTM model of superheater pipe explosion described in step 3 is specifically implemented as follows.
The LSTM model of the superheater pipe explosion comprises an input layer, a hidden layer and an output layer.
The input layer is used for inputting sample data, the sample data is input into the input layer according to the time sequence of the sample data generated by the label, the hidden layer is used for learning the relation between the input data and the output data, and the output layer outputs a result predicted by the model.
Further, the parameters of the LSTM model of superheater squibs are set as follows.
Input data length: 35, corresponding to 35 points of superheater data.
The length of the time series is 30, corresponding to the length of a single sample of 30 min.
Number of neurons in the hidden layer: 128.
the early warning threshold value is 20/30.
The invention has the following beneficial effects.
The creative key link of the invention is to perform matrixing and time serialization on superheater data, learn and prejudge the superheater pipe explosion problem by utilizing the learning and predicting capability of LSTM on time series data, and predict whether new superheater data is subjected to pipe explosion or not by learning the rule on the time dimension through LSTM.
The following advantages can be obtained through the early warning of the invention.
1 rate of accuracy is high, and this scheme carries out the early warning to the superheater pipe explosion problem with data drive's mode, and after data test stage and model come into operation, the early warning has been carried out to the superheater pipe explosion that the model is all accurate.
2, timeliness and comprehensiveness, compared with manual equipment troubleshooting, the model can monitor the operation of the whole superheater in real time and send out early warning to pipe explosion in advance.
3, the cost is low. The manual inspection of the superheater is not required to be stopped. The operation efficiency of the boiler is improved, and the personnel investment is reduced.
4, the operability is strong. The learning and early warning of the model can be automatically operated without the operation of field operators. The given early warning index is easy to understand and convenient for operators to judge.
Drawings
FIG. 1 is a schematic diagram of a single sample superheater data matrix generation in accordance with the present invention.
FIG. 2 is a block diagram of a single cell of the LSTM model of the invention.
FIG. 3 is a graph showing that the loss value of the LSTM model of the present invention approaches to be stable after 100 rounds of training.
FIG. 4 is a diagram illustrating that the accuracy of the LSTM model approaches 99.65% after 100 training rounds.
FIG. 5 is a graph of predicted values for a model collected every half hour in an embodiment of the present invention.
FIG. 6 is a flow chart of the present invention.
Detailed Description
The invention is further illustrated by the following figures and examples.
According to the invention, the measured point data of the temperature and the pressure of the boiler superheater are regarded as the matrixed time sequence data, and the rule of the input data on the time dimension is learned by utilizing the good learning and predicting capability of a Long Short-Term Memory network (LSTM) on the time sequence data, so that the analysis of the superheater pipe explosion problem based on data driving is realized, and the purpose of early warning the superheater pipe explosion is achieved.
The invention refers to the measuring point data of the temperature and pressure of the boiler superheater as the superheater data for short, the continuous superheater data forms time sequence data, the number of the measuring points is 35, and the long-short term memory network is referred to as LSTM for short.
Example 1.
As shown in fig. 6, a superheater explosion and leakage early warning method based on a long-short term memory network is specifically implemented as follows.
And 2, selecting superheater data from 74 minutes before to 15 minutes before the time point corresponding to the label of the tube explosion as original data for each label and the corresponding time in the step 1, and selecting superheater data from 59 minutes before to the corresponding time point according to the label of the tube explosion as original data. Each tag can generate a 59 x 35 size matrix of raw data, which is preprocessed as follows.
2-1, carrying out normalization processing on the obtained original data matrix to obtain a normalization matrix of each label, wherein the normalization calculation formula is as follows;
wherein:
respectively the indices of the rows and columns in the original data matrix,
after expressing normalization
The matrix elements of the position are,
before expressing normalization
The matrix elements of the position are,
to normalize the smallest matrix element in the first j columns,
is normalized to the largest matrix element in the first j columns.
2-2, performing sliding sampling on the normalized matrix of the single label obtained in the step 2-1 by taking 30 minutes as a time period and taking 1 minute as a step length in a time dimension to form 30 sample data, namely each label can generate 30 samples, and the single sample is a superheater data matrix with the size of 30 x 35, which is specifically shown in the following figure 1;
for a sample generated by a certain tube bursting label, the corresponding label value is marked as n/30 in the generation process, wherein
. For example, the tag generates a first sample with a tag value of 1/30, a second sample with a tag value of 2/30, and so on.
For a non-detonator label, the generated sample data label value is directly marked as 0.
This is intended to change the pipe bursting problem into a classification problem and then into a regression problem. For the pipe bursting problem, the closer to the pipe bursting time point, the greater the possibility of the pipe bursting problem occurring, and vice versa. It is obviously not appropriate to set the sample labels of the detonators to 1, so the labels are given different values in chronological order.
