Diesel engine cylinder cover vibration signal inversion virtual key phase pulse and whole period feature extraction method

Abstract

The invention discloses a diesel engine cylinder cover vibration signal inversion virtual key phase pulse and whole period feature extraction method, and belongs to the field of mechanical signal processing. Collecting a diesel engine cylinder cover vibration signal; based on the autocorrelation function of the signal smooth envelope energy gradient, identifying the whole-cycle signal length of the diesel engine; judging whether a reference signal exists or not, if yes, identifying a key phase triggering position based on interpolation alignment and a cross-correlation function, and if not, intercepting a single full-period signal and storing the single full-period signal as the reference signal; inverting the virtual key phase pulse signal so as to intercept a required full-period signal; typical signal features are extracted, and a whole-period signal feature parameter set is constructed. The method overcomes the problem of difficult key phase pulse signal acquisition caused by difficult installation of a key phase sensor on a high-power-density diesel engine, fully considers the quasi-periodic characteristics of the working cycle of the diesel engine, can accurately identify the whole cycle length, efficiently inverts the virtual key phase pulse signal, and improves the accuracy of the key phase pulse signal. And full-period feature extraction of the vibration signal of the diesel engine under the keyless phase sensor is realized.

Description

Technical Field

[0001] The invention relates to the field of mechanical signal processing, and in particular to a method for extracting virtual key phase pulses and whole cycle features from a diesel engine cylinder head vibration signal inversion. Background Art

[0002] As an important power device, high power density diesel engines play an important role in key fields such as energy and electricity, transportation, and defense industry. Especially under complex and changeable working conditions, diesel engine heavy parts such as crank-connecting rod mechanism, crankshaft, transmission system and other components frequently fail, which will lead to huge economic losses and even casualties. Therefore, it is particularly important to monitor the condition of diesel engines.

[0003] Cylinder head vibration signal monitoring is one of the real-time status monitoring methods for diesel engines. Since the diesel engine working cycle has quasi-periodic characteristics, full-cycle identification and interception are necessary prerequisites for cylinder head vibration signal feature extraction. The alternating ignition of each cylinder of the diesel engine makes the crankshaft speed unstable, and there are periodic fluctuations within and outside each working cycle. Under the condition of changing and unpredictable working conditions, it is unrealistic to obtain the full-cycle vibration signal of the cylinder head and gearbox in a time-varying state by time interval. In order to ensure that the vibration signal is synchronized with the crankshaft angle for full-cycle acquisition, the traditional method needs to use a key phase block, or additional equipment such as an optical pulse encoder, an electronic phase-locked loop multiplier, etc. to generate a pulse signal to achieve synchronous acquisition, but for high-power-density diesel engines, the traditional method has problems such as high cost and difficult installation. In recent years, some scholars have proposed a method for acquiring a keyless whole-cycle signal through the vibration signals of the cylinder head and gearbox. The method first extracts the instantaneous frequency from the gearbox vibration signal based on the synchronous compression generalized S transform to estimate the speed and the whole-cycle length; then uses the Teager energy operator to enhance the impact characteristics of the cylinder head vibration signal; finally, uses the adaptive correlation analysis method to accurately calculate the whole-cycle length through a variable-length sliding combination window. However, since this method needs to use the time-frequency transformation method to process the gearbox vibration signal, it is easy to cause inaccurate speed estimation when the background noise is large. At the same time, this method also has problems such as more input signal types, more prior parameters, and low calculation efficiency. Therefore, in view of the limitations of the existing methods, the present invention discloses a method for inverting virtual key phase pulses and whole-cycle feature extraction of diesel engine cylinder head vibration signals, which can realize accurate and efficient inversion of diesel engine virtual key phase pulse signals and whole-cycle feature extraction based only on the cylinder head vibration signal. Summary of the invention

[0004] In order to solve the problems that it is difficult to install key phase sensors on high power density diesel engines, resulting in difficulty in obtaining key phase signals, and the existing technology has many input signal types, many prior parameters, and low calculation efficiency, the present invention proposes a method for inverting virtual key phase pulses and whole cycle features of diesel engine cylinder head vibration signals. The method fully considers the quasi-periodic characteristics of the diesel engine working cycle, can accurately identify the whole cycle length, efficiently invert virtual key phase pulse signals, and realize the whole cycle feature extraction of diesel engine vibration signals without key phase sensors.

