CN110163433B - Ship flow prediction method - Google Patents
Ship flow prediction method Download PDFInfo
- Publication number
- CN110163433B CN110163433B CN201910423705.7A CN201910423705A CN110163433B CN 110163433 B CN110163433 B CN 110163433B CN 201910423705 A CN201910423705 A CN 201910423705A CN 110163433 B CN110163433 B CN 110163433B
- Authority
- CN
- China
- Prior art keywords
- neural network
- imf
- ship flow
- data
- independent
- 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
-
- 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
- G06F18/2134—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on separation criteria, e.g. independent component analysis
-
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Artificial Intelligence (AREA)
- Strategic Management (AREA)
- Evolutionary Computation (AREA)
- Human Resources & Organizations (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Game Theory and Decision Science (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a ship flow prediction method based on ensemble empirical mode decomposition and an independent circulation neural network. Firstly, in the aspect of data processing, a set empirical mode decomposition algorithm is adopted to decompose nonlinear non-stationary ship flow data into a series of high and low frequency eigenmode function sequences with stationarity and a monotonous remainder sequence, so that the information of an original sequence is retained to the maximum extent, the internal rules of the sequences are fully utilized, and the prediction precision is improved. And then, constructing a deep learning neural network by overlapping a plurality of hidden layers by using an independent circulation neural network, and fully extracting time hidden characteristic information of the ship flow in data training by combining a large amount of ship flow data so as to finish prediction. The invention uses the independent cyclic neural network with optimized parameters to separately process the decomposed ship flow data components, thereby improving the prediction precision while refining the data component processing; the invention has better self-adaptability.
Description
Technical Field
The invention relates to the technical field of time series prediction, in particular to a ship flow prediction method based on ensemble empirical mode decomposition and an independent cyclic neural network.
Background
The accuracy of the marine traffic flow prediction is improved, and a basic basis can be provided for planning, designing and navigation management of a channel. Traffic flow prediction is a rather complex process that is influenced by many factors, such as weather changes, economic indices, etc.
In past methods of traffic flow prediction, researchers often modeled based on some basic assumptions applied in the literature or selected traffic characteristics from the obtained data. That is, the correctness of these assumptions and the selected features may have a significant impact on the prediction accuracy. However, if statistical information and potential correlations hidden in the data are employed, such factors affecting vessel traffic flow will be revealed by the data set itself. Therefore, if this method is adopted, potential errors and mistakes caused by assumptions and empirical judgment can be avoided, and the accuracy of flow prediction can be improved by learning the information and correlation hidden in the data.
At present, the research on the ship traffic flow prediction at home and abroad is very rich, and the research mainly comprises an artificial neural network, wavelet analysis, a support vector machine, a long-term and short-term memory network and the like. The artificial neural network can well fit the nonlinear part of the ship flow sequence, wavelet analysis can select a proper basis function to predict the ship flow sequence, and the Markov model can well utilize the data correlation of the ship flow sequence, so that the prediction precision can be further improved. In addition, ship traffic prediction belongs to a time series problem, and currently, mainstream methods and models related to time series research are richer, such as a Convolutional Neural Network (CNN), a cyclic neural network (RNN) and various variants thereof. But have different effects for different fields. The convolutional neural network has good performance for processing data with spatial structure correlation such as two-dimensional data, and the cyclic neural network has the problems of gradient explosion and gradient disappearance when processing time sequence problems, and usually has the problem of processing by a long-short-term memory network (LSTM).
In view of a series of problems of low prediction precision, low adaptability, cyclic neural gradient explosion and the like in the prior art, the development of a ship flow prediction method based on ensemble empirical mode decomposition and an independent cyclic neural network, which can predict precision, is really necessary, and the adaptability problem of time series prediction for different time scales can be solved.
Disclosure of Invention
The invention aims to provide a ship flow prediction method based on ensemble empirical mode decomposition (EEMD-IndRNN), which adopts an independent circular neural network (IndRNN) model based on Ensemble Empirical Mode Decomposition (EEMD), belongs to the framework of a neural network method, can solve the problem of adaptability to time series prediction of different time scales, and simultaneously improves the prediction precision.
