JP5511913B2 - Frequency component analyzer - Google Patents

Description

  The present invention relates to a frequency component analysis device that performs frequency component analysis on detected values of operating parameters of an internal combustion engine in synchronization with engine rotation.

  In Patent Document 1, an internal combustion engine is used by using a discrete Fourier transform algorithm (hereinafter referred to as “DFT algorithm”) and a fast Fourier transform algorithm (hereinafter referred to as “FFT algorithm”). A signal processing apparatus for performing frequency component analysis of an output signal of a mounted knock sensor is shown. Since the FFT algorithm employs a special algorithm to increase the calculation speed, the intensity of the necessary frequency component (specifically, the center frequency component of the knock sensor output signal) may not be obtained. Therefore, in the signal processing device, the intensity is calculated using the DFT algorithm for the center frequency component (the component having the maximum intensity) of the knock sensor output signal, and the intensity is calculated using the FFT algorithm for frequency components other than the center frequency component. Is calculated.

  Patent Document 2 discloses a technique for increasing the calculation speed when the engine rotational speed is high using the DFT algorithm.

JP-A-11-303673 Japanese Patent No. 4879329

  Since the frequency component analysis of the knocking sensor output signal basically needs to be performed in synchronization with the rotation of the engine, it is necessary to further shorten the calculation time of the frequency component analysis when the engine rotational speed is high. Since this point is not taken into consideration in the apparatus shown in Patent Document 1, the time that can be secured for the knocking determination process using the frequency component intensity obtained by frequency component analysis is shortened, and accurate determination becomes difficult. There is a fear.

  The technique disclosed in Patent Document 2 can increase the calculation speed when the engine speed is high by using only the DFT algorithm, but uses the FFT algorithm when the engine speed is low. The calculation speed is higher. On the other hand, the apparatus disclosed in Patent Document 1 does not select either the DFT algorithm or the FFT algorithm from the viewpoint of calculation speed, and there is room for improvement in any of the techniques disclosed in any document.

  The present invention has been made paying attention to this point, and when performing the frequency component analysis of the detection parameter in synchronization with the rotation of the internal combustion engine, the frequency component intensity calculation is more appropriately executed, and the engine rotational speed is wide. An object of the present invention is to provide a frequency component analyzer capable of increasing the calculation speed in a range.

  In order to achieve the above object, the invention according to claim 1 is the frequency component analyzer for performing the frequency component analysis on the detected value (VKNK) of the operation parameter of the internal combustion engine in synchronization with the rotation of the engine. Sampling means for sampling parameters at predetermined time (TSMP) intervals and converting the sample values into digital values, and the intensity (STFT1) of a plurality of frequency components included in the detected value (VKNK) using a fast Fourier transform algorithm FFT calculation means for calculating, DFT calculation means for calculating the intensity (STFT2) of a plurality of frequency components included in the detected value (VKNK) using a discrete Fourier transform algorithm, and a rotational speed (NE) of the engine are set. When it is lower than the threshold (NETH), the FFT operation means uses the intensity of the plurality of frequency components. When the engine speed (NE) is equal to or higher than the set threshold value (NETH), a process for calculating the intensity of the plurality of frequency components by the DFT calculation means is executed in synchronization with the rotation of the engine. And an arithmetic control means.

  According to a second aspect of the present invention, in the frequency component analyzing apparatus according to the first aspect, the calculation control means switches between the calculation by the FFT calculation means and the calculation by the DFT calculation means. It is performed in an angular range (RCASW).

  According to a third aspect of the present invention, in the frequency component analyzing apparatus according to the first or second aspect, the DFT calculation means includes the first element strength (DMFTS) and the first corresponding to the plurality of frequency components. Element strength calculating means for calculating the intensity (DMFTC) of the second element whose phase is shifted by 90 degrees with respect to the element for a predetermined number (ND) of sample values, the first element intensity (DMFTS), and the second element Frequency component intensity calculating means for calculating the intensity (STFT2) of the plurality of frequency components using intensity (DMFTC), wherein the frequency component intensity calculating means includes a part of an integrated value of the first element intensity and the frequency component intensity calculating means. The frequency component intensity (STFT2) is calculated by replacing a part of the integrated value of the second element intensity with the previously calculated values (TRMS1, TRMC1).

  According to a fourth aspect of the present invention, in the frequency component analyzer according to the third aspect, the set threshold (NETH) is the number of processing steps per unit time required for calculating the frequency component strength (STFT). It is set according to (NS).

  The invention according to claim 5 is the frequency component analyzer according to any one of claims 1 to 4, wherein the operating parameter is a detection signal of a knock sensor (11) attached to the engine, The apparatus further comprises knocking determination means for determining knocking in the engine based on the intensities (STFT1, STFT2) of a plurality of frequency components calculated by the FFT calculation means or the DFT calculation means, and the FFT calculation means includes a plurality of first calculations. Intensities (STFT1) of a plurality of first frequency components corresponding to the frequency group (FG1) are calculated, and the DFT calculation means corresponds to a second frequency group (FG2) different from the first frequency group (FG1). The intensities (STFT2) of a plurality of second frequency components are calculated, and the knocking determination means is configured to output the first or second frequency included in a predetermined frequency range. A process of selecting representative values (GSTFT1, GSTFT2) of component strengths (STFT1, STFT2) is executed for a plurality of predetermined frequency ranges, and the determination of knocking is performed using the selected representative values (GSTFT1, GSTFT2). Characteristic to do.