[0005] The present invention provides a method for extracting virtual key phase pulses and whole cycle features from inversion of diesel engine cylinder head vibration signals, which is characterized by comprising the following steps:

[0006] S1: Collect diesel engine cylinder head vibration signal x(t), where t is the sampling time;

[0007] S2: Identify the length of the diesel engine full cycle signal based on the autocorrelation function of the signal smoothed envelope energy gradient (SEEG);

[0008] S3: Determine whether there is a reference signal. If so, identify the key phase trigger position based on interpolation alignment and cross-correlation function. If not, first intercept a single full-cycle signal and store it as a reference signal, and then identify the key phase trigger position based on interpolation alignment and cross-correlation function.

[0009] S4: Invert the virtual key phase pulse signal to intercept the required full cycle signal;

[0010] S5: Extract typical signal features and construct a whole-cycle signal feature parameter set;

[0011] In step S2, the identification of the diesel engine full cycle signal length based on the autocorrelation function of the signal smoothed envelope energy gradient (SEEG) includes the following steps:

[0012] S201: Loading cylinder head vibration signal x(t);

[0013] S202: Calculate the envelope of x(y), the calculation formula is as follows:

[0014]

[0015] In the formula, represents the Hilbert envelope, pv represents the Cauchy principal value, t=1,2,…,N, N represents the number of sampling points of the signal x(t), τ represents the time delay, x(τ) represents the value of the signal x(t) at time τ, and X(t) represents the envelope of the signal x(t);

[0016] S203: Calculate the energy gradient x of X(t) EEG :

[0017] x EEG (i) = (X(i+1)) 2 -(X(i)) 2

[0018] In the formula, x EEG (i) represents the energy gradient of X(t), i = 1, 2, ..., N-1;

[0019] S204: Set the sliding window size W and calculate the smoothed envelope energy gradient x(t) SEEG , the calculation formula is as follows:

[0020]

[0021] In the formula, x SEEG (i) represents the i-th sliding average, x EEG (n) represents x EEG The nth sample value in , W represents the sliding window size;

[0022] S205: Calculate x SEEG The autocorrelation function R xx (τ), the calculation formula is as follows:

[0023]

[0024] In the formula, R xx (τ) represents x SEEG The autocorrelation function of SEEG The sequence length, τ represents the delay;

[0025] S206: Use the "findpeaks" function in MATLAB to obtain R xx (τ), after sorting the local maxima in descending order, take the interval between the indexes corresponding to the first two local maxima as the whole cycle length T of the signal x(t);

[0026] In one embodiment of the present invention, in step S3, the determination of whether a reference signal exists, if so, identifying the key phase trigger position based on interpolation alignment and cross-correlation function, if not, first intercepting a single full cycle signal and storing it as a reference signal, and then identifying the key phase trigger position based on interpolation alignment and cross-correlation function, includes the following steps:

[0027] S301: According to the sequence variable x r (t) Determine whether there is a reference signal, sequence variable x r(t) represents a full-cycle cylinder head vibration signal, which can be obtained in two ways, including: a reference signal stored after running the complete algorithm flow of the present invention, or a full-cycle cylinder head vibration signal collected and intercepted by the diesel engine under the condition of key phase triggering. The method for determining whether a reference signal exists is as follows:

[0028] If the sequence variable x r (t) is empty, indicating that the reference signal does not exist, and steps S302-S306 are executed. If the sequence variable x r (t) is not empty, indicating that the reference sequence exists, and there is no need to execute S302, and steps S303-S306 are directly executed;

[0029] S302: Cut x(t) with length T to obtain a full-cycle signal and store it in the sequence variable x r (t) is used as the reference signal;

[0030] S303: Cut x(t) with length T to obtain the full-period signal x p (t);

[0031] S304: Based on the spline interpolation method, that is, the "interp1" function in MATLAB, the whole periodic signal x p (t) and the reference signal x r The number of points of (t) is aligned, and the number of sampling points of the two signals after alignment is N r ;

[0032] S305: Calculate x p (t) and x r The cross-correlation function R of (t) xy (τ), calculated as follows:

[0033]

[0034] In the formula, R xy (τ) represents x p (t) and x r (t), N r Represents x p (t) and x r (t) is the number of sampling points, τ represents the time delay;

