TWI228646B - A signal processing method and its computational algorithm in the diagnosis of component defects for mechanical systems - Google Patents

Abstract

This invention is to propose both a signal processing method and its computational algorithm to apply in the diagnosis of component defects for mechanical systems. A demodulation function for signal processing is used to excellently demodulate the vibration signal and can be applied to present a 3-D spectrum. To implement the function, both the signal processing method and its computational algorithm is proposed and would provide a convenient tool in constructing the 3-D spectrum. From the sense of sight, the 3-D spectral pattern would be very useful in the defect diagnosis.

Description

1228646 发明 Description of the invention: [Technical field to which the invention belongs] The present invention relates to the signal processing and execution method of the operating limbs of the aging system, especially the heart-turning tune analysis function is effective, and it has a fairly perfect signal. Demodulation capability 'and using the result of the demodulation function to present it in a three-dimensional spectrum for users to visually make clear and convenient spectrum mode judgments of the operating conditions of mechanical system elements Applicant. [Previous technology] The analysis of the shipment conditions of the surface system can generally be divided into three stages: New Zealand acquisition, signal processing and frequency spectrum. First use a vibration sensor (such as a displacement meter, speed gauge, or acceleration gauge) to convert the physical quantity of vibration into a voltage form, and then moderately amplify the voltage through an amplifier, and then input the voltage into the computer through an analog-to-digital conversion interface for digital signals After processing, the vibration frequency spectrum is taken to show the frequency characteristics of the vibration signal, and the frequency distribution and pattern of the frequency spectrum are used to judge the damage of the mechanical system components. Damage to mechanical system components will cause knocking vibrations in the operation of the system, and since most moving components of the system are rotating, if such components are damaged, periodic knocks will occur. Therefore, the operating conditions of the mechanical components are diagnosed. It is to detect the occurrence of this periodic vibration signal. However, due to the infiltration of a considerable amount of noise in the vibration signal of the mechanical system, if the component is damaged slightly, the energy of the vibration signal generated will be small, so it will be covered by the noise and difficult to detect, only when the severe damage occurs The energy of the vibration signal generated by it becomes larger, and it is obviously larger than the energy of noise vibration. However, at this time, the damage of the component has become serious, so the mechanical system may cause serious failure and stop operation at any time. This phenomenon is the difficulty faced by general spectrum analyzers when analyzing mechanical vibration signals, and it is also a major lack of damage and early warning effects in practical use. If the vibration caused by the damage of the component and the modulation caused by the resonance of the mechanical system 1228646, the bearing component is taken as an example to explain the implementation result of the demodulation variable function and the two-dimensional spectrum effect. The fourth figure is normal. Comparing the three-dimensional spectrum of the bearing with the three-dimensional spectrum of the fifth, sixth, and seventh damaged bearings, it can be found that the fourth diagram is a sharp spectrum and the other three diagrams are sheet-like spectrums. The difference lies in the impact of normal bearings. The energy comes from the non-true circle of the geometric shape. Generally speaking, the energy it can excite is mostly distributed at low frequency and is small. Therefore, the energy of the higher frequency band signal acquired during demodulation is quite low and insufficient. To fully excite the resonance of the mechanical system. However, for damaged bearings, the vibrational striking energy is derived from the damage of the components. Generally speaking, the energy bandwidth that it can excite is wide and strong. Therefore, its energy will completely excite the resonance of the mechanical system, thus generating a sheet. The spectrum resembles the resonance mode of a system. In summary, the embodiments of the present invention can indeed achieve the expected use effect, and the specific structure disclosed has not only been seen in similar products, nor has it been disclosed before the application. It has fully complied with the provisions of the Patent Law. In accordance with the law, the application for an invention patent was submitted in accordance with the law, and it is imperative that it be examined and the patent granted. 