  According to the first aspect of the present invention, the operating parameters of the internal combustion engine are sampled at predetermined time intervals, the sample values are converted into digital values, and the intensities of a plurality of frequency components are calculated based on the obtained digital values. The When the engine rotational speed is lower than the set threshold, the intensities of a plurality of frequency components included in the detected value are calculated using a fast Fourier transform algorithm (FFT algorithm), while the engine rotational speed is equal to or higher than the set threshold. The intensities of a plurality of frequency components included in the detected value are calculated using a discrete Fourier transform algorithm (DFT algorithm), and the frequency component intensity calculating process is executed in synchronization with the engine rotation. When the engine speed is lower than the set threshold, the calculation speed can be increased by using the FFT algorithm as compared with the case of using the DFT algorithm. When the engine speed is equal to or higher than the set threshold, the calculation speed can be increased as compared with the FFT algorithm by improving the DFT algorithm. As a result, a high calculation speed can be realized in a wider range of the engine rotation speed.

  According to the second aspect of the present invention, switching between the calculation by the FFT algorithm and the calculation by the DFT algorithm is executed within a predetermined crank angle range of the engine. If switching is performed during the period when the frequency component intensity calculation is being performed, the calculation result before the switching is not completed, and thus the obtained calculation result may be incomplete. Therefore, by performing switching in the crank angle range where the calculation is not executed, it is possible to avoid interruption of calculation due to switching and perform smooth switching.

  According to the third aspect of the present invention, when performing the calculation by the DFT algorithm, the intensity of the first element and the intensity of the second element whose phase is shifted by 90 degrees with respect to the first element are the predetermined number of sample values The first element strength and the second element strength are used to calculate the intensity of a plurality of frequency components, and a part of the integrated value of the first element strength and a part of the integrated value of the second element strength are respectively calculated. The frequency component intensity is calculated in place of the previously calculated value. Therefore, when calculating the frequency component intensity using the DFT algorithm, it is not necessary to perform part of the integration calculation of the first and second element intensities again, and the calculation speed can be increased.

  According to the fourth aspect of the invention, the setting threshold value for the engine speed is set according to the number of processing steps per unit time required for calculating the frequency component intensity. Since the previously calculated values of the first and second element strengths applied to the calculation of the frequency component strength by the DFT algorithm partially overlap with the current calculated values, it is possible to use the previously calculated values for the overlapping strengths. When possible, the number of processing steps NSD per unit time of the DFT algorithm becomes constant even when the engine speed increases, while the number of processing steps NSF of the FFT algorithm per unit time is proportional to the increase of the engine speed. Then increase. Therefore, by setting the engine speed at which the processing step number NSF is equal to the processing step number NSD as the set threshold value, it is possible to switch the appropriate calculation algorithm.

  According to the fifth aspect of the present invention, the intensity of the first frequency component corresponding to the first frequency group is calculated using the FFT algorithm, and the second frequency group corresponding to the second frequency group different from the first frequency group is calculated. The intensity of the frequency component is calculated using the DFT algorithm. A process of selecting a representative value of the intensity of the first or second frequency component included in the predetermined frequency range is executed for a plurality of predetermined frequency ranges, and knocking is determined using the selected representative values. Since the corresponding frequency does not match between the first frequency component calculated by the FFT algorithm and the second frequency component calculated by the DFT algorithm, representative of the intensity of the first or second frequency component included in the predetermined frequency range By using the value, it is possible to appropriately perform the knocking determination process even when switching between the FFT algorithm and the DFT algorithm.

It is a figure which shows the structure of the internal combustion engine and its control apparatus concerning one Embodiment of this invention. It is a figure for demonstrating sampling of a knock sensor output, and frequency component analysis. It is a time chart for demonstrating the procedure and timing relationship which calculate frequency component intensity | strength (STFT1) by a FFT algorithm. It is a time chart for demonstrating the procedure and timing relationship which calculate frequency component intensity | strength (STFT2) by a DFT algorithm. It is a time chart for demonstrating the relationship between the calculation period (TS) of frequency component intensity | strength (STFT), and the generation period (CRME) of a CRK interruption. It is a figure which shows the relationship between an engine speed (NE) and the number of calculation steps per unit time. It is a flowchart for demonstrating the procedure which calculates frequency component intensity | strength (STFT). It is a figure which shows a numerical formula in order to demonstrate the method of calculating a frequency component intensity | strength using a FFT algorithm. It is a figure which shows the switching period (RCASSW) of the algorithm to be used. It is the figure which showed the intensity | strength data obtained by a frequency component analysis as a spectrum time series map. It is a figure which shows the representative value spectrum time series map produced based on the spectrum time series map shown in FIG. 10, and the reference | standard spectrum time series map for knock determination. It is a flowchart of a knocking determination process. It is a figure which shows the frequency component intensity | strength of the vibration detected by a knock sensor.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is an overall configuration diagram of an internal combustion engine (hereinafter referred to as “engine”) and a control device thereof according to an embodiment of the present invention. It is arranged. A throttle valve opening sensor 4 for detecting the throttle valve opening TH is connected to the throttle valve 3, and a detection signal of the sensor 4 is supplied to an electronic control unit (hereinafter referred to as “ECU”) 5.