[0035] S306: Take R xy The time corresponding to the maximum value of (τ) is the time when the virtual key phase is triggered after the point alignment T s After scaling, the actual virtual key phase triggering time T is obtained start , the scaling formula is as follows:

[0036]

[0037] Where, T start Indicates the actual virtual key phase triggering time, T s Indicates the virtual key phase triggering moment after the whole cycle signal is aligned with the points, N single represents the full period signal x obtained in step S303 p The number of sampling points of (t), N r Represents the reference signal x r (t) The number of sampling points;

[0038] In step S4, the inverted virtual key phase pulse signal is used to intercept the required full cycle signal, including the following steps:

[0039] S401: Based on the obtained virtual key phase triggering time T start The whole period length T of the sum signal x(t) is used to invert the virtual key phase pulse signal. The inversion formula is as follows:

[0040]

[0041] In the formula, x KEY (n) represents the inverted virtual key phase pulse signal, T start represents the virtual key phase triggering moment, n represents the time index of the signal, k represents the integer sequence from 0 to K, t=0,1,2,…,K, δ represents the Kronecker delta function, which is defined as follows:

[0042]

[0043] Where p represents the time index of δ;

[0044] K represents the total number of full-cycle signals that can be intercepted from the signal x(t). The calculation formula of K is as follows:

[0045]

[0046] In the formula, floor(·) represents the floor function, which can be implemented using the floor function in MATLAB, N represents the number of sampling points of the signal x(t), and T represents the length of the integer cycle of x(t);

[0047] S402: Based on the virtual key phase pulse signal x KEY , intercept and obtain the full-cycle signal of key phase triggering from x(t) in turn;

[0048] In one embodiment of the present invention, in step S5, extracting typical signal features and constructing a whole-cycle signal feature parameter set includes the following steps:

[0049] S501: Calculate the typical time domain characteristics and frequency domain characteristics of the intercepted full-cycle signal. The time domain characteristics include the peak value X max , mean X mean , effective value X rms , standard deviation X std , Kurtosis X kurtosis , peak factor X cf , waveform factor X wf , the frequency domain features include the center frequency F cf , RMS frequency F mf , the calculation formula is as follows:

[0050] X max =max(x p )

[0051]

[0052]

[0053] In the formula, max(·) represents the maximum value function, x p Represents a full-cycle signal, N single Represents x p The number of sampling points, f(p) represents the signal x p The frequency value of the pth frequency point after fast Fourier transform, A(p) represents the signal x p The amplitude of the pth frequency point after fast Fourier transform, P represents half of the data points in the fast Fourier transform result (single-sided spectrum);

[0054] S502: Record characteristic parameters of all integer periodic signals and store them as an integer periodic signal characteristic parameter set.

[0055] The present invention aims to solve the problems that it is difficult to install key phase sensors on high power density diesel engines, which leads to difficulty in obtaining key phase signals, the existing methods have many prior parameters and low computational efficiency, and proposes a method for inverting virtual key phase pulses and extracting whole cycle features of diesel engine cylinder head vibration signals. The method accurately describes the energy change characteristics of the impact component in the cylinder head vibration signal and calculates the smooth envelope energy gradient of the signal. The method uses the periodicity of the diesel engine working cycle to calculate the autocorrelation function of the smooth envelope energy gradient and identify the whole cycle length of the signal. Furthermore, the key phase trigger position is identified based on interpolation alignment and cross-correlation functions, and the virtual key phase pulse signal is inverted to realize the whole cycle signal feature extraction of virtual key phase triggering without key phase sensors. The present invention breaks through the dependence of whole cycle signal interception on key phase sensors, and has broad application prospects and promotion significance in the fields of diesel engine real-time status monitoring and vibration signal processing.

[0056] The beneficial technical effects of the present invention are:

[0057] 1. The whole-cycle identification method of diesel engine cylinder head vibration signal based on signal smoothed envelope energy gradient (SEEG) autocorrelation proposed in the present invention comprehensively considers the energy change characteristics and quasi-periodicity of the impact component. The whole-cycle length of the cylinder head vibration signal can be accurately identified through an efficient calculation method, which effectively improves the preprocessing efficiency of diesel engine vibration signal analysis.

[0058] 2. The key phase trigger position identification method based on spline interpolation and cross-correlation calculation proposed in the present invention can ensure that the full cycle signal is intercepted at the same crankshaft angle position each time it runs, effectively solving the phase matching problem of virtual key phase triggering.

