CN112559827A - Measurement parameter prediction and sewage treatment control method based on deep learning - Google Patents
Measurement parameter prediction and sewage treatment control method based on deep learning Download PDFInfo
- Publication number
- CN112559827A CN112559827A CN202011421250.4A CN202011421250A CN112559827A CN 112559827 A CN112559827 A CN 112559827A CN 202011421250 A CN202011421250 A CN 202011421250A CN 112559827 A CN112559827 A CN 112559827A
- Authority
- CN
- China
- Prior art keywords
- prediction
- level
- frequency components
- sewage treatment
- 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
- 238000005259 measurement Methods 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 28
- 239000010865 sewage Substances 0.000 title claims abstract description 24
- 238000013135 deep learning Methods 0.000 title claims abstract description 11
- 238000012545 processing Methods 0.000 claims abstract description 17
- 238000003062 neural network model Methods 0.000 claims abstract description 11
- 230000009466 transformation Effects 0.000 claims abstract description 9
- 230000009471 action Effects 0.000 claims abstract description 5
- 238000004140 cleaning Methods 0.000 claims abstract description 5
- 238000013528 artificial neural network Methods 0.000 claims description 18
- 238000010606 normalization Methods 0.000 claims description 11
- 230000002159 abnormal effect Effects 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 150000001875 compounds Chemical class 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims description 2
- 238000000354 decomposition reaction Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 210000002569 neuron Anatomy 0.000 description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000004886 process control Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/9035—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/90335—Query processing
- G06F16/90348—Query processing by searching ordered data, e.g. alpha-numerically ordered data
Abstract
The invention relates to a measurement parameter prediction and sewage treatment control method based on deep learning, which comprises the following steps: step 1: acquiring real-time sewage treatment online instrument measurement time sequence data; step 2: performing data cleaning and wavelet transformation on the time sequence data to obtain high-frequency components of each level; and step 3: respectively processing each level of high-frequency components, inputting the processed high-frequency components into a trained GRU neural network model, and outputting a prediction result corresponding to each level of high-frequency components; and 4, step 4: the prediction results corresponding to the high-frequency components of each level are processed again and then combined for wavelet reconstruction, and measurement parameters subjected to prediction processing are obtained; and 5: and performing corresponding control action on the corresponding control link of the sewage treatment according to the measured parameters subjected to the prediction treatment. Compared with the prior art, the method has the advantages of ensuring relatively accurate pre-estimated parameters, avoiding control lag, reducing the measurement frequency, reducing the cost and the like.
Description
Technical Field
The invention relates to the technical field of sewage treatment measurement control, in particular to a measurement parameter prediction and sewage treatment control method based on deep learning.
Background
With the wave of "upgrading" raised in recent years, in order to meet the increasingly strict discharge standard, an online instrument is gradually added for the sewage treatment with relatively low automation degree originally.
However, the existing online sewage treatment instrument has the characteristics of long measurement time and high measurement and maintenance cost (measurement needs to consume measurement medicines, and the measurement cost is increased along with the increase of the measurement frequency), so that the operation cost is increased, and unstable water outlet is caused by relative delay of control measures due to overlong measurement time.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a measurement parameter prediction and sewage treatment control method based on deep learning.
The purpose of the invention can be realized by the following technical scheme:
a measurement parameter prediction and sewage treatment control method based on deep learning comprises the following steps:
step 1: acquiring real-time sewage treatment online instrument measurement time sequence data;
step 2: performing data cleaning and wavelet transformation on the time sequence data to obtain high-frequency components of each level;
and step 3: respectively processing each level of high-frequency components, inputting the processed high-frequency components into a trained GRU neural network model, and outputting a prediction result corresponding to each level of high-frequency components;
and 4, step 4: the prediction results corresponding to the high-frequency components of each level are processed again and then combined for wavelet reconstruction, and measurement parameters subjected to prediction processing are obtained;
and 5: and performing corresponding control action on the corresponding control link of the sewage treatment according to the measured parameters subjected to the prediction treatment.
Further, the step 2 comprises the following sub-steps:
step 201: performing data primary selection on the time sequence data based on normal distribution to obtain alternative abnormal data;
step 202: accurately screening the time series data subjected to data primary selection based on difference to screen out abnormal values;
step 203: after removing the abnormal value, filling the original time series data by adopting a front-back average method;
step 204: and performing wavelet transformation on the time sequence data subjected to data filling to obtain high-frequency components of each level.