1〇1228646 [Simplified description of the figure] The first picture: the exponential attenuation function form represented by the percussion signal. The second picture: a schematic diagram of the modulation from the damage diagnosis signal of hundreds of Hz to the thousands of Hz. Figure: Schematic diagram of the result of applying the present invention to the modulation signal of the second figure. Fourth figure: Three-dimensional spectrum chart of normal bearings. Fifth figure. Three-dimensional spectrum chart of roller damaged bearings. (1) Sixth figure. Outer ring damaged bearings. Three-dimensional spectrum chart (two) Figure seven: Three-dimensional spectrum chart of damaged inner ring bearing (three)

Claims (1)

1228646 One tear) 'Dagger ten, M + 2K + 1 = E2 two 3 (Soap) Λ + 1. Where 3 () means spectrum analysis of vector sequences. 3. The signal processing method for the diagnosis of the operating conditions of the mechanical system components as described in item 1 of the scope of the patent application, wherein: the proportional parameter α can be set to a range of [〇.〇〇5, 〇 · 〇ΐ5] for adjusting an appropriate Bandpass band attenuation slope; The setting of the bandpass band width of Λ is related to the system characteristics, and generally the setting range is [2000, 3000]; the above two parameters are selected fixed parameters when used; as for the bandpass of Λ The change of the center frequency of the frequency band means that different frequency bands in the resonance mode of the mechanical system are selected. For the range of general vibration signal measurement, you can choose [300, 10000]; when X changes from low frequency to high frequency, The vibration energy of the modulation signal also increases and decreases with the resonance mode of the system. Therefore, if the vibration spectrum of the demodulated signal is added to this dimension, it can provide the information of the resonance mode of the mechanical system at the same time. A convenient and clear spectrum mode judgment for the user. 4. · The method of signal processing and execution of the second section of the system for the operation of the equipment, such as the application method, where: Proportion > The number α may be 5 and the range is [〇 · 〇05, 〇〇15] To adjust the slope of an appropriate bandpass band. The mosquitoes that enter the bandwidth of the bandpass band are related to the system characteristics. The general mosquito range is [chant 3000]; the above two parameters are selected when using Fixed parameters; 2 The change of the center frequency of the bandpass band of Yiyi means the selection of the common area system; / 'From: H-band: The range of general vibration signal measurement' can be selected [3000, 10000] Tian c •胄 彳 b to Xiao Xi ’s tuning of the signal ’s vibrational energy i will increase and decrease, so I ’ll take care of it here — mention = / 4 * 1228646 system resonance mode information, and because of the three-dimensional vibration spectrum will provide users with A convenient and clear visual mode judgment. One * ViV 1228646 issued the patent specification (the format, order and bold text of this specification, please do not change it arbitrarily, ※ please do not fill in the marked part) ※ application number: ※ application date: 72 9 7 XIPC classification: ^ 4 / ( 70 I. Title of the invention: (Chinese // English) A signal processing method and its computational algorithm in the diagnosis of component defects for mechanical systems 贰, Applicant: (Total 1 Person) Name or name: (Chinese / English) Southern Taiwan University of Technology Representative: (Chinese / English) Zhang Xinxiong / CHANG, HSI HSIUNG Address of residence or business office: (Chinese / English) Yongkang, Tainan County No. 1, Nantai St., Yongkang City, Tainan County 710 Nationality: (Chinese / English) Republic of China / TW Participants, Inventors: (1 person) Name: (Chinese / English) Shen Yutai / SHEN, YU TAI Address: (Chinese / English) No. 145 Ren He Road, Tain, East District, Konan City an Nationality: (Chinese / English) Republic of China / TW 1228646. It can be found that the vibration signal of the component damage of the mechanical system can be expressed as v (t) = d (t) · s⑴ + n (t) where d⑴ is the component damage The function of the periodic tapping sequence signal is generally about several hundred Hz; s (t) is a time function of the resonance of the mechanical system, and its frequency is generally about several