Prediction of stream-flow by using Artificial Neural Network model with special reference to pre-processing of raw data.
The model is based on daily stream-flow records of many years.
1. Stream flow Forecasting
by
ANN Modeling with
Preprocessing Techniques
for
Time Series Data
PROPOSAL FOR Ph.D. Thesis at N.I.T.K.
,Surathkal,
By Aniruddha Banhatti,
Part Time Ph.D. Student,
Registration Number: AM08P05
2. Importance of Stream Flow
Forecasting
Hydrologic Structures
Irrigation
Flood Control
Hydrologic Planning
Flood Relief
3. Nature of Stream flow Data
Time Series Data
Show following characteristics:
Trend
Seasonality
Cyclic Nature
Irregular Fluctuations – Outliers and
Noise
4. Basics of Artificial Neural
Networks
ANN is a massively parallel information
processing system.
It resembles biological neural networks of
human brain.
Processing occurs at large number of single
elements called Nodes or Neurons.
Signals are passed between neurons using
Links
Each link has a weight associated with it.
Each link applies a nonlinear transformation
called an Activation Function to its net input
8. Algorithms for ANNs
Various algorithms can be used such
as :
Back Propagation Algorithm
Conjugate Gradient Algorithms
Radial Basis Function
Cascade Correlation Algorithm
Recurrent ANNs
Self Organizing feature Maps
9. Back Propagation Algorithm
is found to be best suited for
Time Series Data
and most of the
Hydrologic Modeling Problems.
11. Use of ANNs in Hydrology
Rainfall – Runoff Modeling
Modeling Streamflows
Water quality Modeling
Groundwater Studies
Estimating Precipitation
Other Uses
12. Characteristics of Hydrologic Time
Series
Non-stationary
Auto correlated
Cross related
Chronological dependance
These characteristics manifest as
Trend
Seasonality
Cyclic nature
Irregular fluctuations
13. Data Pre-processing
Techniques
Raw Values – for control group
Normalization – De-trending
Logarithmic transform
Logarithmic plus First Difference
Logarithmic plus Second Difference
14. Problem Identification
An investigation is proposed to use different
data pre- processing techniques for multistep
lead time forecasting using different ANN
architectures to develop best model by
evaluating various performance criteria and
make the data more adaptable than the raw
data for ANN modeling, so as to forecast
streamflow more realistically and also to
improve the performance of the ANN model.
15. Study Area
Gauging station at Pandu along
Brahmaputra River at Guwahati is
taken as the study area.
Daily stream flow data for ten year
period
1st January 1990 to 31st December
1999
will be used for the present study.
18. Plan Of Research Work
Plotting and Visual Observation of
Data
Identification of Features Specific to
the Data
Applying Pre-Processing Techniques
Preparation of Data Sets
19. Data Sets
No. of lagged terms Dataset Lagged terms Data Matrix
Input Output
1 Raw values
Log
Log + first
difference
yt = xt
yt = log xt
yt = log xt + first diff.
y1
y2
y3
…..
yt
y2
y3
y4
…..
yt-1
2 Raw values
Log
Log + first
difference
yt = xt
yt = log xt
yt = log xt + first diff.
y1, y2
y2, y3
.….
yt-1, yt
y3
y4
.….
yt-2
3 Raw values
Log
Log + first
difference
yt = xt
yt = log xt
yt = log xt + first diff.
y1, y2, y3
y2, y3, y4
…..
…..
yt-2, yt-1,
yt
y4
y5
…..
…..
yt-3
20. Architectures of ANN
According to Activating Function
According to Number of Neurons
According to Algorithm Used
22. Architectures Used
According to Activation Function
Sigmoid
Tansig
Logsig
According to number of input neurons
1 to 10
Input Neurons will be used
23. Number of Trials
Nine Datasets
Three Architectures
Ten Input Methods
Thus there will be
9 X 3 X 10 = 270
Model Trials