Further, the normal distribution in step 201 is described by the formula:
wherein σ is the standard deviation, σ2Is the variance and μ is the mean.
Further, the corresponding description formula of the wavelet transform in step 204 is as follows:
in the formula, τ is the translation amount, and a is the scale.
Further, the step 3 specifically includes: and respectively carrying out normalization processing on each level of high-frequency component, inputting the normalized high-frequency component into the trained GRU neural model, and outputting a prediction result corresponding to each level of high-frequency component.
Further, the normalization process is described by the corresponding description formula:
in the formula (I), the compound is shown in the specification,
to normalize the processed sample data, σjIs the deviation of the jth sample component, DijIs the jth sample component of the ith sample, and m is the number of samples of the jth sample component.
Further, the GRU neural network model in step 3 includes a first GRU neural network layer, a first Dropout layer, a second GRU neural network layer, a second Dropout layer, and a sense neural network layer, which are connected in sequence.
Further, the step 4 specifically includes: and performing data inverse normalization processing on the prediction results corresponding to the high-frequency components of each level again, and then combining the prediction results to perform wavelet reconstruction to obtain the measurement parameters subjected to prediction processing.
Further, the corresponding description formula of the GRU neural network model is as follows:
zt＝σ(WZ·[ht-1,xt])
rt＝σ(Wr·[ht-1,xt])
in the formula, ztTo refresh the door, rtTo reset the gate, xtFor the input of a variable, h, at the present momentt-1Is a hidden layer of the previous sequence,
as a candidate hidden layer, htHidden layer for current sequence, WZFor updating the corresponding weight parameter of the gate, WrIn order to reset the weight parameter corresponding to the gate, W is the weight parameter corresponding to the candidate hidden layer.
Furthermore, the GRU neural network model takes Keras as a modeling environment, the loss function adopts root mean square error, and model training is completed through an adam optimization algorithm.
Compared with the prior art, the invention has the following advantages:
(1) the method of the invention comprises the following steps: step 1: acquiring real-time sewage treatment online instrument measurement time sequence data; step 2: performing data cleaning and wavelet transformation on the time sequence data to obtain high-frequency components of each level; and step 3: respectively processing each level of high-frequency components, inputting the processed high-frequency components into a trained GRU neural network model, and outputting a prediction result corresponding to each level of high-frequency components; and 4, step 4: the prediction results corresponding to the high-frequency components of each level are processed again and then combined for wavelet reconstruction, and measurement parameters subjected to prediction processing are obtained; and 5: corresponding control actions are carried out on corresponding control links of sewage treatment according to the measured parameters subjected to prediction treatment, so that the parameters can be accurately predicted in advance, control lag is avoided, the measurement frequency can be reduced, and the cost is reduced.
(2) The control method can accurately pre-estimate the relevant parameters in advance, effectively avoids the phenomenon of unstable effluent quality caused by long measurement time and lag of control measures of the instrument, and reduces energy consumption.
(3) The control method of the invention can reduce the measurement frequency of a certain instrument and reduce the measurement cost.
(4) The model adopted in the control method can be quickly corrected according to the measured data, so that the accuracy is ensured.
Drawings
FIG. 1 is a flow chart of the method steps of the present invention;
FIG. 2 is a diagram illustrating a wavelet decomposition tree structure according to the present invention;
fig. 3 is a schematic diagram of raw COD online meter data (5min x 500 pieces) in an embodiment of the present invention;
FIG. 4 is a diagram illustrating a two-level low-frequency component after wavelet transform in an embodiment of the present invention;
FIG. 5 is a diagram illustrating a first-level high-frequency component after wavelet transform according to an embodiment of the present invention;
FIG. 6 is a diagram illustrating two-level high-frequency components after wavelet transform according to an embodiment of the present invention;
fig. 7 is a comparison diagram of the reconstructed parameter sequence and the original parameter sequence in the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, shall fall within the scope of protection of the present invention.
DETAILED DESCRIPTION OF EMBODIMENT (S) OF INVENTION
Fig. 1 shows a measurement parameter prediction and sewage treatment control method based on deep learning, which comprises the following steps:
step 1: acquiring real-time sewage treatment online instrument measurement time sequence data;
step 2: performing data cleaning and wavelet transformation on the time sequence data to obtain high-frequency components of each level;
and step 3: respectively processing each level of high-frequency components, inputting the processed high-frequency components into a trained GRU neural network model, and outputting a prediction result corresponding to each level of high-frequency components;
and 4, step 4: the prediction results corresponding to the high-frequency components of each level are processed again and then combined for wavelet reconstruction, and measurement parameters subjected to prediction processing are obtained;
and 5: and performing corresponding control action on the corresponding control link of the sewage treatment according to the measured parameters subjected to the prediction treatment.
