stella.util

Class SensorTest

java.lang.Object

util.PropertyContainer

util.PropertyResources

stella.util.SensorTest

All Implemented Interfaces:

Cloneable, PropertySupplying, ResourceSupplying

public class SensorTest
extends PropertyResources

This class test the behaviour of the linear prediction and the derivative properties of the sensors. It takes the raw output of a sensor stored as ascii-data in a file and performs numerical derivative calculation on it according to the KEY_DERIVE property, and linear prediction according to the KEY_DATABASE, KEY_PREDICT, and KEY_FUTURE properties. The output data are writtn together with the input data to a column-ascii file.

The aim of this test is to find the most reliable derivators and linear predictors, possibly together with read-out times, for all sensors. Prediction should only be accurate for the next two minutes (closure time of the roof).

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
private static class	SensorTest.Parser A command line parser class to allow direct parameter hand-over.

Nested classes/interfaces inherited from class util.PropertyResources

PropertyResources.Load, PropertyResources.URLResource

Field Summary

Fields

Modifier and Type	Field and Description
private static boolean	DEFBIAS The default bias removal.
private static int	DEFCOLUMN The default column to read on input.
private static String	DEFFORMAT The default data output format.
private static int	DEFFUTURE The default number of predicted values.
private static String	DEFOUTFILE The default output file name.
private List	derivatives The list to the data derived from applying the derivators.
private NumberFormat	df The data formatter used.
private List	extrapolations The list to the data derived from applying extrapolation.
private double[]	input The input data of the sensor raw readings.
static String	KEY_BIAS The key pointing to bias removal at predicition.
static String	KEY_COLUMN The key to the column number to read in the input file.
static String	KEY_DATABASE The key pointing the number of input data used for prediction.
static String	KEY_DERIVE The key pointing to a list of derivator files.
static String	KEY_EXTRAPOLATE The key pointing to a list of data points used for extrapolation.
static String	KEY_FORMAT The key to the output data formatter.
static String	KEY_FUTURE The key pointing to the number of points to predict/extrapolate.
static String	KEY_INFILE The key to the input file.
static String	KEY_LINEAR If present, only linear extrapolation is done.
static String	KEY_OUTFILE The key to the output file.
static String	KEY_PREDICT The key pointing to a list of linear prediction coefficient-counts.
static String	KEY_RESIDUAL If present, only residuals of the extrapolation are loaded.
static String	KEY_SMOOTH The key pointing to a list of smoother files.
private List	predictions The list to the data derived from applying linear prediction.
private List	smoothers The list to the data derived from applying the smoothers.

Fields inherited from class util.PropertyResources

KEY_NOINITONCREATE, POSTFIX_ALPHABET, POSTFIX_BYTES, POSTFIX_DIR, POSTFIX_EXT, POSTFIX_FILE, POSTFIX_LIST, POSTFIX_MODTIME, POSTFIX_URL

Fields inherited from class util.PropertyContainer

KEY_LISTSEPARATOR, KEY_MAPKEYVALUECHAR, KEY_MAPSEPARATOR

Fields inherited from interface util.ResourceSupplying

KEY_URLRESOURCES, KEY_URLUSECONFIG, KEY_URLUSECURRENT, KEY_URLUSEHOME

Fields inherited from interface util.PropertySupplying

CONFIG, KEY_CLASS

Constructor Summary

Constructors

Constructor and Description
SensorTest(Map<String,String> prop) Constructs a new censor tester.

Method Summary

Modifier and Type	Method and Description
boolean	applyDerivator() Applies the numerical derivators stated in the KEY_DERIVE property.
boolean	applyExtrapolation() Applies extrapolation as stated in the KEY_EXTRAPOLATE property.
boolean	applyPrediction() Applies linear predictions as stated in the KEY_PREDICT property.
boolean	applySmoothing() Applies the savitzky-goolay filters stated in the KEY_SMOOTH property.
private double[]	deriveOne(double[] indata, SavitzkyGolaySmoothing sgs) Numerically derives an entire data set using a savitzky-golay filter.
private double[]	extrapolateOne(double[] indata, int base, Extrapolation ext, int fut) Applies extrapolation to the entire data set.
static void	main(String[] arg) The main method.
private double[]	predictOne(double[] indata, int base, int back, int fut) Applies linear prediction to the entire data set.
boolean	readInput() Reads the input data.
private double[]	readInputFile(File infile) Reads the input data from the given file.
private double[]	smoothOne(double[] indata, SavitzkyGolaySmoothing sgs) Numerically derives an entire data set using a savitzky-golay filter.
private void	writeData(BufferedWriter out) Writes the data sets.
private void	writeHead(BufferedWriter out) Writes a header to the output stream.
void	writeOutput() Writes the output data.

