1. Field of the Invention
This invention relates generally to a technique for measuring the width of a noisy pulse in an analog signal and, more particularly, to a technique for remotely determining whether a valve associated with a cylinder in a locomotive diesel engine is defective by measuring the width of ultrasonic pulses emitted therefrom.
2. Discussion of the Related Art
Locomotives employ diesel engines having sixteen or more cylinders. Each cylinder employs several valves, for example four valves, where some of the valves allow a fuel/air mixture to be emitted into the cylinder and other of the valves allow exhaust gas to be removed from the cylinder. It is important that the seal integrity of the valve relative to the valve seat around the opening in the cylinder is sufficient so that the engine operates properly and is not damaged. However, the rigors of the combustion process within the cylinder sometimes causes cracks and other defects in the valve or the valve seat which may affect seal integrity. Continued operation of the engine with a defective valve may cause valve failure. Further, valve failure in a locomotive engine can result in damage to the engine turbocharger which may cost $100,000.00 or more to replace or fix.
Heretofore, it was usually necessary for a technician to visually inspect the valve and valve seat to look for evidence of cracking or other damage during maintenance over hauls and the like. This is a costly process because the valve head needs to be removed and each valve inspected which includes significant labor and downtime. Further, the inspection process is highly subjective in that the technician determines whether a defect exists by his or her own visual assessment and experience.
To overcome these problems, it is known in the art to employ a detection system including an ultrasonic sensor probe that detects ultrasonic emissions from the engine while it is idling. Ultrasonic emissions are typically emitted from the engine during such times as fuel injection and exhaust emission when the valves are opened. The probe is placed in contact with suitable locations on the cylinder head and a measurement is taken for each cylinder. The detection system includes processing circuitry that provides heterodyning by mixing the ultrasonic signal for frequency down-conversion purposes, for example 40 kHz to 0-3 kHz, to make the sound energy audible. The detection system also includes a headset for listening to the down-converted signal.
Because a defect in the valve causes additional gases to leak during the compression stroke, additional ultrasonic energy is emitted if such a defect exists. Therefore, by knowing the “sounds” that the engine cylinder makes with no defect, a technician can listen for higher intensity signals indicating the presence of a valve defect. However, such a technique for determining valve defects still includes a subjective aspect where the technician must determine the defect by the sound perceived. It would be desirable to provide a more cost effective and objective technique for determining if a defect exists in a valve or valve seat associated with a cylinder of a locomotive engine.
In accordance with the teachings of the present invention, a measurement technique is disclosed that generates a number value identifying the width of a noisy pulse in an analog signal. In one embodiment, the noisy pulse is part of a down-converted ultrasonic signal detected from sound transmissions from a cylinder in a locomotive diesel engine. A wider pulse, and thus a higher number value, provides an indication of whether a valve associated with the cylinder is defective, where the sealing integrity of the valve has been compromised.
The technique employs a mathematical algorithm that samples the down-converted signal at predetermined periods of time to provide an analog-to-digital conversion. The samples are then bunched into successive groups, where each group includes a predetermined number of samples. A root mean square is taken of the samples in each group to generate a representative amplitude value for that group. Successive amplitude values, such as three successive values, are then multiplied together to generate product values. The product values are averaged over a predetermined number of product values to generate the number value that can be analyzed to determine if a defect in the valve exists. If a noisy pulse in the signal generates a number value that is greater than a certain number valve, then the valve can be determined to have a defect. The number value can be ratioed against another product value of the amplitude values to compensate for gain differences in the signal.
Additional objects, advantages and features of the present invention will become apparent from the following description and appended claims, taken in conjunction with the accompanying drawings.
The following discussion of the invention directed to a mathematical technique for determining whether a defect exists in a valve of a cylinder of an engine is merely exemplary in nature, and is in no way intended to limit the invention or its applications or uses. Particularly, the discussion below concerns determining defects in a valve of a locomotive diesel engine. However, as will be appreciated by those skilled in the art, the algorithm of the present invention has a much wider application.
The probe was used to sense the ultrasonic emissions from the engine cylinder where one of the valves of the cylinder had a known specific size defect. Particularly, the valve of the cylinder shown in
By comparing the signals outputs of
According to the invention, a mathematical algorithm is disclosed that generates a number value indicative of the width of noisy pulses in the down-converted analog output from a engine cylinder. By comparing the generated number value against a number value from a cylinder having no valve defects, a substantially objective technique is provided for determining valve defects in a cylinder valve. In one embodiment, the algorithm employs an appropriate sample interval, and assumes that all noisy pulses in the signal are less than three sample intervals long. If the pulses were sufficiently spaced, then, in the binary case, the product of any four successive samples would be zero or near zero. In such a case, the local discreet-time function comprising the product of four adjacent samples would be highly discriminatory against pulses below two or so intervals in duration. In other words, the product of the samples would be less for those pulses having shorter durations and greater for those pulses having a longer duration. Therefore, the present invention proposes computing a product of adjacent time samples to determine the duration of the noisy pulse to determine the width of the pulse, and thus, whether a defect exists.