In order to achieve the purpose, the invention is realized by the following technical scheme:
a ship flow prediction method based on ensemble empirical mode decomposition and an independent cyclic neural network comprises the following steps: s1, preprocessing the original ship flow data; s2, performing stationarity verification on the original ship flow data; s3, performing ensemble empirical mode decomposition on the unstable original ship flow data obtained through verification in the step S2 to obtain a plurality of eigenmode functions and a remainder; and S4, predicting each decomposed eigenmode function and remainder by using an independent circulation neural network model respectively to obtain predicted value components of each independent circulation neural network, and superposing the predicted value components to obtain a ship flow prediction result.
Preferably, the step S2 further includes:
and carrying out stability verification on the time sequence of the original ship flow data by adopting an ADF (automatic document feeder) verification method, wherein when the time sequence of the original ship flow data is stable, a unit root does not exist, otherwise, the unit root exists.
Preferably, the step S3 further includes:
s31, white noise is added into the original ship flow data X (t) to obtain the ship flow data X added with the white noisen1(t) ═ x (t) + n (t), n (t) white noise;
s32, finding the original ship flow data Xn(t) fitting the upper envelope q1 and the lower envelope q2 by cubic spline interpolation, and taking the average value m (t) ═ q1+ q2)/2 to obtain a new sequence h1 ═ Xn1(t) -m (t); when the new sequence h1 has a positive minimum value or a negative maximum value, repeating the step S32 until the first eigenmode function IMF1 is found, and further obtaining new data Xn1(t)-IMF1；
S33, new data X according to the step S32n1(t) -IMF1, and new data Xn1(t) -IMF1 as the next cycleX in step S32n1(t), step S32 is executed in a loop until the original data X is processedn(t) is decomposed into m eigenmode functions and a monotonic remainder r (t), then:
Xn1(t)＝IMF1+IMF2+…+IMFm+r(t)
step S34, adding K times of different white noises into the original data X (t), repeating the steps S31-S33, and correspondingly obtaining m eigenmode functions and a monotonous remainder after the white noises are added each time;
step S35, performing ensemble averaging calculation on each decomposed component, as follows:
of formula (II) IMF'mIs the average value of the sum of the mth eigenmode functions obtained after adding white noise and decomposing K times, wherein i is the ith white noise; r' (t) is the average value of the sum of monotonic remainders r (t) obtained by adding white noise and decomposing K times;
in step S36, the decomposition result of the original ship flow data x (t) is:
X(t)＝IMF′1+IMF′2+…IMF′m+r′(t)。
preferably, the step S4 further includes:
s41, and IMF 'of each eigenmode function output in the step S3'1，IMF′2，…，IMF′mAnd the remainder r' (t) is used as the input of the independent recurrent neural network;
s42, changing the hidden layer state expression of the independent recurrent neural network into:
ht＝σ(WX′t+u⊙ht-1+b)
wherein t is time, X'tIs the input at time t, i.e. the above-mentioned eigenmode function IMF'1，IMF′2，…，IMF′mThe remainder r' (t); w is the weight between hidden layers, σ is the activation function of the neuron; u is the weight between the input layer and the hidden layer; an element product of a matrix is indicated, namely each hidden layer neuron only receives the output at the moment t and the self state at the moment t-1 as input;
s43, the output of the independent circulation neural network is:
Y(t)＝Vht+c
in the formula, V is a weight coefficient between the hidden layer and the output layer, and c is a threshold value;
step S44, outputting each predicted value component through each independent recurrent neural network, and superposing each predicted value component of each independent recurrent neural network to obtain a prediction result:
Y＝Y1+Y2+…+Ym+Yr
wherein, YmAnd YrAre predicted values of different ship flow components through the independent circulation neural network.