Methods inherited from class util.PropertyResources

clone, createFrom, createFrom, createFrom, createInstance, getApplet, getAsResources, getLocate, getPropertiesToKey, getPropertiesToKey, getResource, getResourceAsStream, getResourceFromKey, getResources, hasLocalURLs, initProperties, keyCreate, keyCreate, reload, setApplet

Methods inherited from class util.PropertyContainer

augment, augment, augment, augment, augment, defaultBoolean, defaultChar, defaultDouble, defaultFloat, defaultInt, defaultLong, defaultObject, defaultObject, defaultProperties, defaultProperty, getAsBoolean, getAsChar, getAsDouble, getAsEnums, getAsFloat, getAsInt, getAsList, getAsLong, getAsMap, getAsMap, getAsObject, getAsObject, getProperties, getProperty, has, isNew, parseObject, parseObject, reload, removeProperty, requires, rescanned, setObject, setProperties, setProperty, stringProperties, toString

Methods inherited from class java.lang.Object

equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface util.PropertySupplying

defaultBoolean, defaultChar, defaultDouble, defaultFloat, defaultInt, defaultLong, defaultObject, defaultObject, defaultProperties, defaultProperty, getAsBoolean, getAsChar, getAsDouble, getAsFloat, getAsInt, getAsList, getAsLong, getAsMap, getAsObject, getAsObject, getProperties, getProperty, has, parseObject, removeProperty, setObject, setProperty, stringProperties

Field Detail

KEY_SMOOTH

public static final String KEY_SMOOTH

The key pointing to a list of smoother files.

See Also:

Constant Field Values

KEY_DERIVE

public static final String KEY_DERIVE

The key pointing to a list of derivator files.

See Also:

Constant Field Values

KEY_BIAS

public static final String KEY_BIAS

The key pointing to bias removal at predicition.

See Also:

Constant Field Values

KEY_DATABASE

public static final String KEY_DATABASE

The key pointing the number of input data used for prediction.

See Also:

Constant Field Values

KEY_PREDICT

public static final String KEY_PREDICT

The key pointing to a list of linear prediction coefficient-counts.

See Also:

Constant Field Values

KEY_LINEAR

public static final String KEY_LINEAR

If present, only linear extrapolation is done.

See Also:

Constant Field Values

KEY_RESIDUAL

public static final String KEY_RESIDUAL

If present, only residuals of the extrapolation are loaded.

See Also:

Constant Field Values

KEY_EXTRAPOLATE

public static final String KEY_EXTRAPOLATE

The key pointing to a list of data points used for extrapolation.

See Also:

Constant Field Values

KEY_FUTURE

public static final String KEY_FUTURE

The key pointing to the number of points to predict/extrapolate.

See Also:

Constant Field Values

KEY_INFILE

public static final String KEY_INFILE

The key to the input file.

See Also:

Constant Field Values

KEY_OUTFILE

public static final String KEY_OUTFILE

The key to the output file.

See Also:

Constant Field Values

KEY_COLUMN

public static final String KEY_COLUMN

The key to the column number to read in the input file.

See Also:

Constant Field Values

KEY_FORMAT

public static final String KEY_FORMAT

The key to the output data formatter.

See Also:

Constant Field Values

DEFBIAS

private static final boolean DEFBIAS

The default bias removal.

See Also:

Constant Field Values

DEFFUTURE

private static final int DEFFUTURE

The default number of predicted values.

See Also:

Constant Field Values

DEFOUTFILE

private static final String DEFOUTFILE

The default output file name.

See Also:

Constant Field Values

DEFCOLUMN

private static final int DEFCOLUMN

The default column to read on input.

See Also:

Constant Field Values

DEFFORMAT

private static final String DEFFORMAT

The default data output format.

See Also:

Constant Field Values

input

private double[] input

The input data of the sensor raw readings.

smoothers

private List smoothers

The list to the data derived from applying the smoothers.

derivatives

private List derivatives

The list to the data derived from applying the derivators.

predictions

private List predictions

The list to the data derived from applying linear prediction.

extrapolations

private List extrapolations

The list to the data derived from applying extrapolation.

df

private final NumberFormat df

The data formatter used.

Constructor Detail

SensorTest

public SensorTest(Map<String,String> prop)

Constructs a new censor tester. Default properties set are linked to the keys KEY_FUTURE, KEY_COLUMN, KEY_FORMAT, and KEY_OUTFILE.

Method Detail

readInput

public boolean readInput()

Reads the input data. The input file must be an ascii-column file with at least the number of columns stated at KEY_COLUMN. At least this column must be parsable as a double.

Returns:

True, if at least some data points could be read.

readInputFile

private double[] readInputFile(File infile)

Reads the input data from the given file. The input file must be an ascii-column file with at least the number of columns stated at KEY_COLUMN. At least this column must be parsable as a double.

applySmoothing

public boolean applySmoothing()

Applies the savitzky-goolay filters stated in the KEY_SMOOTH property. This property is a comma-separated list of smoother-definition files. Each valid entry (i.e. entries that can be used to construct a savitsky-golay smoother) produces a data set that is the smoothened of the input data. The data missing due to the necessary pre-current data points is zero.