The values Vi are applied to a multiplying device 18 and a multiplying device 20. The multiplying device 18 computes a product for every three consecutive Vi, or Vi, Vi−1, and Vi−2, and generates a product value fi. Thus, the longer the pulse of noise in the analog input signal, the greater the computed value. Therefore, if a particular pulse is more than a single bunch of 25 successive samples long, then the product will be greater than if the pulse width was less than the bunch sample time for collecting the bunch. The multiplying device 20 cubes each value Vi for normalization purposes and generates a product value gi. This normalization step is important to compensate for gain variations that may occur in the devices 12, 14 and 16 and the ultrasonic pickup of the probe (not shown). Thus, by cubing each value Vi, variations in gain in the signal do not affect the overall output.
The product value fi is applied to an averaging device 22 and the product value gi is applied to an averaging device 24. The averaging devices 22 and 24 can generate an average, a means, a percentile, etc., on a moving basis or on a finite block of data. The averaging device 22 averages successive values fi, and the averaging device 24 averages successive values gi, preferably over each stroke of the engine. For example, each of the averaging devices 22 and 24 averages about 50 samples. The average values are applied to a ratio device 26 that generates a ratio of f/g. Thus, if the value computed in the multiplying device 18 is high because the pulse width is larger than normal as a result of a valve defect, then the ratio f/g will be higher than what is normally seen in a non-defective valve. Therefore, a number greater than a predetermined number value of a non-defective valve indicates a defective valve.
One of normal skill in the art would readily recognize what various circuits could be used for each of the devices 12, 14, 16, 18, 20, 22, 24 and 26 discussed above. Further, the various operations discussed herein could be performed by suitable software programs, also readily apparent to those skilled in the art.
Other embodiments of the invention can apply more sophisticated processing, including employing fourier transforms or other known mathematical algorithms, to multiply continuous samples together consistent with the discussion herein. However, the embodiment discussed above employs reduced processor memory and processing power to perform the same operation. Other approaches include exploiting wavelet analysis to map a function of one variable, time, into a function of two variables, time and duration. Wavelet analysis relies on the computation of convolution integrals at each point in time for a potentially large number of similarly-shaped functions of time-scale or duration. The resulting two-dimensional function, presented as a contour plot, can provide a visual indication of underlying processes characterized by different time scales, periods and phases, such as appear to be present in the data provided in
The foregoing discussion discloses and describes merely exemplary embodiments of the present invention. One skilled in the art will readily recognize from such discussion and from the accompanying drawings and claims that various changes, modifications and variations can be made therein without departing from the spirit and scope of the invention as defined in the following claims.
This application claims the benefit of U.S. provisional application No. 60/357,516, titled “Technique for Computing a Measure on an Ultrasonic Signal Having Application to Identify Valve Defects in Operating Engines,” filed Feb. 15, 2002.
Number | Name | Date | Kind |
---|---|---|---|
4292841 | Wesley | Oct 1981 | A |
4444049 | Hitchcock | Apr 1984 | A |
5121443 | Tomlinson | Jun 1992 | A |
5377131 | Ottesen et al. | Dec 1994 | A |
5467185 | Engeler et al. | Nov 1995 | A |
5930136 | Nakajima et al. | Jul 1999 | A |
6085727 | Nakano | Jul 2000 | A |
6195020 | Brodeur et al. | Feb 2001 | B1 |
6230546 | Wilstermann et al. | May 2001 | B1 |
6278914 | Gaudreau et al. | Aug 2001 | B1 |
6455850 | Coates et al. | Sep 2002 | B1 |
6535810 | Volpe et al. | Mar 2003 | B2 |
6707043 | Coates et al. | Mar 2004 | B2 |
20040031856 | Atsmon et al. | Feb 2004 | A1 |
Number | Date | Country |
---|---|---|
389943 | Feb 1990 | AT |
3207556 | Aug 1983 | DE |
0871 036 | Oct 1998 | EP |
Number | Date | Country | |
---|---|---|---|
20030158712 A1 | Aug 2003 | US |
Number | Date | Country | |
---|---|---|---|
60357516 | Feb 2002 | US |