[0059] 3. The key phase pulse inversion and whole cycle feature extraction method proposed in the present invention can efficiently and accurately invert the virtual key phase pulse signal based on the Kronecker delta function, which solves the key phase information loss problem caused by the difficult arrangement of key phase sensors in high power density diesel engines. The key phase pulse can be accurately inverted based on the cylinder head vibration signal, thus getting rid of the dependence on the key phase sensor. In addition, the time and frequency domain features extracted from the intercepted whole cycle signal will lay the foundation for diesel engine status monitoring and intelligent operation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0061] Figure 1 It is a flow chart of a method for extracting virtual key phase pulses and whole cycle features from a diesel engine cylinder head vibration signal inversion provided by an embodiment of the present invention;

[0062] Figure 2 is a time domain diagram of a diesel engine cylinder head vibration signal provided by an embodiment of the present invention;

[0063] Figure 3 is a diagram of smoothed envelope energy gradient (SEEG) calculation results provided by an embodiment of the present invention;

[0064] Figure 4 is a graph of autocorrelation calculation results of a smoothed envelope energy gradient (SEEG) provided in an embodiment of the present invention;

[0065] Figure 5 is a time domain diagram of a reference signal provided by an embodiment of the present invention;

[0066] Figure 6 is a result diagram of a virtual key phase pulse signal obtained by inversion in an embodiment of the present invention;

[0067] Figure 7 This is a graph of the whole-cycle signal result obtained by using virtual key phase triggering to intercept in an embodiment of the present invention. DETAILED DESCRIPTION

[0068] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0069] Figure 1 This is a flow chart of a method for extracting virtual key phase pulses and full cycle features from a diesel engine cylinder head vibration signal inversion provided by an embodiment of the present invention. Figure 1 The diesel engine cylinder head vibration signal inversion virtual key phase pulse and full cycle feature extraction method includes the following steps:

[0070] S1: Collect diesel engine cylinder head vibration signal x(t), where t is the sampling time;

[0071] S2: Identify the length of the diesel engine full cycle signal based on the autocorrelation function of the signal smoothed envelope energy gradient (SEEG);

[0072] S3: Determine whether there is a reference signal. If so, identify the key phase trigger position based on interpolation alignment and cross-correlation function. If not, first intercept a single full-cycle signal and store it as a reference signal, and then identify the key phase trigger position based on interpolation alignment and cross-correlation function.

[0073] S4: Invert the virtual key phase pulse signal to intercept the required full cycle signal;

[0074] S5: Extract typical signal features and construct a full-cycle signal feature parameter set.

[0075] In one embodiment of the present invention, in step S2, the identification of the diesel engine full cycle signal length based on the autocorrelation function of the signal smoothed envelope energy gradient (SEEG) comprises the following steps:

[0076] S201: Loading cylinder head vibration signal x(t);

[0077] S202: Calculate the envelope of x(t), the calculation formula is as follows:

[0078]

[0079] In the formula, represents the Hilbert envelope, pv represents the Cauchy principal value, t=1,2,…,N, N represents the number of sampling points of the signal x(t), τ represents the time delay, x(τ) represents the value of the signal x(t) at time τ, and C(t) represents the envelope of the signal x(t);

[0080] S203: Calculate the energy gradient x of X(t) EEG :

[0081] x EEG (i) = (X(i+1)) 2 -(X(i)) 2

[0082] In the formula, x EEG (i) represents the energy gradient of X(t), i = 1, 2, ..., N-1;

[0083] S204: Set the sliding window size W and calculate the smoothed envelope energy gradient x(t) SEEG , the calculation formula is as follows:

[0084]

[0085] In the formula, x SEEG (i) represents the i-th sliding average, x EEG (n) represents x EEG The nth sample value in , W represents the sliding window size;

[0086] It should be noted that the sliding window size W considered can be adjusted according to the actual input cylinder head vibration signal. Preferably, W=70;

[0087] S205: Calculate x SEEG The autocorrelation function R xx (τ), the calculation formula is as follows:

[0088]

[0089] In the formula, R xx (τ) represents x SEEG The autocorrelation function of SEEG The sequence length, τ represents the delay;

[0090] S206: Use the "findpeaks" function in MATLAB to obtain R xx (τ), after sorting the local maxima in descending order, take the interval between the indexes corresponding to the first two local maxima as the whole cycle length T of the signal x(t);