  The fuel injection valve 6 is provided for each cylinder between the engine 1 and the throttle valve 3 and slightly upstream of the intake valve (not shown) of the intake pipe 2, and each injection valve is connected to a fuel pump (not shown). At the same time, it is electrically connected to the ECU 5 and the valve opening time of the fuel injection valve 6 is controlled by a signal from the ECU 5. Each cylinder of the engine 1 is provided with a spark plug 7, and the spark plug 7 is connected to the ECU 5. The ECU 5 supplies an ignition signal to the spark plug 7.

  An intake pressure sensor 8 for detecting the intake pressure PBA and an intake air temperature sensor 9 for detecting the intake air temperature TA are provided on the downstream side of the throttle valve 3. A cooling water temperature sensor 10 that detects the engine cooling water temperature TW and a non-resonant knock sensor 11 are mounted on the main body of the engine 1. Detection signals from the sensors 8 to 11 are supplied to the ECU 5. As the knock sensor 11, for example, a sensor capable of detecting vibration in a frequency band from 5 kHz to 25 kHz is used.

An intake air flow rate sensor 13 for detecting the intake air flow rate GA is provided on the upstream side of the throttle valve 3 in the intake pipe 2, and the detection signal is supplied to the ECU 5.
A crank angle position sensor 12 that detects a rotation angle of a crankshaft (not shown) of the engine 1 is connected to the ECU 5, and a signal corresponding to the rotation angle of the crankshaft is supplied to the ECU 5. The crank angle position sensor 12 is a cylinder discrimination sensor that outputs a pulse (hereinafter referred to as “CYL pulse”) at a predetermined crank angle position of a specific cylinder of the engine 1, and relates to a top dead center (TDC) at the start of the intake stroke of each cylinder. A TDC sensor that outputs a TDC pulse at a crank angle position before a predetermined crank angle (every 180 degrees of crank angle in a four-cylinder engine) and one pulse (hereinafter referred to as “CRK”) with a constant crank angle cycle shorter than the TDC pulse (for example, a cycle of 6 °). The CYL pulse, the TDC pulse, and the CRK pulse are supplied to the ECU 5. These pulses are used for various timing controls such as fuel injection timing and ignition timing, and detection of engine speed (engine speed) NE.

  The engine 1 includes a first valve operating characteristic variable mechanism that continuously changes a valve lift amount and an opening angle (valve opening period) of an intake valve (not shown), and a crankshaft rotation angle of a cam that drives the intake valve. Is provided with a variable valve operation characteristic device 20 having a second valve operation characteristic variable mechanism that continuously changes the operation phase with reference to. The ECU 5 supplies the lift amount control signal and the operation phase control signal to the valve operation characteristic variable device 20, and controls the operation of the intake valve. The configurations of the first and second valve operating characteristic variable mechanisms are shown in, for example, Japanese Patent Application Laid-Open No. 2008-25418 and Japanese Patent Application Laid-Open No. 2000-227013.

  The ECU 5 shapes input signal waveforms from various sensors, corrects the voltage level to a predetermined level, converts an analog signal value into a digital signal value, etc., and a central processing unit (hereinafter referred to as “CPU”). A storage circuit (memory) for storing various calculation programs executed by the CPU, calculation results, and the like, an output circuit for supplying a drive signal to the fuel injection valve 6 and the spark plug 7, and the like.

  In the present embodiment, frequency component analysis of the output signal of the knock sensor 11 is performed, and knocking determination is performed based on the frequency component intensity obtained as a result of the analysis. First, an outline of frequency component analysis will be described. FIG. 2A shows an output signal waveform of the knock sensor 11, and FIG. 2B is an enlarged view of the waveform of the period TS in FIG. In the present embodiment, the sampling period TSMP is set to 20 microseconds, and frequency component analysis by fast Fourier transform (hereinafter referred to as “FFT”) for 64 pieces of data detected continuously, or 50 pieces of data are targeted. The frequency component analysis is performed by the discrete Fourier transform (hereinafter referred to as “DFT”). The result of the frequency component analysis is shown in FIG. The vertical axis in FIG. 2C is the frequency component intensity STFT calculated by the DFT algorithm. In this embodiment, 21 predetermined frequencies (5, 6, 7,..., 24 in the frequency band from 5 kHz to 25 kHz are used. , 25 kHz) is calculated every 6 degrees of crank angle (every time the crankshaft of engine 1 rotates 6 degrees). In the following description, the frequency component intensity calculated by the FFT algorithm is referred to as “first frequency component intensity STFT1”, and the frequency component intensity calculated by the DFT algorithm is referred to as “second frequency component intensity STFT2”. Is simply referred to as “frequency component intensity STFT”. Unlike the case of the DFT algorithm, the plurality of predetermined frequencies corresponding to the first frequency component intensity STFT1 are frequencies of “0.781 kHz” intervals from 5.467 kHz to 24.211 kHz.