Compared with the prior art, the invention has the beneficial effects that: the ship traffic flow is influenced by factors such as seasons, climate and human activities in actual life to form irregular fluctuation, has the characteristics of non-stability and nonlinearity and brings great difficulty to prediction. And obtaining a series of stable high-low frequency components, wherein the high-frequency components represent the short-term change of the ship flow, and the low-frequency components represent the long-term change trend of the ship flow. If each component is separately and independently processed, the prediction precision can be greatly improved; because the problems of gradient explosion and gradient disappearance exist when the cyclic neural network processes the time sequence problem, the invention uses the independent cyclic neural network with proper parameters to process the decomposed ship traffic flow sequence, and can achieve good prediction effect while avoiding the problems; compared with the traditional judgment of the factors influencing the ship flow based on the subjective factors, the method has better self-adaptability.
Drawings
FIG. 1 is a flow chart of a ship flow prediction method based on ensemble empirical mode decomposition and an independent cyclic neural network according to the present invention;
FIG. 2 is a schematic diagram of a collective empirical mode decomposition method according to the present invention;
FIG. 3 is an exploded view of the collective empirical mode of the present invention shown in FIG. 2;
FIG. 4 is a schematic diagram of the independent recurrent neural network according to the present invention.
Detailed Description
The features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of the non-limiting embodiments made with reference to fig. 1-4. The present invention will be described in more detail below with reference to fig. 1-4, which illustrate embodiments of the present invention. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein.
The invention is mainly applied to predicting the ship traffic number of a certain port or a water area, and as shown in the combination of figures 1 to 4, the ship flow prediction method based on the ensemble empirical mode decomposition and the independent cyclic neural network comprises the following steps:
step S1, ship flow data preprocessing:
the ship flow data are influenced by subjective factors, objective factors and the like, so that some data are abnormal, and although the quantity is very small, the influence on the whole prediction model is very large. Therefore, it is necessary to pre-process the ship flow data x (t) (raw data), including deleting zero data and abnormal big data, and the ship flow data { x }1,x2,…，xnAnd represents the number of ships entering and exiting the port at time t.
Step S2, carrying out stationarity verification on the ship flow data:
wherein, an ADF (automatic bucket-filler, Augmented DF unit root) inspection method is adopted to carry out stability check on ship flow data X (t) (original data). If the time sequence of the data is stable, the unit root does not exist, otherwise, the unit root exists;
in step S2, the ADF primitive assumption is: if the time sequence has no unit root, the time sequence data is stable; if the time sequence has a unit root, namely non-stationary, for stationary time sequence data, the significance is required to be given on a given confidence level, and the original hypothesis is rejected; if the obtained statistic is significantly smaller than the critical statistic of 3 confidences (1%, 5%, 10%), the original hypothesis is rejected;
step S3, if it is found that the ship flow data is non-stationary after the verification in the step S2, performing ensemble empirical mode decomposition on the non-stationary ship flow data x (t) to subsequently improve the prediction accuracy, so as to obtain a series of stationary high and low frequency components. The high-frequency component represents the short-term change of the ship flow, and the low-frequency component represents the long-term change trend of the ship flow. Fig. 2 is a schematic diagram illustrating a principle of ensemble empirical mode decomposition of a data signal.
The step S3 further includes the following steps:
step S31, white noise is added into the ship flow data X (t) to obtain the ship flow data X added with the white noisen1(t) ═ x (t) + n (t), n (t) is white noise, so that data of different scales are automatically mapped to a proper reference scale to overcome the problem of modal aliasing defect of EMD (Empirical Mode Decomposition), thereby obtaining a better Decomposition result;
step S32, finding ship flow data Xn(t) fitting the upper envelope q1 and the lower envelope q2 by cubic spline interpolation, and taking the average value m (t) ═ q1+ q2)/2 to obtain a new sequence h1 ═ Xn1(t) -m (t). If the new sequence h1 has a positive minimum or a negative maximum, this step S32 is repeated until the first eigenmode is foundFunction IMF1 so that new data X can be obtainedn1(t)-IMF1；
Step S33, New data X according to step S32n1(t) -IMF1, and new data Xn1(t) -IMF1 as X in step S32 of the next cyclen1(t), step S32 is executed in a loop until the original data X is processedn(t) is decomposed into m eigenmode functions and a monotonic remainder r (t), then:
Xn1(t)＝IMF1+IMF2+…+IMFm+r(t) (1)
step S34, based on the above principle, the invention adds K times of different white noises to the original data X (t), and repeats the steps S31-S33 to correspondingly obtain m eigenmode functions and a monotonous data after decomposing each time of adding white noises.
