JPH0643060A - Digital filter - Google Patents

Abstract

PURPOSE:To shorten the time of operation. CONSTITUTION:A digital filer 10 is constituted of two stages of filters 1 and 2. The degrees of the filters 1 and 2 are set so that the operation is completed in less than the sampling period of 2ms are set to that the 1 and the operation is completed in less than the sampling period of 10ms of the filter 2. The signal in the frequency band of 0-400Hz is inputted into the filter 1. The component in the narrow frequency band of 1-7Hz is extracted with the filter 2.

Description

Detailed Description of the Invention

[0001]

FIELD OF THE INVENTION This invention relates to digital filters.

[0002]

2. Description of the Related Art In an automobile, the rotational fluctuation of the engine may cause torsional vibration of the drive system, which may result in a surge phenomenon in which the vehicle vibrates in the front-rear direction. In this surge phenomenon, the frequency of the surge that gives an occupant an uncomfortable feeling is, for example, about 1 to 7 hertz. As one of the methods for measuring the degree of this surge, the fluctuation of the combustion pressure is detected and the fluctuation of the combustion pressure is detected. Therefore, there is a method of detecting a specific frequency band of about 1 to 7 hertz as surge torque.

To detect such surge torque, first, the combustion pressure of the engine is detected by a cylinder pressure sensor.
Based on this detection signal, the Pi meter measures the average effective pressure Pi.
Is calculated and the calculation result is input to the surge meter. The surge meter amplifies the operation result to the optimum amplitude, converts it into frequency components with a filter, and
The strength of the frequency band of 7 hertz is square-mean-squared, and the presence or absence of surge is determined by comparison with the reference level.

Since the surge depends on the ignition cycle,
When filtering, it is necessary to sample in a wide frequency band corresponding to the ignition cycle of the engine. For example, in a 4-cycle 8-cylinder engine, the combustion pressure should be taken into consideration in consideration of the operating speed (generally 0 to 6000 rpm). The detection frequency is equal to 0 to 400 hertz from the unit frequency = number of cylinders × rotation speed / 2, and it is necessary to detect the signal strength in a narrow frequency band of 1 to 7 hertz.

FIG. 4 is a diagram showing a conventional digital filter for digitally filtering the Pi operation result, which is a digital filter having a cascade structure of 32nd order. Further, in order to perform filtering with this digital filter, the sampling period is set to be shorter than the minimum value of the period of the signal including the fluctuation period component. That is, the maximum detection frequency of the combustion pressure Pi is 4
Since it is 00 Hertz and the minimum cycle value is 2.5 ms (= 1/400), set the sampling cycle to about 2 ms.

[0006]

By the way, in the conventional digital filter, the sampling frequency is 0 to 400.
Since a wide band of about Hertz must be used, the filter order must also be increased. However, if the filter order is increased, the number of repetitive calculations increases and the calculation time is required, and the calculation process may not be completed within the sampling period of 2 ms.

The present invention has been made in view of such conventional problems, and an object of the present invention is to provide a digital filter capable of shortening the calculation time.

[0008]

Therefore, according to the present invention, while inputting the latest data detected in each cycle of a signal including a fluctuation cycle component at a constant sampling cycle, a predetermined frequency component is extracted from the data. In the digital filter to be extracted, a plurality of stages of filters are connected in series, the first stage of the filter inputs data at a sampling period shorter than the minimum value of the period of the signal including the fluctuation period component, and the latter stage of the former stage receives the data. From the side, the sampling cycle is set larger than that of each filter, and each filter completes the arithmetic processing in less than the respective sampling cycle, and the order of each filter and the number of filter steps are set so that a predetermined frequency component is extracted from the filter at the final stage. I was supposed to.

[0009]

According to the above construction, the arithmetic processing is completed in less than the sampling period for each filter, and moreover, a predetermined frequency component is extracted from the final stage filter.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS. The digital filter of this embodiment is used, for example, in an engine surge detection device. That is, an in-cylinder pressure sensor (not shown) detects the combustion pressure of, for example, an engine with four cycles and eight cylinders. The detected combustion pressure of the engine is A / D converted, the average effective pressure Pi is calculated by a CPU (not shown), and the calculation result is temporarily stored in the RAM. The calculation result stored in this RAM is amplified to the optimum amplitude and then input to the digital filter.
The frequency component of the calculation result is extracted by the digital filter.