  FIG. 3 is a time chart for explaining the frequency component analysis by the FFT algorithm. FIG. 3A shows detection data VKNK (digital) sequentially obtained by sampling the output signal of the knock sensor 11 every 20 microseconds. Value) indicates an address number (repeated in the range of 1 to 50) corresponding to the memory address in which the value is sequentially stored. FIG. 3B shows a CRK pulse output from the crank angle position sensor 12. In this embodiment, the falling time of the CRK pulse is used as a reference for the calculation execution time (hereinafter referred to as “CRK interrupt”). In the present embodiment, the timing at which the CRK interrupt occurs is stored, and the FFT calculation is performed using 64 detection data obtained during the sampling period TS centered on the timing. Here, “j1” is an index parameter indicating a frequency (hereinafter referred to as “first frequency index”), and j1 = 0, 1, 2,..., 24 are frequencies 5.467, 6.248, 7.029,. , 24.211 kHz (first frequency group FG1). “I” is an index parameter (hereinafter referred to as “crank angle index”) indicating a crank angle CA at which the CRK interruption occurs (an angle position when the piston is located at the top dead center is 0 degree), and i = 0, 1, 3, ..., 11 correspond to crank angles 12, 18, ..., 72 degrees.

  In the example illustrated in FIG. 3, the index parameter m (= 1 to 64) illustrated in FIG. 3C indicates data applied to the calculation of the first frequency component strength STFT1 (j1, i). At the first CRK interrupt timing, the first frequency component strength STFT1 (j1, i) is calculated using 64 pieces of data centering on the detection data of the address number “5” (detection data corresponding to m = 32). At the next CRK interrupt timing, the first frequency component intensity STFT1 (j1, i + 1) is obtained using 64 pieces of data centering on the detection data of the address number “45” (detection data corresponding to m = 32). Is calculated. The frequency component intensity STFT is a dimensionless quantity indicating relative intensity.

  The data applied to the calculation of the first frequency component intensity STFT1 is 64 data centered on detection data corresponding to m = 32, in other words, during the sampling period TS centering on the generation timing of the CRK interrupt. By using the 64 detection data obtained, frequency component analysis centered on the target crank angle can be performed. As a result, it is less susceptible to engine rotation fluctuations than the method using detection data sampled within the sampling period starting from the time when the CRK interrupt occurs or within the sampling period starting from the time when the CRK interrupt occurs. The effect is obtained.

  FIG. 4 is a time chart for explaining frequency component analysis by the DFT algorithm, and FIG. 4A is the same as FIG. 3A. FIG. 4B shows a state where the sine wave component intensity DMFTS and the cosine wave component intensity DMFTC calculated using the detection data VKNK are stored in the memory corresponding to the detection data VKNK. The address number is the same as that in FIG. 4A, but the sine wave component intensity DMFTS and cosine wave component corresponding to 21 frequencies from 5 kHz to 25 kHz corresponding to the sample value of one detection data VKNK. The intensity DMFTC is calculated every 1 millisecond and stored in the memory.

  FIG. 4C is the same as FIG. In the calculation by the DFT algorithm, each frequency (5 to 5) is used by using the sine wave component strength DMFTS and the cosine wave component strength DMFTC corresponding to 50 detection data obtained during the sampling period TS centering on the timing at which the CRK interrupt occurs. Component intensity corresponding to 25 kHz), that is, the second frequency component intensity STFT2 (j2, i) is calculated (FIG. 4D). Here, “j2” is a frequency index used in the DFT calculation, and j2 = 0, 1, 2,..., 20 corresponds to frequencies 5, 6, 7,..., 25 kHz (second frequency group FG2). The crank angle index i is the same as the FFT calculation.

  In the example illustrated in FIG. 4, the index parameter m (= 1 to 50) illustrated in FIG. 4C indicates data applied to the calculation of the second frequency component strength STFT2 (j2, i). At the first CRK interrupt timing, the second frequency component strength STFT2 (j2, i) is calculated using 50 pieces of data centering on the detection data of the address number “5” (detection data corresponding to m = 25). At the next CRK interrupt timing, the second frequency component intensity STFT2 (j2, i + 1) is obtained by using 50 data centered on the detection data of the address number “45” (detection data corresponding to m = 25). Is calculated.

  FIG. 5 is a diagram showing the relationship between the CRK interrupt generation period CRME and the detection data sampling period (displayed as “STFT calculation”) TS applied to the calculation of the frequency component intensity STFT. FIG. 5A shows an example in which the engine speed NE is 1000 rpm, and the sampling period TS and the interrupt generation period CRME match. Therefore, when the engine speed NE exceeds 1000 rpm, the sampling periods TS partially overlap. FIG. 5B shows an example in which the engine speed NE is slightly higher than 2000 rpm, and the overlapping portion of the sampling period TS is long.

  In the present embodiment, as shown in FIG. 5A, the first frequency component strength STFT1 is calculated using the FFT algorithm in a state where the sampling periods TS do not overlap or the overlapping periods are relatively short. As shown in FIG. 5B, when the overlapping period of the sampling period TS is relatively long, the second frequency component intensity STFT2 is calculated using the DFT algorithm, and the sine wave component intensity DMFTS related to the overlapping part is integrated. By using the previously calculated value as the integrated value of the value and the cosine wave component intensity DMFTC, the calculation time required for calculating the second frequency component intensity STFT2 is shortened.