thousand Hz or more, and n (t) is noise. As a function of time, its frequency energy is generally distributed below 2 kilohertz. As shown in the first figure, the tap signal can be expressed as an exponential decay function. From the figure, it can be seen that this modulation phenomenon can be about several hundred Hz. The damage diagnosis signal is modulated to thousands of Hz. Therefore, if the mechanical vibration signal v (t) can be filtered by a band-pass filter (selecting a frequency band above 2 kilohertz) to remove noise n (t), only the modulation can be captured. Signal, and greatly reduce the interference caused by noise; and then demodulate this modulation signal to restore the knock signal of damaged components. Traditionally, the method of completing the above-mentioned demodulation can be divided into hard There are two methods: hardware and software. The method of hardware demodulation is to capture After the mechanical vibration signal passes through the band-pass filter, the signal can be demodulated by passing the signal through a low-pass filter of 1 kHz. The band-pass filter and the low-pass filter used are both implemented in hardware. The demodulated signal obtained by this method will have a considerable degree of distortion. As for the software demodulation method, the captured mechanical vibration signal is input into the computer through analog to digital method, and then the captured signal is band-pass filtered by digital filtering method or fast Fourier transform, and then the band-pass filtering is performed. The latter signal is converted by Hilbert to obtain the final demodulated signal. However, this method has a considerable degree of confusion. If the digital filtering method is used, the signal will also have a considerable degree of distortion, and for the selection of different band-pass frequency bands, the digital filter design also has a number of parameters that need to be set. The software method is used to automatically design the digital filter. Although it is more convenient to use fast Fourier transform, the problems caused by it will also cause signal distortion. In addition, for the digital filtering method or the fast Fourier transform, the Hilbert transformation used will also have many parameter setting problems and transformations with the aforementioned digital filter design. 6 228646 ()) Leakage problems occur. [Summary of the Invention] The present invention is mainly to provide a signal processing method for diagnosing the operating conditions of mechanical system components, using a demodulation analysis function: where: 乂 is the center frequency of the band-pass band; Λ is the bandwidth of the band-pass band; α is The scaling parameter is used to adjust the attenuation slope of the band-pass band (siope). And the demodulated signal I is expressed as the absolute of the convolution (c0nv〇luti〇n) ~ as each: (f ~ Γ) do β to increase the de-vibration of the deviating bottle_spectrum and add dimensionality to remove impurities The age-resonance M-type message, and the presence of a two-dimensional vibrating vehicle will provide a convenient and clear spectrum mode judgment for the user. [Embodiment] The technical means, inventive features, inconsistent purposes, and effects of the present invention are easy to understand, and are illustrated with the following drawings: The present invention uses a demodulation analysis function to apply Kada to vibration signals. The demodulation changes are: (α) (strict (/ Monument) today), where: / c is the center frequency of the band-pass band; Λ is the bandwidth of the band-pass band; α is a proportional parameter that can be used to adjust the band-pass scale The attenuation slope is emissive. 1228646 4 # Form feed 焯: ：： year-month said I and the demodulation signal can be expressed as the absolute value of V (t) and / ^ Λ (0 convolution), that is, in practical application: the proportional parameter α can be set The range is [0 · 005, 0.015], which is used to adjust an appropriate attenuation slope of the band-pass frequency band; and the setting of the band-pass frequency band of Λ is related to the system characteristics. Generally, the settable range is [2000, 3000]. The above two The parameters are selected fixed parameters in use. As for the change of the center frequency of the bandpass band, it means that different frequency bands in the resonance mode of the mechanical system are selected. For the range of general vibration signal measurement, you can choose [3〇 〇 () 10000]. Pan Tian's crimes Ai ~ Yu and the two Ij tanzan change the vibration energy of the signal along with the resonance mode of the system, so if the vibration spectrum of the demodulated signal plus this one dimension can be