The steps correspond to the following specific processes:
1. detecting and processing time series data abnormity;
1) initially selecting based on 3 sigma-rule:
normal distribution formula:
wherein σ is the standard deviation, σ2Is the variance and μ is the mean.
The probability in the interval (μ -3 σ, μ +3 σ) was 99.74. Therefore, when the data distribution interval exceeds this interval, it can be considered as alternative abnormal data.
2) And (3) accurate screening based on difference: and screening the abnormal values determined by the fact that the first-order difference exceeds the threshold value continuously by using the box line graph.
3) Outlier removal is filled with the pre-and post-mean values.
2. Wavelet transformation extraction features:
1) wavelet transformation thought:
the data sequence is subjected to wavelet decomposition, and the result of each layer of decomposition is that the low-frequency signal obtained by the last decomposition is decomposed into two parts, namely a low-frequency part and a high-frequency part. After the N-layer decomposition, the source signal X is decomposed into: x ═ cD1+ cD2+ ·+ cDn + cAn where cD1, cD2,. cDn are high frequency signals decomposed by the first layer, the second layer, and the n-th layer, and cAn is a low frequency signal decomposed by the n-th layer, respectively.
The wavelet decomposition tree is shown in figure 2.
2) Principle formula of wavelet transform:
in the formula, τ is the translation amount, and a is the scale.
Selecting historical data of COD on-line instruments of Ningbo water plant as an example for application:
the original COD on-line meter data (5min × 500 pieces), the wavelet transformed secondary low frequency component, the wavelet transformed primary high frequency component, and the wavelet transformed secondary high frequency component are shown in fig. 3, fig. 4, fig. 5, and fig. 6, respectively.
3. Data normalization
The data normalization formula is as follows:
in the formula (I), the compound is shown in the specification,
to normalize the processed sample data, σjIs the deviation of the jth sample component, DijIs the jth sample component of the ith sample, and m is of the jth sample componentNumber of samples.
4. GRU neural network prediction
1) Summary of the principles of the GRU neural network
zt＝σ(WZ·[ht-1,xt])
rt＝σ(Wr·[ht-1,xt])
In the formula, ztTo refresh the door, rtTo reset the gate, xtFor the input of a variable, h, at the present momentt-1Is a hidden layer of the previous sequence,
as a candidate hidden layer, htHidden layer for current sequence, WZFor updating the corresponding weight parameter of the gate, WrIn order to reset the weight parameter corresponding to the gate, W is the weight parameter corresponding to the candidate hidden layer.
rtFor controlling how much of the previous memory needs to be retained if rtIs 0, then it represents
Only the input information of the current sequence is reserved; last ztControl requires a hidden layer h from the previous sequencet-1How much information is forgotten in the middle and how much hidden layer information of the current sequence needs to be added
Thereby obtaining the output hidden layer information h of the current sequencet。
The neural network structure of the prediction model comprises a first GRU neural network, a first Dropout layer, a second GRU neural network, a second Dropout layer and a third Dense neural network which are sequentially connected. In the embodiment, Keras is used as a modeling environment to model and learn and train the GRU neural network; during training, the network layer of the GRU neural network maintains a state between fixed line numbers of data in the training data set in which the GRU neural network operates before updating the network weights. And determining the emptying time of the state of the GRU neural network layer by adopting a reset _ states function, compiling the GRU neural network by adopting a mean-square error loss function by adopting a loss function, and finishing the training of the GRU neural network by using an adam optimization algorithm.
The Dropout layer forwards propagates a random value (0-1) of each neuron, sets the neuron value of which the random value is smaller than a set threshold value and corresponding to the neuron to be 0, sets the neuron value of which the random value is larger than the set threshold value and corresponding to the neuron to be 2 multiplied by an original value, stores the random values corresponding to all the neurons, and uses the random values in backward propagation, so that overfitting can be prevented, and the generalization capability of the model is improved.
5. Data inverse normalization
The inverse normalization must utilize a corresponding normalization model, requiring the input data to be of the same dimension.
6. Performing wavelet data reconstruction
The comparison of the original parameter sequence after reconstruction with the predicted sequence is shown in fig. 7.