  A solid line L11 shown in FIG. 6 is a diagram showing the relationship between the number of processing steps NSD per unit time and the engine speed NE when the previously calculated value is used in the CPU used in the present embodiment. Until reaching NESAT, the number of processing steps NSD increases, and becomes a constant value NSSAT in the range of NE ≧ NESAT. A broken line L12 shown in FIG. 6 indicates the relationship between the engine speed NE and the number of processing steps NSF per unit time when the FFT algorithm is used. As is apparent from FIG. 6, when the FFT algorithm is used, the processing step number NSF increases linearly with respect to the increase in the engine speed NE, but in the range where the engine speed NE is equal to or less than the crossing speed NECRS, The processing step number NSF is smaller than the processing step number NSD of the DFT algorithm. Therefore, in the present embodiment, the rotation speed threshold value NETH is set to the cross rotation speed NECRS, and when the engine rotation speed NE is lower than the rotation speed threshold value NETH, the first frequency component intensity STFT1 is calculated using the FFT algorithm, When the engine speed NE is equal to or higher than the speed threshold value NETH, the second frequency component strength STFT2 is calculated using the DFT algorithm. As a result, an increase in the number of processing steps per unit time can be suppressed in a wide range of the engine speed NE, and the calculation speed can be increased.

FIG. 7 is a flowchart showing a procedure for calculating the frequency component intensity. In step S1, detection data VKNK is acquired, and in step S2, sine wave component strength DMFTS (j2, k) and cosine wave component strength DMFTC (j2, k) are calculated by the following equations (1) and (2). Here, the index parameter k is an address number shown in FIG. 4B, and takes a value from “1” to ND (“50” in the present embodiment). Δt corresponds to (sampling period × ND), and is 1 millisecond in this embodiment.
DMFTS (j2, k) = VKNK (k) ×
sin {2π × (j2 + 5) × 1000 × Δt × k / ND} (1)
DMFTC (j2, k) = VKNK (k) ×
cos {2π × (j2 + 5) × 1000 × Δt × k / ND} (2)

  Steps S1 and S2 are executed every sampling period (20 microseconds).

  In step S3, it is determined whether or not a CRK interrupt has occurred. If not, the process returns to step S1 and whether or not the engine speed NE is equal to or greater than the speed threshold value NETH at the timing when the CRK interrupt occurs. Is determined. When NE <NETH, the first frequency component strength STFT1 is calculated by the FFT algorithm (step S6), and when NE ≧ NETH, the second frequency component strength STFT2 is calculated by the DFT algorithm (step S6). Step S5).

  In step S5, specifically, the second frequency component strength STFT (j2, i) is calculated by the following equation (3). The index parameter m is a modified address number that is set so that the address number k when the CRK interrupt occurs is “25” (see FIG. 4D). The index parameter mp is a modified address number applied at the time of previous calculation, mx is the previous boundary index, and my is the current boundary index, which are given by the following equations (4) and (5), respectively.

ｍｘ＝ＮＤ−ＮＶ＋１ （４）
ｍｙ＝ＮＶ＋１ （５）
ここでＮＶは前回のＳＴＦＴ２算出に使用したデータのうち、今回のＳＴＦＴ２算出に使用できるデータの数であり、データ数ＮＶはエンジン回転数ＮＥが高くなるほど増加する。

mx = ND-NV + 1 (4)
my = NV + 1 (5)
Here, NV is the number of data that can be used for the current STFT2 calculation among the data used for the previous STFT2 calculation, and the data number NV increases as the engine speed NE increases.

  In the equation (3), the first term TRMS1 and the third term TRMC1 correspond to the previous calculated value of the integrated value, and the second term TRMS2 and the third term TRMC2 correspond to the newly calculated integrated value.

  In this way, a part of the integrated value of the sine wave component intensity DMFTS and a part of the integrated value of the cosine wave component intensity DMFTC are respectively replaced with the previously calculated values (TRMS1, TRMC1), and the frequency component intensity STFT (j, i) By calculating, when calculating the second frequency component intensity STFT2, it is not necessary to perform part of the integration calculation of the sine wave component intensity DMFTS and the cosine wave component intensity DMFTC again, and the calculation speed can be increased.

Next, a method for calculating the first frequency component intensity STFT1 by the FFT algorithm will be described. Since the FFT algorithm is a well-known calculation method and it is difficult to perform a simple description using mathematical formulas, an outline of the calculation executed in the present embodiment will be described below.
FIG. 8 is a diagram showing a group of mathematical expressions corresponding to the first three stages of operations executed by the computer using 64 pieces of detection data VKNK. “W” in these equations is a complex trigonometric function called a rotator defined by the following equation (11). In equation (11), “i” is used as a symbol indicating an imaginary number.

X (n) (n = 0 to 63) in the mathematical expression group EG1 is a detection parameter corresponding to the detection data VKNK (n) (n = 0 to 63) as shown in the following expression (12). For n, a bit inversion operation (indicated by *) is performed.
x (n) = VKNK (n *) (12)

  Since the number of data is 64 (= 2 to the sixth power), for example, when n = 1 is represented by a 6-bit binary number, it becomes “000001”, when the bit is inverted, it becomes “100,000”, and this is represented by a decimal number “32”. " That is, x (1) = VKNK (32).

