CN-121993290-A - Engine knock fault detection method based on vibration signal analysis

CN121993290ACN 121993290 ACN121993290 ACN 121993290ACN-121993290-A

Abstract

The invention relates to the technical field of engine detection, in particular to an engine knock fault detection method based on vibration signal analysis. The method comprises the steps of obtaining a voltage vibration signal and a pulse square wave signal of an engine, preprocessing the voltage vibration signal and the pulse square wave signal to obtain a vibration digital sequence and a pulse time stamp set, extracting a difference value of adjacent time stamps from the pulse time stamp set to serve as a time span, constructing a rigid body dynamic equation by combining a physical mechanical angle interval, calculating transient angular acceleration, mapping the vibration digital sequence to a crankshaft rotation space to obtain an equal angle vibration sequence, performing frequency domain conversion on the equal angle vibration sequence, filtering mechanical noise to obtain a knocking residual signal, calculating knocking characteristic energy in a monitoring window, and issuing a control and regulation instruction when the knocking characteristic energy exceeds a preset safety threshold value. The invention restores the mechanical operation state under transient variable load, effectively inhibits the order ambiguity defect and reduces the bottom operation load and response delay.

Inventors

YANG YONGBO
MA HONGJIANG
ZHANG HAO
YANG WENSEN
XUE YING

Assignees

西安康明斯发动机有限公司

Dates

Publication Date: 20260508
Application Date: 20260407

Claims (10)

1. An engine knock fault detection method based on vibration signal analysis, comprising: acquiring a voltage vibration signal of an engine and a pulse square wave signal with a physical and mechanical angle interval, and preprocessing to obtain a vibration digital sequence and a pulse time stamp set; Extracting the difference value of adjacent time stamps from the pulse time stamp set as a time span, wherein the time span comprises a first time span and a second time span, and calculating to obtain transient angular acceleration and initial angular velocity by combining the physical mechanical angle interval; Mapping the vibration digital sequence to a crankshaft rotation space based on the transient angular acceleration and the initial angular velocity to obtain an equal-angle vibration sequence; And performing frequency domain conversion on the equiangular vibration sequence, filtering mechanical noise to obtain a knocking residual signal, calculating knocking characteristic energy of the knocking residual signal in a preset monitoring window, and issuing a control and regulation instruction when the knocking characteristic energy exceeds a preset safety threshold.
2. The engine knock fault detection method based on vibration signal analysis according to claim 1, wherein the preprocessing comprises the steps of sampling the voltage vibration signal at equal time intervals according to a preset sampling frequency to obtain the vibration digital sequence, capturing level jump edges of the pulse square wave signal, recording absolute time stamps of each level jump edge, and constructing the pulse time stamp set.
3. The method for detecting engine knock fault based on vibration signal analysis according to claim 1, wherein the first time span and the second time span include extracting a first time stamp, a second time stamp, and a third time stamp from the set of pulse time stamps, taking a difference between the second time stamp and the first time stamp as the first time span, and taking a difference between the third time stamp and the second time stamp as the second time span.
4. The engine knock fault detection method based on vibration signal analysis according to claim 1, wherein the transient angular acceleration satisfies a relation: Wherein, the Is the transient angular acceleration; For the physical mechanical angular separation; for the first time span; for said second time span.
5. The engine knock fault detection method based on vibration signal analysis according to claim 1, wherein the initial angular velocity satisfies a relation: Wherein, the For the initial angular velocity; For the physical mechanical angular separation; for the first time span; Is the transient angular acceleration.
6. The method for detecting engine knock fault based on vibration signal analysis according to claim 1, wherein the equiangular vibration sequence includes calculating a crank angle at each sampling time in the vibration digital sequence, resampling the vibration digital sequence based on the crank angle to obtain the equiangular vibration sequence; The crank angle satisfies the relation: Wherein, the Index the sampling time; Is the first The sampling moments; Is the first -Said crank angle for each of said sampling instants; is a first timestamp; the crank angle being the first timestamp; For the initial angular velocity; Is the transient angular acceleration.
7. The engine knock fault detection method based on vibration signal analysis according to claim 1 is characterized in that the knock residual signal comprises the steps of carrying out frequency domain conversion on the equiangular vibration sequence to obtain an order domain frequency spectrum, configuring a multi-order stop band filter with a center stop band aligned with the mechanical excitation order of the engine, carrying out zeroing filtering on pulse spectral line characteristics in the order domain frequency spectrum by utilizing the multi-order stop band filter to obtain a pure order domain frequency spectrum, and carrying out reverse frequency domain conversion on the pure order domain frequency spectrum to obtain the knock residual signal.
8. The method for detecting engine knock fault based on vibration signal analysis according to claim 1, wherein the knock characteristic energy includes extracting discrete voltage magnitudes of the knock residual signals falling within the preset monitoring window and performing square accumulation summation to obtain the knock characteristic energy.
9. The method for detecting engine knock fault based on vibration signal analysis according to claim 1, wherein the preset safety threshold is obtained through an off-line bench calibration experiment.
10. The method for detecting engine knock failure based on vibration signal analysis according to claim 1, wherein the control adjustment command includes retarding an ignition advance angle and a main injection advance angle of a cylinder in which knocking occurs, and reducing a fuel injection amount.