In FIG. 1 showing the present embodiment, a digital filter 10 has a two-stage structure including a filter 1 which is a first stage filter and a filter 2 which is a final stage filter, and is constructed in a RAM. . R for filter 1
The calculation result stored in the AM is input. As described above, in order to detect a surge of an engine with four cycles and eight cylinders, the surge depends on the ignition cycle. Therefore, the detection frequency band is set to a rotation frequency of 0 to 6000 rpm. From rotation speed / 2, it will be about 0 to 400 hertz. Therefore, the sampling period is set to the minimum value of the period (2.5m
Set to 2 ms, which is shorter than s). This sampling cycle 2
The configuration of the filter 1 is a 16th-order cascade type configuration of the delay units x _{1 to} x ₁₅ , the multipliers f _{1 to} f ₁₆ , and the adders a _{1 to} a ₁₆ so that the arithmetic processing is completed in less than ms. .

The sampling period of the filter 2 is 10 ms, which is larger than that of the filter 1. This is because the frequency band to be extracted is 1 to 7 hertz, and what is added by the filter 1 may be added. This sampling period
The configuration of the filter 2 includes delay devices y _{1 to} y ₆₃ and multipliers g _{1 to} g ₆₄ so that the arithmetic processing is completed in less than 10 ms and the desired frequency band of 1 to 7 hertz can be extracted. ,
And an adder b _{1 to} b _{64 in} a 64th-order cascade configuration.

The intensity of the frequency band component output from the filter 2 is square-mean-squared, and the presence or absence of surge is determined based on this value. Next, the operation of detecting a surge will be described. The combustion pressure of the engine detected by the in-cylinder pressure sensor is processed based on the flowchart of FIG. That is, in step (denoted as "S" in the drawing, the same applies hereinafter) 1, the arithmetic processing of the average effective pressure Pi is performed.

In step 2, the Pi calculation result is Pine.
It is stored in RAM as w. The Pinew stored in this RAM is processed based on the flowchart of FIG. Note that this processing is performed every 2 ms. In step 11, the Pinew data is input to the multiplier f ₁ and the delay unit x
Stored at ₁ .

In step 12, the Pinew data is processed by the digital filter 10. That is,
The data stored in the delay unit x ₁ is multiplied by the multiplier f ₂ and added by the adder a ₂ . Every time the Pinew data is input, the Pinew data input from the RAM is n →
(N-1)-> (n-2) ... are sequentially updated, and the Pinew data is processed.

P calculated by the filter 1 in this way
The inew data is input to the delay device y ₁ of the filter 2 every 10 ms, is processed in the same manner by the filter 2, and becomes 1 as a surge.
The intensities of the components in the ~ 7 hertz frequency band are calculated. According to such a configuration, the digital filter 10 has a two-stage configuration of filters 1 and 2, and the arithmetic processing of the filter 1 is completed within a sampling period of 2 ms and the arithmetic processing of the filter 2 is completed within a sampling period of less than 10 ms. Since the orders of filters 1 and 2 are set in, the calculation speed of the digital filter 10 is improved, and the maximum detection frequency of combustion pressure including the fluctuation period component is 400 hertz (minimum period value 2.5 ms).
It is possible to filter up to and to extract components in the frequency band of 1 to 7 hertz which is the surge frequency. Further, since the order of the filter 1 is reduced and the sampling period of the filter 2 is also increased, the number of delay units for storing data is reduced, and thus the memory capacity of the filter configuration RAM can be reduced.

[0017]

As described above, according to the present invention, the digital filter is composed of a plurality of stages and the arithmetic processing is completed within the predetermined sampling period of each filter from the first stage of inputting the signal including the fluctuation period component. By configuring the order and the number of stages in this way, the operation speed of the digital filter can be improved and a predetermined frequency component can be extracted.

[Brief description of drawings]

FIG. 1 is a diagram showing an embodiment of a digital filter of the present invention.

FIG. 2 is a flowchart of Pi calculation processing.

FIG. 3 is a flowchart of a processing operation of a digital filter.

FIG. 4 is a diagram showing a conventional digital filter.

[Explanation of symbols]

 1, 2 filter 10 digital filter

Claims (1)

[Claims] 1. A digital filter for extracting a predetermined frequency component from the data while inputting the latest data detected in each period of a signal including a fluctuation period component at a constant sampling period, The first-stage filter is configured by connecting filters in series, and the first-stage filter inputs data at a sampling period shorter than the minimum value of the period of the signal containing the fluctuation period component, and the latter-stage side has a larger sampling period than the front-stage side and each filter The digital filter is characterized in that the order of each filter and the number of stages of the filters are set so that the arithmetic processing is completed in less than each sampling period and a predetermined frequency component is extracted from the filter of the final stage.