By applying the converted detected parameters x (n) to the equation group EG1 first by Equation (12), the calculated frequency component intensity parameters X _0, by applying the frequency component intensity parameter X ₀ in formula group EG2 The frequency component strength parameter X ₁ is calculated, and by applying the frequency component strength parameter X ₁ to the mathematical formula group EG3, the frequency component strength parameter X ₂ is calculated. By repeating the same operation, the frequency component strength parameter X ₁ is calculated. ₅ (0, jx) (jx = 1 to 31) is obtained, which corresponds to the frequency component FFT (jx) to be obtained. “Jx” is a discrete frequency, and corresponds to (50 × jx / 64) [kHz] when the sampling period is 20 microseconds (sampling frequency 50 kHz) and the number of data is 64.

Accordingly, the first frequency component intensity STFT1 is given by the following equation (13). “Re” and “Im” in Expression (13) indicate a real part and an imaginary part of a complex number, respectively. The first frequency index j1 corresponds to a value obtained by subtracting “7” from the index parameter jx. Therefore, when Expression (13) is expressed using the first frequency index j1, Expression (13a) is obtained, and thereby the first frequency component intensity STFT1 (j1) is calculated by calculation using the FFT algorithm. Similarly to the calculation by the DFT algorithm, when the frequency component intensity in the frequency range of 5 kHz to 25 kHz is calculated, the index parameter jx takes 25 values from “7” to “31”.

  FIG. 9 is a time chart for explaining the switching timing when switching the calculation algorithm when the determination result in step S4 of FIG. 7 changes from “YES” to “NO” or vice versa. The generation timing (crank angle 6 degrees) is indicated by a vertical line. In this embodiment, knocking is performed using detection data acquired in a period of 12 degrees to 72 degrees after compression top dead center + α (a period (32 × 20 microseconds before or after that) or (25 × 20 microseconds)). Since the determination is performed, the period from 96 degrees before the compression top dead center to the compression top dead center is set as the switching period RCASW while avoiding the data acquisition period TDR used for knocking determination. Thereby, interruption of the frequency component intensity calculation before and after switching can be avoided and smooth switching can be performed.

Next, an outline of the knocking determination method using the frequency component intensity STFT (STFT1 or STFT2) will be described with reference to FIGS.
As described above, the frequency component intensity STFT is calculated as time series data as a two-dimensional matrix (hereinafter referred to as “spectrum time series map”) as shown in FIG. 10 (a) or FIG. 10 (b). In the spectrum time series map, the vertical direction is the frequency f, and the horizontal direction is the crank angle (crank angle after compression top dead center at which the combustion stroke starts) CA. FIG. 10A shows a first spectral time-series map calculated using the FFT algorithm, and the frequency index j1 (j1 = 0 to 24) and the crank angle index i ( i = 0 to 10), and the map value of each lattice point is displayed as the first frequency component intensity STFT1 (j1, i).

  FIG. 10B shows a second spectral time-series map calculated using the DFT algorithm, and the frequency index j2 (j2 = 0 to 20) and the crank angle index i ( i = 0 to 10), and the map value of each lattice point is displayed as the second frequency component intensity STFT2 (j2, i).

  In the present embodiment, the spectral time-series maps shown in FIGS. 10A and 10B are divided into seven frequency ranges from a plurality of predetermined frequency ranges Rf0 to Rf6, and the frequency component intensity STFT in each frequency range is divided. By selecting the maximum value as the representative value, the representative value spectrum time-series map shown in FIGS. 11A and 11B is generated. Then, knocking determination is performed by comparing the representative value spectrum time series map with a reference spectrum time series map (FIG. 11C) corresponding to the representative value spectrum time series map at the time of occurrence of knocking. That is, the matching rate between the map value of the representative value spectrum time-series map and the map value of the reference spectrum time-series map is calculated, and it is determined that knocking has occurred when the matching rate exceeds the determination threshold.

In the present embodiment, the first frequency range Rf0 to the seventh frequency range Rf6 are defined as follows.
Rf0 5kHz ≦ f <8kHz
Rf1 8kHz ≦ f <11kHz
Rf2 11 kHz ≦ f <14 kHz
Rf3 14 kHz ≦ f <17 kHz
Rf4 17 kHz ≦ f <20 kHz
Rf5 20 kHz ≦ f <23 kHz
Rf6 23 kHz ≦ f <26 kHz

  FIG. 11A shows a first representative value spectrum time-series map generated from the first spectrum time-series map of FIG. 10A, and frequency indexes jg (jg = 0) as vertical and horizontal indexes, respectively. -7) and the crank angle index i (i = 0 to 10), the map value of each lattice point is displayed as the first representative frequency component intensity GSTFT1 (jg, i).

  FIG. 11B shows a second representative value spectrum time-series map generated from the second spectrum time-series map of FIG. 10B, and the vertical and horizontal indexes are the same as those in FIG. 11A. It is. The map value of each lattice point is displayed as the second representative frequency component intensity GSTFT2 (jg, i).