[0091] In one embodiment of the present invention, in step S3, the determination of whether a reference signal exists, if so, identifying the key phase trigger position based on interpolation alignment and cross-correlation function, if not, first intercepting a single full cycle signal and storing it as a reference signal, and then identifying the key phase trigger position based on interpolation alignment and cross-correlation function, includes the following steps:

[0092] S301: According to the sequence variable xr (t) Determine whether there is a reference signal, sequence variable x r (t) represents a full-cycle cylinder head vibration signal, which can be obtained in two ways, including: a reference signal stored after running the complete algorithm flow of the present invention, or a full-cycle cylinder head vibration signal collected and intercepted by the diesel engine under the condition of key phase triggering. The method for determining whether a reference signal exists is as follows:

[0093] If the sequence variable x r (t) is empty, indicating that the reference signal does not exist, and steps S302-S306 are executed. If the sequence variable x r (t) is not empty, indicating that the reference sequence exists, and there is no need to execute S302, and steps S303-S306 are directly executed;

[0094] S302: Cut x(t) with length T to obtain a full-cycle signal and store it in the sequence variable x r (t) is used as the reference signal;

[0095] S303: Cut x(t) with length T to obtain the full-period signal x p (t);

[0096] S304: Based on the spline interpolation method, that is, the "interp1" function in MATLAB, the whole periodic signal x p (t) and the reference signal x r The number of points of (t) is aligned, and the number of sampling points of the two signals after alignment is N r ;

[0097] S305: Calculate x p (t) and x r The cross-correlation function R of (t) xy (τ), calculated as follows:

[0098]

[0099] In the formula, R xy (τ) represents x p (t) and x r (t), N r Represents x p (t) and x r (t) is the number of sampling points, τ represents the time delay;

[0100] S306: Take R xy The time corresponding to the maximum value of (τ) is the time when the virtual key phase is triggered after the point alignment T s After scaling, the actual virtual key phase triggering time T is obtained start , the scaling formula is as follows:

[0101]

[0102] Where, T start Indicates the actual virtual key phase triggering time, T s It represents the virtual key phase triggering time after the whole cycle signal is aligned with the points, S single represents the full period signal x obtained in step S303 p The number of sampling points of (t), N r Represents the reference signal x r (t) is the number of sampling points.

[0103] In one embodiment of the present invention, in step S4, the inverting virtual key phase pulse signal to intercept the required full cycle signal includes the following steps:

[0104] S401: Based on the obtained virtual key phase triggering time T start The whole period length T of the sum signal x(t) is used to invert the virtual key phase pulse signal. The inversion formula is as follows:

[0105]

[0106] In the formula, x KEY (n) represents the inverted virtual key phase pulse signal, T start represents the virtual key phase triggering moment, n represents the time index of the signal, k represents an integer sequence from 0 to K, k = 0, 1, 2, ..., k, δ represents the Kronecker delta function, which is defined as follows:

[0107]

[0108] Where p represents the time index of δ;

[0109] K represents the total number of full-cycle signals that can be intercepted from the signal x(t). The calculation formula of K is as follows:

[0110]

[0111] In the formula, floor(·) represents the floor function, which can be implemented using the floor function in MATLAB, N represents the number of sampling points of the signal x(t), and T represents the length of the integer cycle of x(t);

[0112] S402: Based on the virtual key phase pulse signal x KEY , intercept and obtain the full-cycle signal of key phase triggering from x(t) in turn;

[0113] In one embodiment of the present invention, in step S5, extracting typical signal features and constructing a whole-cycle signal feature parameter set includes the following steps:

[0114] S501: Calculate the typical time domain characteristics and frequency domain characteristics of the intercepted full-cycle signal. The time domain characteristics include the peak value X max , mean X mean , effective value X rms , standard deviation X std , Kurtosis X kurtosis , peak factor X cf , waveform factor X wf , the frequency domain features include the center frequency F cf , RMS frequency F mf , the calculation formula is as follows:

[0115] X max =max(x p )

[0116]

[0117]

[0118] In the formula, max(·) represents the maximum value function, x p Represents a full-cycle signal, N single Represents x p The number of sampling points, f(p) represents the signal x p The frequency value of the pth frequency point after fast Fourier transform, A(p) represents the signal x p The amplitude of the pth frequency point after fast Fourier transform, P represents half of the data points in the fast Fourier transform result (single-sided spectrum);

[0119] S502: Record characteristic parameters of all integer periodic signals and store them as an integer periodic signal characteristic parameter set.

