CN-115712853-B - Engine sound-vibration multi-information fusion diagnosis equipment and fault diagnosis method

CN115712853BCN 115712853 BCN115712853 BCN 115712853BCN-115712853-B

Abstract

The invention relates to a sound-vibration multi-information fusion diagnosis device and a fault diagnosis method for an engine, which are used for stripping interference parts in sound time domain data of a fault engine, respectively classifying and positioning fault characteristics rapidly by a table lookup method according to three layers of sound-vibration time domain characteristic correlation, namely sound order data and vibration order data, so as to rapidly determine possible fault forms of the engine, and respectively adopting a time domain/frequency domain multi-layer signal fault information extraction method to better process abnormal sound noise signals with strong periodicity and unsteady state signals, so as to facilitate the identification work of the fault forms with unsteady state noise characteristics. Compared with the existing engine site abnormal sound diagnosis scheme, the method has good adaptability, and through comparing the characteristics of the multidimensional information result and carrying out correlation analysis, the method can effectively filter external interference signals in the test process and locate abnormal sound sources from a mechanism layer with higher dimensionality.

Inventors

CAI YIXIAO
JIN YEQI
LI XIAOQI

Assignees

上海大众动力总成有限公司

Dates

Publication Date: 20260505
Application Date: 20221117

Claims (3)

1. The engine sound-vibration multi-information fusion diagnosis equipment is characterized by comprising an acoustic measurement module, a vibration measurement module, a data acquisition module, a data operation analysis module and an interactive display module; The acoustic measurement module and the vibration measurement module are arranged on the engine side end body, and are used for collecting engine noise signals and vibration signals on site and sending the signals to the data collection module; The data acquisition module acquires real-time engine noise signals and vibration signals output by the acoustic measurement module and the vibration measurement module and real-time rotation speed signals detected by the engine rotation speed sensor signals, processes the signals and then sends the processed signals to the data operation analysis module; The data operation analysis module is used for performing operation analysis according to the data sent by the data acquisition module, outputting fault types and fault information judgment, sending the fault types and the fault information judgment to the interaction display module, and performing information interaction with detection staff; the data operation analysis module executes the following steps: 1) The method comprises the following steps that an acoustic measurement module is arranged on an engine side end body, background noise time domain data are collected in an engine-non-started environment, then the background noise time domain data are converted into a frequency domain through a time-frequency domain conversion method, and frequency domain characteristics of the background noise time domain data are extracted; 2) The method comprises the steps of extracting acoustic time domain data of an engine, namely starting a generator, collecting the acoustic time domain data of the engine by an acoustic measurement module, and acquiring effective engine acoustic data by a filtering and denoising algorithm by combining the background noise frequency domain characteristics acquired in the step 1); 3) The engine sound-vibration information preprocessing comprises the steps of periodically segmenting an acquired engine sound-vibration time domain signal through an input rotating speed signal, engine vibration time domain data and engine effective acoustic data by a self-adaptive data period segmentation method; 4) Multi-layer data feature extraction, namely processing data on two layers of time domain and frequency domain respectively aiming at the segmented vibration and acoustic time domain signals, The time domain layer data processing is that the kurtosis, margin, waveform and peak information of acceleration and acoustic signals are compared and analyzed, and signals which are irrelevant to the mechanical system of the engine and have non-strong periodicity are analyzed; The frequency domain layer data processing is that the order information of vibration and acoustic signals is analyzed by combining the engine rotating speed signal, and the acoustic vibration signals with different dimensionality categories are distinguished; 5) Multidimensional signal correlation analysis and fault classification: Classifying fault types by a table look-up method according to correlation analysis of acoustic order data, vibration order data and acoustic-vibration time domain characteristics; 6) And (3) fault information judgment: step 5) after the signal correlation analysis and fault classification work are completed, fault information judgment is carried out, wherein a preset fault threshold method is adopted for fault judgment aiming at a fault form with obvious characteristics; the frequency domain layer data processing comprises the steps of rapidly distinguishing and classifying mechanical noise generated by different mechanical faults and fluid noise generated by valve body faults in different dimensions through order frequency domain characteristics related to engine rotating speed orders, analyzing the order information of vibration and acoustic signals by combining engine rotating speed signals, extracting and sequencing the energy of the orders and the appointed orders in energy concentration, and analyzing the vibration and noise introduced by engine crankshafts, camshafts and bearing mechanical parts; The engine sound-vibration time domain signals acquired in the step 3) are segmented in a periodic time domain, and segmented data are used as training input data of an artificial intelligent self-learning method based on big data; The time domain segmentation interval of the data in the self-adaptive data period segmentation method is judged through the time domain period characteristics of the engine rotating speed signal or the original vibration signal in the engine vibration time domain data, so that data processing is carried out on sound-vibration information in a minimum period, and the table look-up method classifies possible fault types through acoustic energy order sorting and vibration energy order sorting indexes.
2. The method for establishing the data operation analysis module in the engine sound-vibration multi-information fusion diagnostic equipment according to claim 1, wherein the acoustic time domain data and the engine vibration time domain data of the engine in the steps 2) and 3) comprise normal engine operation data and various engine fault operation data.
3. The fault diagnosis method based on the engine sound-vibration multi-information fusion diagnosis equipment according to claim 1 is characterized in that a field acoustic measurement module collects background noise and engine operation noise signals in the measurement process, a field vibration measurement module collects vibration signals after the engine operates, a rotation speed sensor collects engine rotation speed signals, a data operation analysis module receives signals collected by the field acoustic measurement module, the field vibration measurement module and the rotation speed sensor through a data collection module, and the data operation analysis module carries out background noise measurement and feature extraction, acoustic information noise elimination and sound-vibration information preprocessing according to collected data, carries out fault information judgment after multi-layer data feature extraction, multidimensional signal correlation analysis and fault classification, and outputs fault diagnosis results and corresponding fault features; in the acoustic-vibration information preprocessing, the time domain segmentation interval of data in the self-adaptive data period segmentation method is judged through the time domain period characteristics of the engine rotating speed signal or the original vibration signal in the engine vibration time domain data, so that the acoustic-vibration information is processed with the minimum period; in the multidimensional signal correlation analysis and fault classification, when a table look-up method is adopted to classify fault types, possible fault types are classified through acoustic energy order sorting and vibration energy order sorting indexes.