7. Predicting step time intervals as measurement frequencies
The model prediction step length is used as the measurement frequency, and the process control and regulation at the middle moment can refer to the predicted value.
While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A measurement parameter prediction and sewage treatment control method based on deep learning is characterized by comprising the following steps:
step 1: acquiring real-time sewage treatment online instrument measurement time sequence data;
step 2: performing data cleaning and wavelet transformation on the time sequence data to obtain high-frequency components of each level;
and step 3: respectively processing each level of high-frequency components, inputting the processed high-frequency components into a trained GRU neural network model, and outputting a prediction result corresponding to each level of high-frequency components;
and 4, step 4: the prediction results corresponding to the high-frequency components of each level are processed again and then combined for wavelet reconstruction, and measurement parameters subjected to prediction processing are obtained;
and 5: and performing corresponding control action on the corresponding control link of the sewage treatment according to the measured parameters subjected to the prediction treatment.
2. The method for prediction of measurement parameters and sewage treatment control based on deep learning of claim 1, wherein the step 2 comprises the following sub-steps:
step 201: performing data primary selection on the time sequence data based on normal distribution to obtain alternative abnormal data;
step 202: accurately screening the time series data subjected to data primary selection based on difference to screen out abnormal values;
step 203: after removing the abnormal value, filling the original time series data by adopting a front-back average method;
step 204: and performing wavelet transformation on the time sequence data subjected to data filling to obtain high-frequency components of each level.
3. The method according to claim 2, wherein the normal distribution corresponding description formula in step 201 is as follows:
wherein σ is the standard deviation, σ2Is the variance and μ is the mean.
4. The method according to claim 2, wherein the wavelet transform corresponding description formula in step 204 is as follows:
in the formula, τ is the translation amount, and a is the scale.
5. The method for prediction of measurement parameters and sewage treatment control based on deep learning according to claim 1, wherein the step 3 specifically comprises: and respectively carrying out normalization processing on each level of high-frequency component, inputting the normalized high-frequency component into the trained GRU neural model, and outputting a prediction result corresponding to each level of high-frequency component.
6. The method of claim 5, wherein the normalization process is performed according to a description formula:
in the formula (I), the compound is shown in the specification,
to normalize the processed sample data, σjIs the deviation of the jth sample component, DijIs the jth sample component of the ith sample, and m is the number of samples of the jth sample component.
7. The method as claimed in claim 1, wherein the GRU neural network model in step 3 comprises a first GRU neural network layer, a first Dropout layer, a second GRU neural network layer, a second Dropout layer and a sense neural network layer, which are connected in sequence.
8. The method for prediction of measurement parameters and sewage treatment control based on deep learning according to claim 1, wherein the step 4 specifically comprises: and performing data inverse normalization processing on the prediction results corresponding to the high-frequency components of each level again, and then combining the prediction results to perform wavelet reconstruction to obtain the measurement parameters subjected to prediction processing.
9. The method of claim 7, wherein the GRU neural network model is described by the following formula:
zt＝σ(WZ·[ht-1,xt])
rt＝σ(Wr·[ht-1,xt])
in the formula, ztTo refresh the door, rtTo reset the gate, xtFor the input of a variable, h, at the present momentt-1Is a hidden layer of the previous sequence,
as a candidate hidden layer, htHidden layer for current sequence, WZFor updating the corresponding weight parameter of the gate, WrIn order to reset the weight parameter corresponding to the gate, W is the weight parameter corresponding to the candidate hidden layer.