[0120] Taking a certain model of diesel engine as an example, in order to fully demonstrate the complete content of the present invention, a case with a reference signal is selected as an embodiment. The diesel engine cylinder head vibration signal x(t) is collected under the condition of a speed of 1500rpm, the sampling frequency is 51.2kHz, and the number of sampling points is 30k. Figure 2 The time domain diagram of the collected diesel engine cylinder head vibration signal is shown. As can be seen from the figure, the time domain signal exhibits obvious quasi-periodic characteristics.

[0121] The method of the present invention is used to analyze and process the collected cylinder head vibration signal. According to step S2, first, the smoothed envelope energy gradient (SEEG) of the cylinder head vibration signal is calculated. The sliding window size W is set to 70, and the calculated smoothed envelope energy gradient result is as follows: Figure 3As shown in the figure, it can be seen that SEEG can well locate the impulse component in the signal, highlighting the quasi-periodic characteristics. Then, the autocorrelation function of SEEG is calculated, and the calculation result is as follows: Figure 4 As shown in the figure, the autocorrelation function presents equally spaced local peaks, indicating that SEEG has strong periodicity. Finally, the local maximum of the autocorrelation function is retrieved, and after the local maximum is sorted in descending order, the interval between the indexes corresponding to the first two local maximums is taken as the entire period length T = 4088 of the signal x(t). According to step S3, first, it is determined whether there is a reference signal. In this embodiment, there is a reference signal x r (t), x r (t) is the full cycle signal of diesel engine cylinder head vibration collected at a speed of 1000 rpm, such as Figure 5 Secondly, based on the interpolation alignment and cross-correlation function, the key phase trigger position is identified and the virtual key phase trigger time T is calculated. start =3639. According to step S4, first, based on the obtained virtual key phase triggering time T start The whole cycle length T of the sum signal x(t) is used to invert the virtual key phase pulse signal, such as Figure 6 Secondly, based on the virtual key phase pulse signal x KEY , intercept and obtain the key-phase triggered full-cycle signals from x(t) in turn, and intercept 6 full-cycle signals in total, Figure 7 One of the whole periodic signals is shown. According to step S5, the typical time domain features and frequency domain features of the six intercepted whole periodic signals are calculated, a total of 9 features, the calculation results are shown in Table 1, and the feature extraction of the whole periodic signal and the construction of the feature parameter set are realized.

[0122] Table 1 Characteristic parameter set of 6 full-cycle signals intercepted by key phase trigger

[0123]

[0124] Based on the above embodiments, it can be seen that the present invention overcomes the problem that it is difficult to install a key phase sensor on a high power density diesel engine, resulting in difficulty in obtaining a key phase pulse signal. It fully considers the quasi-periodic characteristics of the diesel engine working cycle, can accurately identify the length of the entire cycle, and efficiently invert the virtual key phase pulse signal, thereby realizing the extraction of the entire cycle features of the diesel engine vibration signal without a key phase sensor.

[0125] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope involved in the present application is not limited to the technical solution formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the inventive concept, such as a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in this application.

[0126] Except for the technical features described in the specification, the remaining technical features are technologies known to those skilled in the art. In order to highlight the innovative features of the present invention, the remaining technical features will not be described here in detail.