And 2-3, dividing all samples generated by all the labels into a training set and a testing set, wherein the training set is from 1 month in 2019 to 7 months in 2019, 240 samples with pipe bursting are obtained, and 240 samples without pipe bursting are obtained. The time of the test set is from 7 months in 2019 to 12 months in 2019, the number of the pipe bursting samples is 60, and the number of the non-pipe bursting samples is 60.
Step 3, constructing an LSTM model of superheater pipe explosion, which comprises the following steps: an input layer, a hidden layer, and an output layer.
The input layer is used for inputting sample data, the sample data is input into the input layer according to the time sequence of the sample data generated by the label, the hidden layer is used for learning the relation between the input data and the output data, and the output layer outputs a result predicted by the model.
Each cell (cell) of the hidden layer passes between the last cell and the next cell two states, one a long-term state c and one a short-term state h. LSTM is just using the long-term state c and the short-term state h to learn the regularity in the time dimension. Each unit contains four gates, namely a forgetting gate, an input gate, a status gate and an output gate.
And after the short-term state h transmitted at the last moment and the single sample data x input at the current moment are calculated by different matrixes W and activation functions carried by the hidden layer, four gates are formed.
Forget the door: determines the long-term state c of the last momentt− 1How much to keep the long-term state c to the current timetAnd t represents the current time t.
The input gate and the status gate together determine how much information is input at the current time to be added to the result of the forgetting gate.
An output gate: after the information processed by the first three gates is added and processed by the tanh activation function, the output gate determines how much information is passed to the next time as a short-term state.
The short-term state is processed by a matrix and sigmoid activation function as the output of the hidden layer.
The structure of a single cell of the LSTM model is shown in fig. 2.
The parameters to be set in the LSTM model are as follows.
Input data length: 35, corresponding to 35 points of superheater data.
The length of the time series is 30, corresponding to the length of a single sample of 30 min.
Number of neurons in the hidden layer: 128, the model is of sufficient complexity to guarantee learning ability, and 128 is a power of 7 of 2, facilitating training of the model.
And 4, inputting the training set data processed in the step 2 into the LSTM model set in the step 3, and inputting a single training sample into a 30 x 35-sized superheater data matrix, wherein 30 corresponds to the time sequence length, and 35 is the data length input for each unit. The output is the label value for that sample. The number of training rounds was set to 600 and the size of each batch of training was 12. And obtaining the trained LSTM model.
As can be seen from fig. 3 and 4, the loss value of the model after 100 rounds of training approaches to be stable, and the accuracy is about 99.65%.
And 5, inputting the test set data processed in the step 2 into the LSTM model trained in the step 4, wherein the input of a single test sample is a 30 x 35-sized superheater data matrix, wherein 30 corresponds to the time sequence length, and 35 is the input data length of each unit. The output is the label value for that sample. Comparing the output label value with the label value of the test sample correspondingly input in the test set data to obtain the error rate of the test set data; if the error rate of the test set data is beyond the range, the LSTM model is trained and optimized again.
And 6, taking superheater data with the real-time length of 30 minutes, processing the data according to the steps 2-1-2-3, and inputting the trained LSTM model to obtain a prediction result (namely a label value), wherein the value of the prediction result is between 0 and 1, the closer the prediction result is to 0, the smaller the probability of tube explosion, and the closer the prediction result is to 1, the greater the probability of tube explosion. And when the output prediction result value is larger than the set early warning threshold value, performing pipe explosion early warning on the superheater.
The early warning threshold value is 20/30.
The creative key link of the invention is to perform matrixing and time serialization on superheater data, learn and prejudge the superheater pipe explosion problem by utilizing the learning and predicting capability of LSTM on time series data, and predict whether new superheater data is subjected to pipe explosion or not by learning the rule on the time dimension through LSTM.
Example 2.
As shown in fig. 5, the predicted value of the model is collected every half hour from 12 noon 2/7 to 4 am 2/7 in the morning of 2020, the occurrence probability of the pipe explosion is gradually increased, the output value of the model is close to 0.8, and the case that the risk of pipe explosion is large is shown. The field worker finds that a plurality of water pipes on the right side of the superheater are overheated and have a rough expansion phenomenon in the subsequent one-time overhaul. The prediction result of the model is proved to have certain guiding significance for superheater pipe explosion.
The min-max normalization method can be converted into z-score normalization, and the common points are as follows: and the data dimension is unified, so that the small-dimension data is not phagocytized, and the convergence of the network is accelerated.