In step S35, to eliminate the added noise, the ensemble average calculation is performed on each decomposed component as follows:
of formula (II) IMF'mIs the average value of the sum of the mth eigenmode functions obtained after adding white noise and decomposing K times, wherein i is the ith white noise; r' (t) is an average value of the sum of monotonous data r (t) obtained by adding white noise K times and decomposing.
Therefore, the decomposition result of the original ship flow data x (t) is:
X(t)＝IMF′1+IMF′2+…IMF′m+r′(t) (4)
in step S4, of the IMF components obtained in step S3, the high frequency component represents a short-term change in the ship flow rate, and the low frequency component represents a long-term change trend in the ship flow rate. Hence, the intrinsic mode function IMF 'to decomposition is required'1，IMF′2，…，IMF′mAnd the remainder r' (t) are respectively predicted by using independent recurrent neural network models optimized by different parameters, and the flow chart is shown in figure 1.
Fig. 4 is an Independent cyclic Neural network (Independent Neural Networks) model structure, a time sequence t represents ship flow data of different dates, ship data with a length of time _ step (assumed to be in a range of 1-10) is required to be input at each moment, data with the same length of time _ step (assumed to be in a range of 2-11) is output, and a model with optimized weight is obtained through massive data training, so that the purpose of prediction is achieved.
The independent recurrent neural network (IndRNN) prediction in step S4 of the present invention further comprises the following processes:
step S41, and IMF 'of each eigenmode function output in the step S3'1，IMF′2，…，IMF′mAnd the remainder r '(t) as input to the independent recurrent neural network, the different components (IMF'1，IMF′2，…，IMF′mThe remainder r '(t)) requires the use of IndRNN models of different parameters, and is input in the following step and is collectively denoted as X'tInstead of the intrinsic mode function IMF 'in step 3'1，IMF′2，…，IMF′mAnd remainder r' (t).
Step S42, the hidden layer state expression of the independent recurrent neural network is:
ht＝σ(WX′t+u⊙ht-1+b) (5)
wherein t is time, X'tIs input at time t, i.e. IMF 'as described above'1，IMF′2，…，IMF′mThe remainder r' (t), W is the weight between hidden layers, σ is the activation function of the neuron; u is the weight between the input layer and the hidden layer; an indication of a matrix element product, i.e. each hidden layer neuron accepts as input only the output at that moment at time t and the state itself at time t-1.
In the ship flow prediction, the weights W and u can be trained through an independent cyclic neural network model to extract characteristic information of data. Whereas a conventional RNN accepts as input at time t the state of all neurons at time t-1 for each neuron. Namely, the hidden layer state expression of the conventional RNN is:
ht＝σ(WX′t+Uht-1+b) (6)
as can be seen from the formula 4 and the formula 5, the independent circulation neural network is simplified on the connection of the hidden layer weight, the problems of gradient disappearance and gradient explosion can be effectively solved, the neural network is subjected to multilayer superposition, a deep learning network is constructed, more characteristic information is extracted from ship flow data, and therefore the ship flow prediction precision is more effectively improved.
Step S43, the output of the independent recurrent neural network is:
Y(t)＝Vht+c (7)
in the formula, V is a weight coefficient between the hidden layer and the output layer, and c is a threshold value;
step S44, obtaining predicted value components through the independent cyclic neural networks with different parameters, and superposing the predicted value components of the independent cyclic neural networks to obtain a prediction result:
Y＝Y1+Y2+…+Ym+Yr (8)
wherein, YmAnd YrAre predicted values of different ship flow components through the independent circulation neural network.