Returns:

True, if the smootherss were sucessfully applied.

smoothOne

private double[] smoothOne(double[] indata,
                           SavitzkyGolaySmoothing sgs)

Numerically derives an entire data set using a savitzky-golay filter. The returned data is of the same dimension as the input data and contains the numerically calculated derivative of the input data at all indices. Note that the first few output data must be zero, as they are needed as the prediction-base for the filter. The amount of leading zeros depends on the filter used.

applyDerivator

public boolean applyDerivator()

Applies the numerical derivators stated in the KEY_DERIVE property. This property is a comma-separated list of derivator-definition files. Each valid entry (i.e. entries that can be used to construct a savitsky-golay derivator) produces a data set that is the first derivative of the input data. The data missing due to the necessary pre-current data points is zero.

Returns:

True, if the derivators were sucessfully applied.

deriveOne

private double[] deriveOne(double[] indata,
                           SavitzkyGolaySmoothing sgs)

Numerically derives an entire data set using a savitzky-golay filter. The returned data is of the same dimension as the input data and contains the numerically calculated derivative of the input data at all indices. Note that the first few output data must be zero, as they are needed as the prediction-base for the filter. The amount of leading zeros depends on the filter used.

applyPrediction

public boolean applyPrediction()

Applies linear predictions as stated in the KEY_PREDICT property. This property is a comma-separated list of integers. Each integer corresponds to the number of linear-prediction coefficients used for the particular prediction. The input data base is constructed from the sensor readings and narrowed down to the KEY_DATABASE data-points back from the current sensor reading. Then, the linear prediction coefficients according to the KEY_PREDICT numbers are calculated. The linear prediction is then carried out for KEY_FUTURE data points starting from the current data point. This yields a single predicted value, stored in the output file. The data (sum of KEY_DATABASE and KEY_FUTURE) missing due to the necessary pre-current data points are zero.

Returns:

True, if the predicitons were sucessfully applied.

predictOne

private double[] predictOne(double[] indata,
                            int base,
                            int back,
                            int fut)

Applies linear prediction to the entire data set. Predictions can only start after the requested input-data base length. From there, again a certain number of values (the number of predicted values) are missing. From there, each input data is represented as a linear predicted value on output.
The prediction procedure is as follows:

• In the input data array, start current index at index base-1.
• Use the base data points from the current index to the past, transfer the data points into an auxilliar array and construct the linear prediction data base out of it.
• Use the back integer to calculate this number of LP coefficients.
• From current index, extrapolate fut data points into the future from current index.
• The single extrapolated value is stored at index current+fut on the output data set.
• Increase current index until the input data set length minus fut is reached.

applyExtrapolation

public boolean applyExtrapolation()

Applies extrapolation as stated in the KEY_EXTRAPOLATE property. This property is a comma-separated list of integers. Each integer corresponds to the number of measurements used for the particular prediction. The input data base is constructed from the sensor readings and narrowed down. Then, the extrapolating polynom is calculated. If the KEY_LINEAR switch is present, linear extrapolation is used, otherwise fitting to a parabolic is done. The extrapolation is then carried out for the single KEY_FUTURE data point, starting from the current data point. This yields a single predicted value, stored in the output file. The data (sum of KEY_EXTRAPOLATE and KEY_FUTURE) missing due to the necessary pre-current data points are zero.

Returns:

True, if the extrapolation were sucessfully applied.

extrapolateOne

private double[] extrapolateOne(double[] indata,
                                int base,
                                Extrapolation ext,
                                int fut)

Applies extrapolation to the entire data set. Extrapolation can only start after the requested input-data base length. From there, again a certain number of values (the number of predicted values) are missing. From there, each input data is represented as an extrapolated value on output.
The extrapolation procedure is as follows:

• In the input data array, start current index at index base-1.
• Use the base data points from the current index to the past, transfer the data points into an auxilliar array and set them as the data points in the extrapolator.
• From current index, extrapolate fut data points into the future from current index.
• The single extrapolated value is stored at index current+fut on the output data set.
• Increase current index until the input data set length minus fut is reached.

writeOutput

public void writeOutput()

Writes the output data. The output data is an ascii-file with columned data. The first column is the input data, followed by the derivative data. Each derivative column corresponds to a derivator named in the KEY_DERIVE property. After all derivative data, the linear predicted data is added. Each predicted column corresponds to a number of prediction coefficients, as stated in the KEY_PREDICT property.

writeHead

private void writeHead(BufferedWriter out)
                throws IOException

Writes a header to the output stream. The header starts with a '#' char to indicate a comment line. After that, the keyword ' in' follows. Each derivative data set present in the derivatives list adds a ' D#' entry, where '#' indicates the index of the derivator used. Using the predicted data sets in the predictions list, similar ' P#' character sequences are added.

Parameters:

out - The stream to write to

Throws:

IOException

writeData

private void writeData(BufferedWriter out)
                throws IOException

Writes the data sets. First data column is the input data, followed by the derivatives data, again followed by the data predictions.

Throws:

IOException

main

public static void main(String[] arg)
                 throws ParseException

The main method. Reads tthe command line arguments and parses them according to SensorTest.Parser. Note that the input file may only be specified with a switch, a command line argument is always a properties file.

Throws:

ParseException