Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Mechanical measurement system
Reexamination Certificate
2003-05-01
2004-11-16
Barlow, John (Department: 2863)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Mechanical measurement system
C702S036000, C702S054000, C073S592000
Reexamination Certificate
active
06820016
ABSTRACT:
The present invention relates to the field of detecting leaks in pipes. More specifically, the present invention relates to apparatus and methods for detecting leaks which require transmission of data recorded at the pipe to a remote processor.
Fluid flowing through a pipe constantly generates an acoustic signal which propagates along the walls of the pipe and the through the fluid itself. If there is a leak in the pipe, the escaping fluid and the fluid passing over the leak, also generates an acoustic signal. Therefore, a leak can be detected by listening for such an acoustic signal.
There are known methods for accurately determining the position of a leak. For example, a popular method is to use a ‘leak noise correlator’. This comprises a plurality of fixed sensors which are located at intervals along the pipe. If a leak occurs in a section of pipe between two sensors, both of the sensors detect the acoustic signal from the leak. The acoustic signal will propagate from the leak at the speed of sound in the pipe. Therefore, the time at which the two sensors detect the leak signal will depend on their relative distances from the leak. Comparing the arrival time of the leak signal at the two fixed sensors allows the position of the leak to be determined.
Generally, acoustic signals measured by the two sensors are transmitted away from the sensors to a processing unit for comparison. The acoustic signal from the leak is buried in the background acoustic signal from the pipe and is hard to extract, especially when the leak is small. Therefore, it is advantageous to process a large amount of data from the sensors in order to detect and pinpoint a leak. This causes a problem as there is a needed to transmit a large amount of data from the sensors to the remote processor.
For example, a typical leak noise correlator will require acoustic data sampling at a rate of 10 kHz. If the signal is digitised using a 16 bit analogue to digital converter, it will be necessary to transmit at least 160,000 bits per second of information for each sensor, if the pipe is to be monitored in real time. This transmission rate is beyond the capabilities of current, readily available, radio modem technology.
A lower sampling frequency could be used, but this results in a higher uncertainty in the predicted position of the leak. The data could be compressed using a standard compression technique, such as logarithmic compression. However, standard compression techniques create unacceptable loss in the resolution of the signal, which makes detection of a weak leak signal virtually impossible. Spread spectrum radio modems can transmit such a volume of data. However, they generally work at very high frequencies (typically 0.9 to 3 GHz). As a result, transmission distances can be very short, and usually, a line of sight is required between the transmitter and receiver. Therefore, they are not of use where the line of sight can be constantly interrupted by traffic and other such obstructions.
The present invention addresses the above problems and, in a first aspect provides, an apparatus for detecting a leak in a pipe, the apparatus comprising: a sensor located at a pipe configured to detect a signal from the pipe; converting means to convert the signal detected by the sensor into a digital signal; transform means to transform the digital signal into a different orthogonal space; and a transmitter for transmitting the transformed digital signal back to a remote processor.
Transforms which transform the digital data to a different orthogonal space are of particular use in processing acoustic signals from leaks. The transform should form an unconditional basis for the information. An unconditional basis results in expansion coefficients of a largely low order with a magnitude which decreases rapidly with increasing order.
The widely accepted definition of an unconditional basis follows loosely the form developed by Donoho in 1993 (D. L. Donoho, “Unconditional Bases Are Optimal Basis For Data Compression And For Statistical Estimation”, Applied and Computational Harmonic Analysis, I (1): 100-115, December 1993). An unconditional basis is formally defined by considering a function class F with a norm defined and denoted by ∥.∥F and a basis set f
k
such that any function g&egr;F has a unique representation g=&Sgr;
k
a
k
f
k
with equality defined as a limit using the norm, we consider the infinite expansion:
f
(
t
)
=
∑
k
=
-
∞
∞
c
k
ϕ
(
t
-
k
)
+
∑
k
=
-
∞
∞
∑
j
=
0
∞
d
j
,
k
ψ
(
2
j
t
-
k
)
if for all g&egr;F, the infinite sum converges for all |m
k
|≦1, the basis is called an unconditional basis. Using such an unconditional basis, all subsequences of wavelets converge, and all sequences of subsequences converge. The convergence does not depend on the order of terms in the summation or on the sign of the coefficients. This implies a very robust basis, where the coefficients drop off rapidly for all members of the function class. This is the case for wavelets and leak noise data.
The transform is more preferably a discrete time wavelet transform of the form.
ψ
(
t
)
=
ⅇ
-
t
2
cos
(
π
·
t
2
ln
(
2
)
)
Where f(t) represents the digital data outputted from the analogue to digital conversion means and t is the time. The &phgr;(t−k) and &psgr;(2
j
t−k) functions represent the mother scaling functions and wavelet functions respectively which are used for the discrete wavelet transform.
The coefficients c
k
and d
j,k
are calculated from the inner product of f(t) with the scaling functions and wavelet functions such that:
c
k
=∫f
(
t
)&phgr;(
t−k
)
dt
d
j,k
=∫f
(
t
)&psgr;(2
j
t−k
)
dt
As the wavelet transforms used form an unconditional basis for the data, the magnitude of expansion coefficients (c
k
and d
j,k
above) drop off rapidly with increasing j and k. Therefore, large numbers of the coefficient array elements are very small or zero. The coefficients are preferably calculated using a fast transform technique. In common with the classic fast Fourier transform, the technique employed in this algorithm preferably requires that the number of data points in a packet is an exact power of 2.
Preferably, the discrete wavelet transform is a Fourier transform or a wavelet which encompasses a harmonic form. For example, a Morlet wavelet which is essentially a harmonic waveform modulated by a Gaussian envelope. A generic form of a Morlet wavelet can be expressed as:
g
(
t
)
=
∑
k
m
k
a
k
f
k
(
t
)
More simplified forms of wavelet transforms could also be used, for example the so-called Haar transform. It is well known to those skilled in the art that the choice mother scalar function is dependent on the wavelet function.
The use of the above transform allows a data efficient compression technique to be performed before the signal is transmitted. Many known methods of comparing the signal taken from two adjacent sensors require some sort of wavelet transform to be performed on the signal (e.g. a Fourier Transform which is a specific type of wavelet transform). This is typically performed at the remote processor. Therefore, by performing the transform at the pipe, a more efficient data compression technique can be achieved without actually requiring any more processing steps.
The “raw” data can be examined at the remote processor by performing the inverse transform at the remote processor.
Preferably, the transformed signal is passed through scalar quantising means. More preferably, the quantising means is configured to optimise the number of bits, to minimise the information loss in the reconstructed datastream. In typical use, the scalar quantiser is configured such that the number of bits in the outputted data stream is two more than the number of significant bits in the raw unprocessed acoustic datastream.
For transmission, the data whi
Brown Ian J.
Smith Brynley D.
Williams Andrew J.
Dickstein , Shapiro, Morin & Oshinsky, LLP
Palmer Environmental Limited
Walling Meagan S
LandOfFree
Leak detection apparatus and method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Leak detection apparatus and method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Leak detection apparatus and method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3362847