In summary, the ship flow data is preprocessed, and then stability verification is performed. And decomposing the data into a series of high and low frequency components by integrating empirical modes, and converting the components into a stable data sequence. The high-frequency component represents the short-term change of the ship flow, and the low-frequency component represents the long-term change trend of the ship flow. The processing of the ship flow data can greatly improve the prediction precision, and can overcome the problem that the ship traffic flow is influenced by a plurality of complex factors such as human and nature and the like and the great difficulty is brought to the prediction by non-stable and non-linear data. Meanwhile, the problems of gradient explosion and gradient disappearance exist when the cyclic neural network processes the time sequence problem, and the decomposed ship traffic flow sequence is processed by using the independent cyclic neural network with proper parameters, so that the problems can be avoided, and meanwhile, a multilayer neural network can be constructed, and a good prediction effect is achieved.
While the present invention has been described in detail with reference to the preferred embodiments, it should be understood that the above description should not be taken as limiting the invention. Various modifications and alterations to this invention will become apparent to those skilled in the art upon reading the foregoing description. Accordingly, the scope of the invention should be determined from the following claims.
Claims (2)
1. A ship flow prediction method based on ensemble empirical mode decomposition and an independent cyclic neural network is characterized by comprising the following steps:
s1, preprocessing the original ship flow data;
s2, performing stationarity verification on the original ship flow data;
s3, performing ensemble empirical mode decomposition on the unstable original ship flow data obtained through verification in the step S2 to obtain a plurality of eigenmode functions and a remainder;
s4, predicting each decomposed eigenmode function and remainder by using an independent cyclic neural network model respectively to obtain predicted value components of each independent cyclic neural network, and superposing the predicted value components to obtain a ship flow prediction result;
the step S3 further includes:
s31, white noise is added into the original ship flow data X (t) to obtain the ship flow data X added with the white noisen1(t) ═ x (t) + n (t), n (t) white noise;
s32, finding the original ship flow data Xn(t) fitting the upper envelope q1 and the lower envelope q2 by cubic spline interpolation, and taking the average value m (t) ═ q1+ q2)/2 to obtain a new sequence h1 ═ Xn1(t) -m (t); when the new sequence h1 has a positive minimum value or a negative maximum value, repeating the step S32 until the first eigenmode function IMF1 is found, and further obtaining new data Xn1(t)-IMF1；
S33, new data X according to the step S32n1(t) -IMF1, and new data Xn1(t) -IMF1 as X in step S32 of the next cyclen1(t), step S32 is executed in a loop until the original data X is processedn(t) is decomposed into m eigenmode functions and a monotonic remainder r (t), then:
Xn1(t)＝IMF1+IMF2+…+IMFm+r(t)
step S34, adding K times of different white noises into the original data X (t), repeating the steps S31-S33, and correspondingly obtaining m eigenmode functions and a monotonous remainder after the white noises are added each time;
step S35, performing ensemble averaging calculation on each decomposed component, as follows:
of formula (II) IMF'mIs the average value of the sum of the mth eigenmode functions obtained after adding white noise and decomposing K times, wherein i is the ith white noise; r' (t) is the average value of the sum of monotonic remainders r (t) obtained by adding white noise and decomposing K times;
in step S36, the decomposition result of the original ship flow data x (t) is:
X(t)＝IMF′1+IMF′2+…IMF′m+r′(t)；
the step S4 further includes:
s41, and IMF 'of each eigenmode function output in the step S3'1，IMF′2，…，IMF′mAnd the remainder r' (t) is used as the input of the independent recurrent neural network;
s42, changing the hidden layer state expression of the independent recurrent neural network into:
ht＝σ(WX′t+u⊙ht-1+b)
wherein t is time, X'tIs the input at time t, i.e. the above-mentioned eigenmode function IMF'1，IMF′2，…，IMF′mThe remainder r' (t); w is the weight between hidden layers, σ is the activation function of the neuron; u is the weight between the input layer and the hidden layer; an element product of a matrix is indicated, namely each hidden layer neuron only receives the output at the moment t and the self state at the moment t-1 as input;
s43, the output of the independent circulation neural network is:
Y(t)＝Vht+c
in the formula, V is a weight coefficient between the hidden layer and the output layer, and c is a threshold value;
step S44, outputting each predicted value component through each independent recurrent neural network, and superposing each predicted value component of each independent recurrent neural network to obtain a prediction result:
Y＝Y1+Y2+…+Ym+Yr
wherein, YmAnd YrAre predicted values of different ship flow components through the independent circulation neural network.
