JP2572644Y2 - Vehicle interior noise reduction device - Google Patents

Description

[Detailed description of the invention]

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for reducing vehicle interior noise, and more particularly to a device for efficiently reducing vehicle interior noise without using a microphone for detecting noise at a control point.

[0002]

2. Description of the Related Art As a conventional noise reduction device of this kind, there is a technology disclosed in Japanese Patent Application Laid-Open No. Hei 4-11291. The publication discloses an engine vibration detecting means, an adaptive controller for identifying an inverse transfer function of a transfer function from a detection signal of the detecting means to an ear of a vibration system of a vehicle body, and outputting an output signal of the controller to a vehicle interior. Speaker for detecting noise in the entire vehicle, and means for detecting vehicle body vibration from the road surface. The controller controls the speaker so that the input of the microphone is minimized, and the vehicle body vibration is set to a threshold value. A technique is disclosed in which the output to the speaker is stopped for a certain period of time when the distance exceeds the limit.

[0003] In this technique, LMS (Least Mean Square Method) is used as a means for obtaining an input to a microphone from engine vibration. And such LMS
When using the method, a microphone for detecting noise was required at the control point.

[0004]

As described above, the conventional method requires a control microphone, and it is difficult to install the microphone at a desired control point. However, there is a problem that the effect cannot be expected at the control point originally desired.

[0005] In addition, noise outside the control target, such as a person's voice picked up by the control microphone or noise due to vehicle body vibration, often adversely affects the control.

Another problem is that control cannot be applied to sudden changes in noise.

The present invention has been made in view of the above-mentioned problems of the prior art, and it is possible to output an effective additional sound in accordance with the generation of noise without using a control microphone, and has good tracking ability. It is an object of the present invention to provide a vehicle interior noise reduction device.

[0008]

SUMMARY OF THE INVENTION The present invention provides an engine vibration detecting means for detecting engine noise, and a noise estimation parameter for detecting parameters relating to engine speed, vehicle interior temperature, occupant number, and control point position. Detecting means, a neural network for calculating a filter coefficient corresponding to each signal obtained from the noise estimation parameter detecting means by a learning function, and a filter coefficient calculated by the neural network for a signal obtained from the engine vibration detecting means. It comprises variable filter means for performing convolution operation to create an additional sound signal for canceling noise, and a speaker driven by the additional sound signal.

[0009]

By appropriately calculating the filter coefficient for each condition of the noise estimation parameter using the neural network, the filter coefficient can be kept appropriate even for a sudden change in the condition.

Further, a control microphone is not required,
By providing the position information of the control point to the input of the neural network, noise reduction at an arbitrary control point can be achieved.

Further, a control error due to a speech or the like which has occurred in the conventional LMS algorithm is also eliminated.

[0012]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of a vehicle interior noise reduction apparatus according to the present invention;

FIG. 1 shows an embodiment of a device for reducing vehicle interior noise according to the present invention, wherein 1 is a vehicle compartment, 2 is the engine of the vehicle 1, and 3 is the engine vibration of the engine 2 which is directly detected. An engine vibration sensor (a knock sensor may be used) or an engine speed sensor for detecting the engine frequency from the engine speed, 4 is the engine speed,
A noise estimation parameter detection device for detecting parameters such as the temperature in the passenger compartment, the number of occupants, and the position of a control point (set before starting operation by the control position setting device 5 attached to the seat);
Is a neural network for calculating a filter (tap) coefficient based on a detection signal from the noise estimation parameter detecting device 4, and 6 is a neural network for calculating a filter (tap) coefficient from the noise signal by the engine vibration sensor 3 using the filter coefficient calculated by the neural network 5. Is a variable filter unit that creates a speaker output signal that cancels out,, an amplifier that amplifies the output signal of the variable filter unit, and 8 is a speaker that is installed under the driver's feet.

FIG. 2 shows an embodiment of the position of the variable filter 6 shown in FIG. 1, wherein n Z ^-1 are delay elements for delaying the engine vibration X (n) for each sample. , H (0) to h (n−1) are filter (tap) coefficients for multiplying each delay element, and each filter coefficient is updated for each sample by the neural network 5. . An output signal y (n) to the speaker 8 is obtained by performing a convolution operation of multiplying and adding the filter coefficient to the engine vibration X (n) of each sample.
Is required.

The noise is reduced by amplifying the speaker output y (n) by the amplifier 7 and subtracting it from the sound pressure Y (n) actually observed at the driver's ear.

The neural network 5 has a three-layer structure including an input layer 51, an intermediate layer 52, and an output layer 53, as shown in FIG. Further, the number of units of the input layer 51 is set to the number of parameters detected by the noise estimation parameter detecting device 4, and the number of units of the intermediate layer 52 for performing a predetermined operation by multiplying the value of the input layer 51 by an appropriate weighting coefficient. 1
5. The number of filter coefficients of the output layer 53 for calculating the target filter coefficient by adding the outputs of the intermediate layer 52 was set to 200 taps. After the learning of the neural network 5, the value of the weight coefficient in the intermediate layer was fixed.

[0017]

[Effects of the Invention] As described above, the present invention does not require a microphone to be installed beside the driver's ear, and controls the ear position only by giving ear position information before driving. Noise can be reduced.

In addition, since the neural network instantaneously calculates an appropriate filter coefficient for each changing noise condition, it is possible to expect an effect of obtaining controllability and stability.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram showing a position embodiment of a vehicle interior noise reduction device according to the present invention.

FIG. 2 is a block diagram illustrating a detailed configuration of a variable filter unit illustrated in FIG. 1;

FIG. 3 is a schematic diagram for explaining an outline of the neural network shown in FIG. 1;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Cabin 2 Engine 3 Engine vibration sensor 4 Noise estimation parameter detection device 5 Neural network 6 Variable filter 7 Amplifier 8 Speaker

Claims (1)

(57) [Scope of request for utility model registration] An engine vibration detecting means for detecting engine noise; a noise estimation parameter detecting means for detecting parameters relating to an engine speed, a vehicle interior temperature, a number of occupants and a position of a control point; A neural network for calculating a filter coefficient corresponding to each signal obtained from the noise estimation parameter detection means, and a filter coefficient for calculating a convolution of the filter coefficient calculated by the neural network with the signal obtained from the engine vibration detection means to cancel noise. A vehicle interior noise reduction device comprising variable filter means for generating an additional sound signal and a speaker driven by the additional sound signal.