The term learning described in the present invention: in a machine learning task, a certain evaluation standard (precision) improves evaluation indexes in the task under continuous experience accumulation.
The term neuron as described in the present invention: in machine learning, neurons are the basic components of artificial neural networks, and these neurons can receive signals transmitted by n neurons and transmit the signals through weighted connections.
Claims (5)
1. A superheater explosion and leakage early warning method based on a long-term and short-term memory network is characterized in that measuring point data of temperature and pressure of a boiler superheater are abbreviated as superheater data, and continuous superheater data form time series data, and the method specifically comprises the following steps:
step 1, collecting all records of historical boiler superheater pipe explosion, randomly taking 10 records, and using each record as a label corresponding to a time point; simultaneously, randomly taking 10 normal operation records, wherein each record is used as a label and corresponds to a time point;
step 2, selecting superheater data of a T1 time period before the time point corresponding to the label of the pipe explosion as original data for each label and the corresponding time point in the step 1, and selecting superheater data of a T1 time period before the time point corresponding to the label of the pipe explosion as original data for the label of the pipe explosion; each label can obtain an original data matrix, and the original data is preprocessed; obtaining preprocessed training set data and test set data;
step 3, constructing an LSTM model of superheater pipe explosion, wherein the long-term and short-term memory network is abbreviated as LSTM;
step 4, inputting the training set data processed in the step 2 into the LSTM model set in the step 3, and inputting a single training sample into a superheater data matrix; outputting a label value of the training sample; obtaining a trained LSTM model;
step 5, inputting the test set data processed in the step 2 into the LSTM model trained in the step 4, and inputting a single test sample into a superheater data matrix; outputting a label value of the training sample; obtaining a trained LSTM model;
step 6, taking superheater data with specified real-time length, processing the superheater data according to the step 2, and inputting the trained LSTM model to obtain a predicted label value, wherein the predicted label value is between 0 and 1, the closer the predicted label value is to 0, the smaller the probability of pipe explosion is, the closer the predicted label value is to 1, the greater the probability of pipe explosion is; when the output predicted label value is larger than a set early warning threshold value, performing pipe explosion early warning on the superheater;
the pretreatment in the step 2 is specifically realized as follows:
2-1, carrying out normalization processing on the obtained original data matrix to obtain a normalization matrix of each label, wherein the normalization calculation formula is as follows:
wherein: i, j are the indices of the rows and columns, x, respectively, in the original data matrixijMatrix element, v, representing i, j position after normalizationijThe matrix elements representing the i, j positions before normalization,
to normalize the smallest matrix element in the first j columns,
is the maximum matrix element in the normalized front j columns;
2-2, performing sliding sampling on the normalized matrix of the single label obtained in the step 2-1 by taking 30 minutes as a time period and taking 1 minute as a step length in a time dimension to form 30 sample data, namely each label can generate 30 samples, and the single sample is a superheater data matrix with the size of 30 x 35;
2-3, dividing all samples generated by all labels into a training set and a testing set, wherein the training set comprises 240 pipe bursting samples and 240 non-pipe bursting samples; testing 60 detonation tube samples and 60 non-detonation tube samples;
the sample of the step 2-2 is specifically as follows:
for a sample generated by a certain tube bursting label, in the generation process, the corresponding label value is marked as n/30, wherein n is more than 0 and less than or equal to 30; the label value of the first sample generated by the label is marked as 1/30, the label value of the second sample is marked as 2/30, and the like;
for a non-detonator label, the generated sample data label value is directly marked as 0.
2. The superheater explosion and leakage early warning method based on the long and short term memory network as claimed in claim 1, wherein the LSTM mode of superheater pipe explosion in step 3 is implemented as follows:
the LSTM model of the thermal explosion tube comprises an input layer, a hidden layer and an output layer;
the input layer is used for inputting sample data, the sample data is input into the input layer according to the time sequence of the sample data generated by the label, the hidden layer is used for learning the relation between the input data and the output data, and the output layer outputs a result predicted by the model.
3. The superheater explosion and leakage early warning method based on the long and short term memory network as claimed in claim 2, wherein the LSTM model of the superheater explosion tube has the following parameter settings:
input data length: 35, corresponding to 35 measurement points of superheater data;
time series length 30, corresponding to a length of 30min for a single sample;
number of neurons in the hidden layer: 128.
4. the superheater explosion and leakage early warning method based on the long-short term memory network as claimed in claim 3, wherein the early warning threshold value is 20/30.