  FIG. 11C shows a reference spectrum time-series map corresponding to the representative value spectrum map when knocking occurs, and the vertical and horizontal indexes are the same as those in FIG. The map value of each lattice point is displayed as the reference frequency component intensity RSTFT (jg, i).

  When the representative value spectrum time series map is generated, the forms (number of rows) of the first and second representative value spectrum time series maps are unified, and when two algorithms are switched and used, FIG. It is possible to perform knocking determination by comparing with one reference spectrum time series map shown in the figure, and it is possible to reduce the amount of calculation required for the comparison.

  Actually, as shown in FIG. 12, noise removal processing and map value binarization processing are performed, and the binarized representative value spectrum time-series map and binarization reference spectrum time-series map after the processing are performed. Knocking determination is performed by comparing. Details of the noise removal process, the binarization process, and the comparison process with the reference map are disclosed in Patent Document 2, for example.

  FIG. 12 is a flowchart of a process for performing knocking determination by the above-described method, and this process is executed by the CPU of the ECU 5 in synchronization with the generation of the TDC pulse.

  In step S11, a spectrum time series map shown in FIG. 10A or 10B is calculated. In step S12, a representative value time series map is generated from the spectrum time series map shown in FIG. 10A or 10B. To do. In step S13, noise removal processing is performed using a noise map generated by learning processing (step S20) from the representative value time-series map when knocking has not occurred. In step S14, the binarized representative value time series map is generated by binarizing the map value of the representative value time series map from which noise has been removed to “0” or “1”.

  In step S15, the matching rate calculation process of the map value of the binarized representative value time-series map and the map value of the corresponding reference time-series map is executed, and the matching rate PFIT of both is calculated. In step S16, a determination threshold SLVL is calculated according to the engine speed NE and the intake pressure PBA. Determination threshold SLVL is set so as to increase as engine speed NE increases and to increase as intake pressure PBA increases.

  In step S17, it is determined whether or not the relevance ratio PFIT calculated in step S15 is larger than the determination threshold SLVL. If the answer is affirmative (YES), it is determined that knocking has occurred, and the knocking flag FKNOCK is set. “1” is set (step S18).

  If PFIT ≦ SLVL in step S17, it is determined that knocking has not occurred, and the knocking flag FKNOCK is set to “0” (step S19). Next, noise learning processing is executed based on the representative value time series map generated in step S12, and the noise map is updated (step S20).

  The noise map reflects noise that occurs regularly, for example, the intake noise of the intake valve. For this reason, noise removal using the noise map eliminates the effects of stationary noise and provides high accuracy. Can be determined. FIG. 13 shows a frequency spectrum (solid line) of seating noise and a frequency spectrum (dashed line) at the time of occurrence of knocking, and the influence of such noise can be reliably removed by the processing of FIG. .

  As described above in detail, in this embodiment, the output signal of the knock sensor 11 is sampled at the sampling period TSMP, the sample value is converted to a digital value, and the frequency range from 5 kHz to 25 kHz based on the obtained digital value VKNK. Intensities STFT of a plurality of frequency components are calculated. When the engine rotational speed NE is lower than the rotational speed threshold value NETH, the frequency component strength STFT is calculated using the FFT algorithm, while when the engine rotational speed NE is equal to or higher than the rotational speed threshold value NETH, the frequency component strength STFT is calculated as DFT. The frequency component intensity STFT is calculated using an algorithm and is executed in synchronization with the engine rotation. When the engine speed NE is lower than the speed threshold value NETH, the calculation speed can be increased by using the FFT algorithm as compared with the case of using the DFT algorithm. When the engine speed NE is equal to or higher than the speed threshold value NETH, the calculation speed can be increased as compared with the FFT algorithm by improving the DFT algorithm. As a result, a high calculation speed can be realized in a wider range than the engine speed NE.

  Further, switching between the calculation by the FFT algorithm and the calculation by the DFT algorithm is executed in the predetermined crank angle range RCASW. If switching during the period during which frequency component intensity calculation is being performed, the calculation before switching is not complete, so the calculation result obtained may be incomplete, but the predetermined crank angle at which calculation is not performed By performing switching in the range RCASW, it is possible to avoid computation interruption due to switching and perform smooth switching.

  In addition, when using the DFT algorithm, a part of the integrated value of the sine wave component intensity DMFTS and a part of the integrated value of the cosine wave component intensity DMFTC are respectively replaced with the previous calculated values (TRMS1, TRMC1), respectively. Since STFT (j, i) is calculated, when calculating the second frequency component strength STFT2, it is not necessary to perform part of the integration calculation of the sine wave component strength DMFTS and the cosine wave component strength DMFTC again. The calculation speed when using the algorithm can be increased.

  The rotation speed threshold value NETH is set according to the number of processing steps per unit time required for calculating the frequency component intensity. When using the DFT algorithm, by replacing the part of the integrated value of the sine wave component intensity DMFTS and the part of the integrated value of the cosine wave component intensity DMFTC with the previous calculated values (TRMS1, TRMC1), respectively, The number of processing steps NSD per unit time of the DFT algorithm is constant as the engine speed NE increases, whereas the number of processing steps NSF of the FFT algorithm per unit time is proportional to the increase of the engine speed NE. Increase (Figure 6). Therefore, by setting the engine speed NE at which the process step number NSF is equal to the process step number NSD as the engine speed threshold value NETH, it is possible to switch the calculation algorithm appropriately.