2. The ship flow prediction method based on ensemble empirical mode decomposition and independent recurrent neural network as claimed in claim 1,
the step S2 further includes:
and carrying out stability verification on the time sequence of the original ship flow data by adopting an ADF (automatic document feeder) verification method, wherein when the time sequence of the original ship flow data is stable, a unit root does not exist, otherwise, the unit root exists.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910423705.7A CN110163433B (en) | 2019-05-21 | 2019-05-21 | Ship flow prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910423705.7A CN110163433B (en) | 2019-05-21 | 2019-05-21 | Ship flow prediction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110163433A CN110163433A (en) | 2019-08-23 |
| CN110163433B true CN110163433B (en) | 2021-10-26 |
Family
ID=67631619
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910423705.7A Active CN110163433B (en) | 2019-05-21 | 2019-05-21 | Ship flow prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110163433B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110852527B (en) * | 2019-11-20 | 2022-05-31 | 成都理工大学 | Reservoir physical property parameter prediction method combining deep learning |
| CN110763830B (en) * | 2019-12-04 | 2022-04-05 | 济南大学 | Method for predicting content of free calcium oxide in cement clinker |
| CN111027775A (en) * | 2019-12-12 | 2020-04-17 | 中国长江三峡集团有限公司 | Step hydropower station generating capacity prediction method based on long-term and short-term memory network |
| CN111241466B (en) * | 2020-01-15 | 2023-10-03 | 上海海事大学 | Ship flow prediction method based on deep learning |
| CN111415008B (en) * | 2020-03-17 | 2023-03-24 | 上海海事大学 | Ship flow prediction method based on VMD-FOA-GRNN |
| CN111353828B (en) * | 2020-03-30 | 2023-09-12 | 中国工商银行股份有限公司 | Method and device for predicting number of people coming to store at website |
| CN111507530B (en) * | 2020-04-17 | 2022-05-31 | 集美大学 | RBF neural network ship traffic flow prediction method based on fractional order momentum gradient descent |
| CN111897851A (en) * | 2020-07-01 | 2020-11-06 | 中国建设银行股份有限公司 | Abnormal data determination method and device, electronic equipment and readable storage medium |
| CN112052623A (en) * | 2020-08-13 | 2020-12-08 | 哈尔滨工业大学 | Short-term ship attitude prediction method based on empirical mode decomposition and support vector regression |
| CN112949932B (en) * | 2021-03-18 | 2024-02-02 | 自然资源部第二海洋研究所 | Ship traffic flow prediction method, device, computer equipment and storage medium |
| CN115828736B (en) * | 2022-11-10 | 2024-08-30 | 大连海事大学 | EEMD-PE-LSTM-based short-time ship traffic flow prediction method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160125297A1 (en) * | 2014-10-30 | 2016-05-05 | Umm Al-Qura University | System and method for solving spatiotemporal-based problems |
| CN106126896A (en) * | 2016-06-20 | 2016-11-16 | 中国地质大学（武汉） | The mixed model wind speed forecasting method learnt based on empirical mode decomposition and the degree of depth and system |
| CN106295798A (en) * | 2016-08-29 | 2017-01-04 | 江苏省电力试验研究院有限公司 | Empirical mode decomposition and Elman neural network ensemble wind-powered electricity generation Forecasting Methodology |
| CN107292453A (en) * | 2017-07-24 | 2017-10-24 | 国网江苏省电力公司电力科学研究院 | A kind of short-term wind power prediction method based on integrated empirical mode decomposition Yu depth belief network |
| CN108564046A (en) * | 2018-04-19 | 2018-09-21 | 南京大学 | Based on the steel construction dynamic strain signal processing method for improving EEMD |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108985515B (en) * | 2018-07-24 | 2021-11-26 | 国网河南省电力公司电力科学研究院 | New energy output prediction method and system based on independent cyclic neural network |