5. The superheater explosion and leakage early warning method based on the long-short term memory network as claimed in claim 4, wherein the time period T1 is greater than or equal to 60 minutes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010912762.4A CN111765449B (en) | 2020-09-03 | 2020-09-03 | Superheater explosion and leakage early warning method based on long-term and short-term memory network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010912762.4A CN111765449B (en) | 2020-09-03 | 2020-09-03 | Superheater explosion and leakage early warning method based on long-term and short-term memory network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111765449A CN111765449A (en) | 2020-10-13 |
| CN111765449B true CN111765449B (en) | 2020-12-29 |
Family
ID=72729205
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010912762.4A Active CN111765449B (en) | 2020-09-03 | 2020-09-03 | Superheater explosion and leakage early warning method based on long-term and short-term memory network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111765449B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| ES2946940A1 (en) * | 2022-01-28 | 2023-07-28 | Navarro Vicente Blazquez | GLOBAL DIAGNOSTIC SYSTEM OF STEAM AND CONDENSATE NETWORKS IN REAL TIME (Machine-translation by Google Translate, not legally binding) |
| CN117077839B (en) * | 2023-07-13 | 2024-04-12 | 华能国际电力股份有限公司上海石洞口第二电厂 | AM-BOA-LSTM-based wall temperature prediction method and system for superheater of coal-fired power plant |
| CN117558116A (en) * | 2024-01-11 | 2024-02-13 | 山东奥深智能工程有限公司 | Fire control early warning system based on big data |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108644752B (en) * | 2018-05-11 | 2019-09-13 | 中国神华能源股份有限公司 | For analyzing the method, apparatus and machine readable storage medium of four main tubes of boiler leakage |
| CN109521735B (en) * | 2018-10-25 | 2021-02-09 | 云达世纪（北京）科技有限公司 | On-line assessment method for use state risk of high-temperature heating surface of boiler |
| CN110263846B (en) * | 2019-06-18 | 2021-12-28 | 华北电力大学 | Fault diagnosis method based on fault data deep mining and learning |
| CN111340238B (en) * | 2020-03-12 | 2023-12-15 | 中南大学 | Fault diagnosis method, device, equipment and storage medium of industrial system |
| CN111426816A (en) * | 2020-04-10 | 2020-07-17 | 昆明理工大学 | Method for predicting concentration of dissolved gas in transformer oil based on PSO-L STM |
-
2020
- 2020-09-03 CN CN202010912762.4A patent/CN111765449B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN111765449A (en) | 2020-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111765449B (en) | Superheater explosion and leakage early warning method based on long-term and short-term memory network | |
| CN112749509B (en) | Intelligent substation fault diagnosis method based on LSTM neural network | |
| CN104732276B (en) | One kind metering production facility on-line fault diagnosis method | |
| CN110738274A (en) | nuclear power device fault diagnosis method based on data driving | |
| CN112817280A (en) | Implementation method for intelligent monitoring alarm system of thermal power plant | |
| CN112597696B (en) | Boiler four-pipe leakage early warning method based on extreme learning machine principle | |
| CN113339204B (en) | Wind driven generator fault identification method based on hybrid neural network | |
| CN111651933B (en) | Industrial boiler fault early warning method and system based on statistical inference | |
| CN108921230A (en) | Method for diagnosing faults based on class mean value core pivot element analysis and BP neural network | |
| CN114519382B (en) | Nuclear power plant key operation parameter extraction and abnormality monitoring method | |
| CN109492790A (en) | Wind turbines health control method based on neural network and data mining | |
| CN111860839A (en) | Shore bridge fault monitoring method based on multi-signal fusion and Adam optimization algorithm | |
| CN105930629A (en) | On-line fault diagnosis method based on massive amounts of operating data | |
| CN112149750A (en) | Water supply network pipe burst identification data driving method | |
| CN117556347A (en) | Power equipment fault prediction and health management method based on industrial big data | |
| CN118430092B (en) | Data general acquisition method based on MCC system | |
| CN112580858A (en) | Equipment parameter prediction analysis method and system | |
| Xu et al. | Anomaly detection with gru based bi-autoencoder for industrial multimode process | |
| CN111563685B (en) | Power generation equipment state early warning method based on auto-associative kernel regression algorithm | |
| CN116028849B (en) | Emulsion pump fault diagnosis method based on depth self-coding network | |
| CN114548701B (en) | Full-measurement-point-oriented coupling structure analysis and estimation process early warning method and system | |
| CN111815069B (en) | Satellite lithium battery voltage prediction method based on time series analysis principle | |
| Najar et al. | Comparative Machine Learning Study for Estimating Peak Cladding Temperature in AP1000 Under LOFW | |
| CN100440089C (en) | Industrial process nonlinear fault diagnosis system and method based on fisher | |
| Zhang et al. | Spatio-temporal fusion model of natural gas pipeline condition monitoring based on convolutional neural network and long short-term memory neural network |
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 |