Description

Engine sound-vibration multi-information fusion diagnosis equipment and fault diagnosis method Technical Field The invention relates to a vehicle fault diagnosis technology, in particular to engine sound-vibration multi-information fusion diagnosis equipment and a fault diagnosis method applied to field diagnosis. Background At the heart of the vehicle's powertrain, the engine is primarily responsible for providing the vehicle with the mechanical energy in the form required. Engine system failure is also a major component of vehicle powertrain failure. Among them, engine abnormal failures are the main cause of customer complaints. Because the engine system has a complex structure and a large number of accessories, and in the running process, the noise of the engine system comprises a plurality of components such as mechanical component noise, hydrodynamic noise, combustion noise and the like. Therefore, the faults exist in various forms, the sources are wide, and the generation mechanism is complex. In the field maintenance link, the abnormal sound fault of the engine is usually carried out by a subjective evaluation method which does not depend on instruments and equipment or an objective evaluation method using data acquisition equipment, and the abnormal sound is subjectively judged by hearing of a fault maintainer. The method is simple to operate, does not need to rely on any equipment, and has good identification accuracy on abnormal sound faults with obvious characteristics (such as serious engine cylinder lack shake, serious mechanical interference noise and the like). The method has higher technical level requirements for maintenance personnel. It is difficult to form a set of engine abnormal sound fault identification method and process capable of objectively and quantitatively. It is also difficult to analyze the source of abnormal sound faults of the engine in a theoretical level. The latter uses data acquisition devices (such as audio recording devices, acceleration acquisition devices and the like) to acquire the sound-vibration characteristics of the fault engine, so as to objectively analyze the engine fault data. At present, however, the device generally acquires the sound-vibration characteristics and other engine parameters (such as rotating speed and the like) independently, and the fault judgment also generally adopts a threshold judgment method and the like. Therefore, it is only possible to determine from the appearance whether the engine has a fault (such as whether the designated order is out of limit or not, whether the working noise is out of limit or not, etc.), it is difficult to comprehensively analyze the correlation between the data information of the sound-vibration multiple dimensions of the fault engine, and the determination of the abnormal sound fault source is also relatively original (usually only performed by a single sound-vibration order table look-up method), and the capability of processing and distinguishing the unstable signal faults (such as intermittent faults, pneumatic element faults) and external disturbances is lacking. Disclosure of Invention Aiming at the problems existing in the prior engine fault judgment, the engine sound-vibration multi-information fusion diagnosis equipment and the fault diagnosis method are provided. The invention has the technical scheme that the engine sound-vibration multi-information fusion diagnosis equipment comprises an acoustic measurement module, a vibration measurement module, a data acquisition module, a data operation analysis module and an interactive display module, wherein the acoustic measurement module and the vibration measurement module are arranged on an engine side end body, and are used for acquiring engine noise signals and transmitting the vibration signals to the data acquisition module on site; The data acquisition module acquires real-time engine noise signals and vibration signals output by the acoustic measurement module and the vibration measurement module and real-time rotation speed signals detected by the engine rotation speed sensor signals, processes the signals and then sends the processed signals to the data operation analysis module; And the data operation analysis module is used for carrying out operation analysis according to the data sent by the data acquisition module, outputting fault types and fault information judgment, and sending the fault types and the fault information judgment to the interaction display module for carrying out information interaction with detection staff. The method for establishing the data operation analysis module in the engine sound-vibration multi-information fusion diagnosis equipment specifically comprises the following steps: 1) The method comprises the following steps that an acoustic measurement module is arranged on an engine side end body, background noise time domain data are collected in an engine-non-started environment, then the background noise time