-
2019
- 2019-05-21 CN CN201910423705.7A patent/CN110163433B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160125297A1 (en) * | 2014-10-30 | 2016-05-05 | Umm Al-Qura University | System and method for solving spatiotemporal-based problems |
| CN106126896A (en) * | 2016-06-20 | 2016-11-16 | 中国地质大学（武汉） | The mixed model wind speed forecasting method learnt based on empirical mode decomposition and the degree of depth and system |
| CN106295798A (en) * | 2016-08-29 | 2017-01-04 | 江苏省电力试验研究院有限公司 | Empirical mode decomposition and Elman neural network ensemble wind-powered electricity generation Forecasting Methodology |
| CN107292453A (en) * | 2017-07-24 | 2017-10-24 | 国网江苏省电力公司电力科学研究院 | A kind of short-term wind power prediction method based on integrated empirical mode decomposition Yu depth belief network |
| CN108564046A (en) * | 2018-04-19 | 2018-09-21 | 南京大学 | Based on the steel construction dynamic strain signal processing method for improving EEMD |
Non-Patent Citations (2)
| Title |
|---|
| 基于集合经验模态分解和灰色神经网络的船舶交通流预测;肖进丽等;《武汉理工大学学报(交通科学与工程版)》;20171215;第41卷(第6期);第938-942页 * |
| 肖进丽等.基于集合经验模态分解和灰色神经网络的船舶交通流预测.《武汉理工大学学报(交通科学与工程版)》.2017,第41卷(第6期),第938-942页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110163433A (en) | 2019-08-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110163433B (en) | Ship flow prediction method | |
| CN110147450B (en) | Knowledge complementing method and device for knowledge graph | |
| CN109492808B (en) | Method for predicting remaining parking spaces of indoor parking lot | |
| CN111563706A (en) | Multivariable logistics freight volume prediction method based on LSTM network | |
| CN107544904B (en) | Software reliability prediction method based on deep CG-LSTM neural network | |
| CN111324990A (en) | Porosity prediction method based on multilayer long-short term memory neural network model | |
| CN111063194A (en) | Traffic flow prediction method | |
| CN110222826A (en) | One kind being based on improved EEMD-IndRNN ship method for predicting | |
| Leite et al. | Fuzzy granular evolving modeling for time series prediction | |
| CN113392961A (en) | Method for extracting mesoscale eddy track stable sequence and predicting cyclic neural network | |
| CN112949932B (en) | Ship traffic flow prediction method, device, computer equipment and storage medium | |
| CN110956309A (en) | Flow activity prediction method based on CRF and LSTM | |
| CN116822920B (en) | Flow prediction method based on cyclic neural network | |
| CN111191823A (en) | Production logistics prediction method based on deep learning | |
| CN115185937A (en) | SA-GAN architecture-based time sequence anomaly detection method | |
| CN114139893A (en) | Carbon emission influence factor analysis and index evaluation method and device | |
| CN116626753A (en) | Microseism event identification method and system based on multi-modal neural network | |
| CN109376937B (en) | Self-adaptive scheduling end-of-term water level prediction method based on ensemble empirical mode decomposition | |
| CN113035348A (en) | Diabetes diagnosis method based on GRU feature fusion | |
| CN116757309A (en) | Gas load prediction method and device based on long-short-period time sequence network | |
| CN117009758A (en) | Oil well yield prediction method based on PSO-VMD-LSTM | |
| CN114942480A (en) | Ocean station wind speed forecasting method based on information perception attention dynamic cooperative network | |
| US20220277200A1 (en) | Fast quantised training of trainable modules | |
| CN108417204A (en) | Information security processing method based on big data | |
| CN116405368B (en) | Network fault diagnosis method and system under high-dimensional unbalanced data condition |
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 |