Description

Engine knock fault detection method based on vibration signal analysis Technical Field The invention relates to the technical field of engine detection, in particular to an engine knock fault detection method based on vibration signal analysis. Background A heavy-duty commercial vehicle, such as a dump truck and a mining tractor carrying a large-displacement diesel or natural gas engine, has frequent transient abrupt load under extreme working conditions such as complex road conditions, so that accurate knock fault detection is a core technology for protecting the engine from destructive mechanical impact, a step tracking algorithm is generally adopted in the prior art, and a core logic of the step tracking algorithm is to resample a vibration signal sampled in equal time in a time domain into a step domain signal sampled at equal angles by utilizing a pulse signal output by a crank shaft position sensor, so that mechanical noise strongly related to the engine speed is accurately stripped through a filter. However, the prior art is usually limited by an interdental constant speed assumption when mapping from execution time to angle, the assumption is determined between two adjacent physical teeth of a crank sensor, the instantaneous rotation speed of an engine is absolutely constant, under strong transient working conditions such as heavy load starting or rapid acceleration of a dump truck, the engine can burst huge angular acceleration in an extremely short working cycle, at the moment, the assumption can thoroughly fail, so that a calculated virtual angle deviates from a real crank angle, engineering practice shows that when the rotation speed of the engine is in a specific rotation speed section and rapid acceleration, the phase error of an order domain induced by the method can even exceed twelve percent, in order to compensate the inherent defect, the prior art carries out smooth fitting interpolation by introducing nonlinear mathematical means, such as using a cubic spline function to carry out diagonal displacement, the processing mode of the purely mathematical layer not only depends on a smooth penalty term or a damping coefficient set by manpower, but also consumes extremely large edge calculation force, so that an underlying control unit cannot timely retract to the ignition angle in advance at the early stage of knocking occurrence, and finally causes fatigue damage of the top part of the piston and the cylinder. Disclosure of Invention In order to solve the problems of order ambiguity caused by failure of interdental constant speed hypothesis and control hysteresis caused by mathematical fitting in the prior art, the invention provides an engine knock fault detection method based on vibration signal analysis, which comprises the following steps: The method comprises the steps of obtaining a voltage vibration signal of an engine and a pulse square wave signal with a physical mechanical angle interval, preprocessing the voltage vibration signal to obtain a vibration digital sequence and a pulse time stamp set, extracting a difference value of adjacent time stamps from the pulse time stamp set to serve as a time span, calculating a transient angular acceleration and an initial angular velocity by combining the physical mechanical angle interval, mapping the vibration digital sequence to a crankshaft rotation space to obtain an equal-angle vibration sequence based on the transient angular acceleration and the initial angular velocity, performing frequency domain conversion on the equal-angle vibration sequence, filtering mechanical noise to obtain a knocking residual signal, calculating knocking characteristic energy of the knocking residual signal in a preset monitoring window, and issuing a control adjustment instruction when the knocking characteristic energy exceeds a preset safety threshold. The invention eliminates order ambiguity caused by the traditional interdental constant speed assumption by restoring the mechanical motion track under the extremely variable load working condition, optimizes the mapping logic from physical to digital space, effectively reduces the calculation load of bottom hardware and provides a high-precision and agile protection mechanism for the engine. Further, the preprocessing comprises the steps of sampling the voltage vibration signal at equal time intervals according to preset sampling frequency to obtain the vibration digital sequence, capturing level jump edges of the pulse square wave signal, recording absolute time stamps of each level jump edge, and constructing the pulse time stamp set. According to the invention, the voltage vibration signal is sampled at equal time intervals according to the preset sampling frequency, and the level jump edge of the pulse square wave signal is captured to record the absolute timestamp, so that the alignment of the high-frequency vibration characteristic and the physical clock is realized in the bottom data acquisition st