10. The method of claim 7, wherein the GRU neural network model is modeled in a Keras environment, the loss function is root mean square error, and model training is completed through an adam optimization algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011421250.4A CN112559827A (en) | 2020-12-08 | 2020-12-08 | Measurement parameter prediction and sewage treatment control method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011421250.4A CN112559827A (en) | 2020-12-08 | 2020-12-08 | Measurement parameter prediction and sewage treatment control method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112559827A true CN112559827A (en) | 2021-03-26 |
Family
ID=75059501
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011421250.4A Pending CN112559827A (en) | 2020-12-08 | 2020-12-08 | Measurement parameter prediction and sewage treatment control method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112559827A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102867132A (en) * | 2012-10-16 | 2013-01-09 | 南京航空航天大学 | Aviation direct-current converter online fault combined prediction method based on fractional order wavelet transformation |
| CN108428023A (en) * | 2018-05-24 | 2018-08-21 | 四川大学 | Trend forecasting method based on quantum Weighted Threshold repetitive unit neural network |
| CN109190184A (en) * | 2018-08-09 | 2019-01-11 | 天津大学 | A kind of heating system historical data preprocess method |
| CN109376848A (en) * | 2018-09-01 | 2019-02-22 | 哈尔滨工程大学 | A kind of door control unit neural network of simplification |
| CN110738010A (en) * | 2019-10-17 | 2020-01-31 | 湖南科技大学 | Wind power plant short-term wind speed prediction method integrated with deep learning model |
| CN111242377A (en) * | 2020-01-15 | 2020-06-05 | 重庆大学 | Short-term wind speed prediction method integrating deep learning and data denoising |
| CN111709588A (en) * | 2020-06-23 | 2020-09-25 | 苏州联电能源发展有限公司 | Power consumption prediction method and system |
-
2020
- 2020-12-08 CN CN202011421250.4A patent/CN112559827A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102867132A (en) * | 2012-10-16 | 2013-01-09 | 南京航空航天大学 | Aviation direct-current converter online fault combined prediction method based on fractional order wavelet transformation |
| CN108428023A (en) * | 2018-05-24 | 2018-08-21 | 四川大学 | Trend forecasting method based on quantum Weighted Threshold repetitive unit neural network |
| CN109190184A (en) * | 2018-08-09 | 2019-01-11 | 天津大学 | A kind of heating system historical data preprocess method |
| CN109376848A (en) * | 2018-09-01 | 2019-02-22 | 哈尔滨工程大学 | A kind of door control unit neural network of simplification |
| CN110738010A (en) * | 2019-10-17 | 2020-01-31 | 湖南科技大学 | Wind power plant short-term wind speed prediction method integrated with deep learning model |
| CN111242377A (en) * | 2020-01-15 | 2020-06-05 | 重庆大学 | Short-term wind speed prediction method integrating deep learning and data denoising |
| CN111709588A (en) * | 2020-06-23 | 2020-09-25 | 苏州联电能源发展有限公司 | Power consumption prediction method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112364975B (en) | Terminal running state prediction method and system based on graph neural network | |
| RU2459254C2 (en) | Method for computer-aided training of one or more neural networks | |
| Liu et al. | Intelligent prognostics of machining tools based on adaptive variational mode decomposition and deep learning method with attention mechanism | |
| Slade et al. | Bimodal distribution removal | |
| CN112488415A (en) | Power load prediction method based on empirical mode decomposition and long-and-short-term memory network | |
| CN112468326B (en) | Access flow prediction method based on time convolution neural network | |
| CN107292446B (en) | Hybrid wind speed prediction method based on component relevance wavelet decomposition | |
| CN112000084B (en) | Intelligent BIT design method of controller module based on 1D-CNN and GRU-SVM | |
| CN116861201B (en) | Power supply system based on artificial intelligence | |
| CN111784061B (en) | Training method, device and equipment for power grid engineering cost prediction model | |
| CN113065703A (en) | Time series prediction method combining multiple models | |
| CN111931983A (en) | Precipitation prediction method and system | |
| CN114298134A (en) | Wind power prediction method and device and electronic equipment | |
| CN108984851B (en) | Weighted Gaussian model soft measurement modeling method with time delay estimation | |
| CN112559827A (en) | Measurement parameter prediction and sewage treatment control method based on deep learning | |
| CN116933643A (en) | Intelligent data monitoring method based on partial robust M regression and multiple interpolation | |
| Kotenko et al. | Formation of Indicators for Assessing Technical Reliability of Information Security Systems | |
| Su et al. | Self-organizing pipelined recurrent wavelet neural network for time series prediction | |
| CN115389743A (en) | Method, medium and system for predicting content interval of dissolved gas in transformer oil | |
| Niyogi | A novel method combines moving fronts, data decomposition and deep learning to forecast intricate time series | |
| Kalinina et al. | Construction of Forecast Models based on Bayesian Structural Time Series | |
| Xie | Application Model of Public Health Visual Art Creation Concept Based on Artificial Intelligence Technology in Venice Biennale | |
| CN117195954A (en) | Macroscopic economic prediction method and macroscopic economic prediction system based on transfer learning and neural network | |
| Noyes | Training and generalization | |
| Hu et al. | A baseline similarity attention-based dual-channel feature fusion network for machine remaining useful life prediction |
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 |