  Further, a process of selecting the representative value GSTFT1 or GSTFT2 of the first frequency component intensity STFT1 or the second frequency component intensity STFT2 included in the first to seventh frequency ranges Rf0 to Rf6 is performed, and the representative value spectrum time series map (FIG. 11) is selected. (A) and (b)) are generated, and knocking is determined using the representative value spectrum time-series map. The first frequency component intensity STFT1 calculated by the FFT algorithm and the second frequency component intensity STFT2 calculated by the DFT algorithm do not coincide with each other, so the first frequency component intensity included in the frequency range Rf0 to Rf6. By using the representative value of STFT1 or second frequency component intensity STFT2, knocking determination processing can be performed appropriately even when switching between the FFT algorithm and the DFT algorithm.

  In the present embodiment, the ECU 5 constitutes sampling means, FFT computing means, DFT computing means, computation control means, element strength calculating means, frequency component strength calculating means, and knocking determining means.

  The present invention is not limited to the embodiment described above, and various modifications can be made. For example, in the above-described embodiment, the representative value spectrum time series map is generated from the spectrum time series map, and the knocking determination is performed using the representative value spectrum time series map. You may make it perform knocking determination by comparing with a reference | standard spectrum time series map. In that case, the comparison process is performed using the reference spectrum time-series map corresponding to the first spectrum time-series map and the one corresponding to the second spectrum time-series map. Or you may make it make the 2nd frequency group FG2 which calculates a frequency component intensity | strength using a DFT algorithm be the same as the 1st frequency group FG1. In that case, only one reference spectrum time-series map is sufficient.

  Further, the sampling period of the knock sensor output and the crank angle interval for performing frequency component analysis are not limited to those described above (20 microseconds, 6 degrees), and can be changed within a range in which the object of the present invention is achieved. . Further, the spectrum time series map (configured by a matrix of 31 rows × 11 columns or 21 rows × 11 columns in the above-described embodiment) can be similarly changed.

  Further, in the above-described embodiment, the device for performing the frequency analysis of the detection value by the knock sensor is shown. However, the present invention is not limited to this, and the frequency component analysis of the engine operation parameter detection value is synchronized with the engine rotation (CRK interruption). It can be applied to For example, the present invention can also be applied to frequency component analysis of the instantaneous rotational speed of the engine 1 obtained as the CRM pulse (interrupt) generation period CRME output from the CRK sensor or the inverse thereof.

DESCRIPTION OF SYMBOLS 1 Internal combustion engine 5 Electronic control unit (Sampling means, FFT calculation means, DFT calculation means, calculation control means, element intensity calculation means, frequency component intensity calculation means, knocking determination means)
11 Knock sensor

Claims (5)

In the frequency component analysis device that performs frequency component analysis on the detected value of the operating parameter of the internal combustion engine in synchronization with the rotation of the engine,
Sampling means for sampling the operating parameters at predetermined time intervals and converting the sample values into digital values;
FFT calculation means for calculating the intensity of a plurality of frequency components included in the detected value using a fast Fourier transform algorithm;
DFT calculation means for calculating the intensity of a plurality of frequency components included in the detected value using a discrete Fourier transform algorithm;
When the rotational speed of the engine is lower than a set threshold value, the FFT calculation means calculates the intensity of the plurality of frequency components, and when the engine rotation speed is equal to or higher than the set threshold value, the DFT calculation means calculates the plurality of frequency components. And a calculation control means for executing a process of calculating the intensity of the frequency component in synchronization with the rotation of the engine.   2. The frequency component analyzing apparatus according to claim 1, wherein the calculation control unit performs switching between calculation by the FFT calculation unit and calculation by the DFT calculation unit in a predetermined crank angle range of the engine. The DFT calculation means calculates the intensity of the first element and the intensity of the second element that are 90 degrees out of phase with respect to the first element, corresponding to the plurality of frequency components, for a predetermined number of sample values. Intensity calculation means;
Frequency component intensity calculating means for calculating the intensity of the plurality of frequency components using the first element intensity and the second element intensity;
The frequency component intensity calculating means calculates the intensity of the frequency component by replacing a part of the integrated value of the first element intensity and a part of the integrated value of the second element intensity with the previous calculated value, respectively. The frequency component analyzer according to claim 1 or 2.   The frequency component analyzer according to claim 3, wherein the set threshold is set according to the number of processing steps per unit time required for calculating the frequency component intensity. The operating parameter is a detection signal of a knock sensor attached to the engine,
A knock determination unit for determining knocking in the engine based on the intensities of a plurality of frequency components calculated by the FFT calculation unit or the DFT calculation unit;
The FFT operation means calculates intensities of a plurality of first frequency components corresponding to the plurality of first frequency groups, and the DFT operation means includes a plurality of second frequency groups different from the first frequency group. Calculating the intensity of the second frequency component;
The knocking determination unit executes a process of selecting a representative value of the intensity of the first or second frequency component included in a predetermined frequency range for a plurality of predetermined frequency ranges, and uses the selected plurality of representative values. The frequency component analyzing apparatus according to any one of claims 1 to 4, wherein knocking is determined.

Images