Claims

1. A method for extracting virtual key phase pulses and full cycle features from inversion of diesel engine cylinder head vibration signals, characterized in that: The following steps are involved: S1: Collect diesel engine cylinder head vibration signal x(t), where t is the sampling time; S2: Identify the length of the diesel engine full cycle signal based on the autocorrelation function of the signal smoothed envelope energy gradient (SEEG); S3: Determine whether there is a reference signal. If so, identify the key phase trigger position based on interpolation alignment and cross-correlation function. If not, first intercept a single full-cycle signal and store it as a reference signal, and then identify the key phase trigger position based on interpolation alignment and cross-correlation function. S4: Invert the virtual key phase pulse signal to intercept the required full cycle signal; S5: Extract typical signal features and construct a whole-cycle signal feature parameter set; Wherein, in step S2, the identification of the diesel engine full cycle signal length based on the autocorrelation function of the signal smoothed envelope energy gradient (SEEG) includes the following steps: S201: Loading cylinder head vibration signal x(t); S202: Calculate the envelope of x(t), the calculation formula is as follows: In the formula, represents the Hilbert envelope, pv represents the Cauchy principal value, t=1,2,...,N, N represents the number of sampling points of the signal x(t), τ represents the time delay, x(τ) represents the value of the signal x(t) at time τ, and X(t) represents the envelope of the signal x(t); S203: Calculate the energy gradient x of X(t) EEG : x EEG (i)=(X(i+1)) 2 -(X(i)) 2 In the formula, x EEG (i) represents the energy gradient of X(t), i = 1, 2, ..., N-1; S204: Set the sliding window size W and calculate the smoothed envelope energy gradient x(t) SEEG , the calculation formula is as follows: In the formula, x SEEG (i) represents the i-th sliding average, x EEG (n) represents x EEG The nth sample value in , W represents the sliding window size; S205: Calculate x SEEG The autocorrelation function R xx (τ), calculated as follows: In the formula, R xx (τ) represents x SEEG The autocorrelation function of SEEG The sequence length, τ represents the delay; S206: Use the "findpeaks" function in MATLAB to obtain R xx (τ), after sorting the local maxima in descending order, take the interval between the indexes corresponding to the first two local maxima as the whole cycle length T of the signal x(t); Among them, in step S3, the determination of whether there is a reference signal, if so, identifying the key phase trigger position based on interpolation alignment and cross-correlation function, if not, first intercepting a single full cycle signal and storing it as a reference signal, and then identifying the key phase trigger position based on interpolation alignment and cross-correlation function, includes the following steps: S301: According to the sequence variable x r (t) Determine whether there is a reference signal, sequence variable x r (t) represents a full-cycle cylinder head vibration signal, which can be obtained in two ways, including: a reference signal stored after the complete algorithm flow of the present invention has been run, or a full-cycle cylinder head vibration signal collected and intercepted by the diesel engine under the condition of key phase triggering; the method for determining whether there is a reference signal is as follows: If the sequence variable x r (t) is empty, indicating that the reference signal does not exist, and steps S302-S306 are executed. If the sequence variable x r (t) is not empty, indicating that the reference sequence exists, and there is no need to execute S302, and steps S303-S306 are directly executed; S302: Cut x(t) with length T to obtain a full-cycle signal and store it in the sequence variable x r (t) is used as the reference signal; S303: Cut x(t) with length T to obtain the full-period signal x p (t); S304: Based on the spline interpolation method, that is, the "interp1" function in MATLAB, the whole periodic signal x p (t) and the reference signal x r The number of points of (t) is aligned, and the number of sampling points of the two signals after alignment is N r ; S305: Calculate x p (t) and x r The cross-correlation function R of (t) xy (τ), calculated as follows: In the formula, R xy (τ) represents x p (t) and x r (t), N r Represents x p (t) and x r (t) is the number of sampling points, τ represents the delay; S306: Take R xy The time corresponding to the maximum value of (τ) is the time when the virtual key phase is triggered after the point alignment T s After scaling, the actual virtual key phase triggering time T is obtained start , the scaling formula is as follows: Where, T start Indicates the actual virtual key phase triggering time, T s Indicates the virtual key phase triggering moment after the whole cycle signal is aligned with the points, N single represents the full period signal x obtained in step S303 p The number of sampling points of (t), N r Represents the reference signal x r (t) The number of sampling points; Wherein, in step S4, the inversion of the virtual key phase pulse signal, thereby intercepting the required full cycle signal, comprises the following steps: S401: Based on the obtained virtual key phase triggering time T start The whole period length T of the sum signal x(t) is used to invert the virtual key phase pulse signal. The inversion formula is as follows: In the formula, x KEY (n) represents the inverted virtual key phase pulse signal, T start represents the virtual key phase triggering moment, n represents the time index of the signal, k represents an integer sequence from 0 to K, k = 0, 1, 2, ..., K, δ represents the Kronecker delta function, which is defined as follows: Where p represents the time index of δ; K represents the total number of full-cycle signals that can be intercepted from the signal x(t). The calculation formula of K is as follows: In the formula, floor(·) represents the floor function, which can be implemented using the floor function in MATLAB, N represents the number of sampling points of the signal x(t), and T represents the length of the integer cycle of x(t); S402: Based on the virtual key phase pulse signal x KEY , intercept from x(t) in turn and obtain the full-cycle signal triggered by the key phase.

Images