simultaneously Provide the information of the resonance mode of the mechanical system, and because the three-dimensional vibration spectrum is presented, it can be used = a visual and convenient and clear spectrum mode judgment. In order to implement the above theoretical analysis in a computer system, the following will propose its implementation method Bekai1 · Based on the specifications of the computer, the sampling time τ of the data capture and the range of the parameter center frequency / c, then the demodulation function with a waveform function length of 2KT \ / c'J = point V2 is called today '广 以 令 打） Ding Wei, where none (-KShK) is the position of a point in the demodulation analysis function graph 2 center rate / c Fan 迩 take (㈣) two equal frequency intervals, namely a / + wash / 0 + «# ， guankuan Λ is a fixed value, then the demodulation analysis analysis can be expressed as ..., 1228646 Μ, 〇UK) Uk + i) h ⑽ :: WxW Κι ^ Λ ^ Z ^^ (K ) Α If κ) person ♦ pair, ... ~ tree ,,,) κ AK where the state represents the demodulation analysis vector of the A: frequency band 3. Pair-mechanical vibration signal sequence v㈣, the demodulation analysis can be Represented as' A * abs (4 * v) Υη + Ι, Μ 2 2Κ + 1 = A ei = abs (4 * v) k + ij _abs (An + i * v) _ where A means that f has no frequency band The obtained demodulated signal, * represents the vector accumulative operation, and abs〇 is to take the absolute value of each element in its vector. 4 · Finally, the spectrum analysis is performed on the demodulated signal, and the three-dimensional spectrum can be expressed as Γ * , Ei -¾) '^ n + l, M + 2K + 1 ~ Ei = 3 (¾) Λ + ι. Among them, 3 (·) means taking a spectrum analysis of the vector sequence. Please refer to the first figure again for the form of the exponential decay function represented by the tap signal. The second figure is the modulation Schematic diagram of thousands of Ήζ modulated by a damage diagnosis signal with a frequency of several hundred Hz; the third diagram illustrates the result of the theoretical analysis of the present invention applied to the modulation signal of the second diagram. The thick line is The thin line is the original damage percussion function [same as the first picture]. As known from the third picture, the modulation analysis function of the present invention has a fairly perfect signal demodulation variable energy Cl η--; j ί £ | Patent scope: 1. A signal processing method for the diagnosis of the operating conditions of mechanical system components, including: a demodulation analysis function: h 'a, fc, fw it)] πί ^-(-) 2 ^ -y2λ · (/ c ~ ^ γ) {-/ 2π (/ c + KW) ί) where: meaning the center frequency of the band-pass band; / w is the bandwidth of the band-pass band; α is the proportional parameter that can be used to adjust the attenuation slope of the band-pass band (Slope attenuation); and the demodulated signal can be expressed as the absolute value of v (t) and / ^ / fr (〇 convolution) That ⑺ (, - Γ) to do. 2 · —Signal processing and execution methods for operating condition diagnosis of mechanical system components, including: Κ, fc, fw ,, 1 · (―) 2 -7 * 2λ · Λ (e 2 jnt V.2 / r (/ c + quart: 乂 is the center frequency of the band-pass band; Λ is the bandwidth of the band-pass band; α is the proportional parameter which can be used to adjust the slope attenuation of the band-pass band; and the demodulation signal can be expressed as v (t ) And / ^ Λ (〇 convolution) absolute value, that is, the above theoretical analysis is implemented in a computer system with the following algorithm: Set the sampling time τ, 彖 α and center frequency of data acquisition based on the computer's specifications and characteristics / e, the demodulation conversion of the length of the job list is 2K ha ha ccJw {k) = T ^ f6 ^) 2 (eJ2 ^) kT ^ iJ2; r (fc + T) k \, where A :( -Ky SK) is the position of a certain point in the demodulation analysis function graph. The center frequency / c depends on (n + 1) equal solution intervals, that is, ^ / 〇phen, Λ + 2 is introduced /. +, And The bandwidth A is a fixed value, and the light moment analysis of the moment is $... Ι (η + 1) χ (2Κ + 1) Two t Two: > (Κ) Others, people (κ) Λμ ^ ( -κ) haJMf ^ K + l) ^ h \ Κ lK + i where the state represents the λ: th frequency band Demodulation analysis vector, for a sequence of mechanical vibration signals), 〇 ～ Μ, the demodulation variational fold ^ 亀 abs (^ * v) y η + 1, Μ + 2Κ + 1 ~ A ei • = abs (ii2 * v) A Λ + ι_ .absA + i * v) 11 can be expressed as β which is shown by ㈣, and _E1 is counted, and abs () is the value of each element in its vector. Absolute value; spectrum analysis is performed on demodulated signals at the end of Xia Zheyun, and the three-dimensional spectrum can be expressed as 13