JPH0671195B2 - Adaptive noise eliminator - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an adaptive noise removing device for adaptively removing noise correlated with an interference azimuth beam output from a target azimuth beam output in a standby beam type sonar or the like. is there.

(Prior Art) Conventionally, in such a technical field, an adaptive noise canceller called an adaptive noise canceller has been used, and such a circuit is disclosed in, for example, the following document, "Adaptive Noise Canceller". Noise Canceling: Principles and Applications "Widlow et al., 1975, Proceeding of Eye Triple E Volume 63 PP1692-1716, (“ Adaptive Noise Ca
ncelling: Principles and Applications ", B.WIDROW et
al., Proc. IEEE, vol.63, pp.1692-1716 (1975)).

FIG. 4 shows a block diagram showing the principle of the adaptive noise elimination circuit. In FIG. 4, 1 is a signal source, 2 is a noise source, 3 is an adaptive noise elimination circuit, 4 is a primary input terminal, 5 is a reference input terminal, 6 is an adaptive filter, 7a is an adder, and 7b is adaptive filter control. The circuit, 7c is a convergence constant input terminal, and 8 is an output terminal. The sum (S + N ₀ ) of the signal S from the signal source and the noise N ₀ from the noise source is applied to the primary input terminal of the adaptive noise elimination circuit, and the noise N ₁ from the noise source is applied to the reference input. . Adaptive filter estimates N ₁ to N ₀ To do. In the adder 7a, from the primary input (S + N ₀ ) Is subtracted, and the output terminal Is obtained.

If is estimated well, only the signal S will be available at the output terminal. The adaptive noise canceling circuit 3 includes a noise source 2
To the primary input terminal 4 due to fluctuations in the propagation state of noise
The feature is that even if N ₀ changes, noise can be automatically removed according to the change. The characteristic of the adaptive filter 6 is calculated by the adaptive filter control circuit 7b and is updated sequentially. The adaptive filter control circuit 7b calculates the characteristic of the adaptive filter 6 so that the power of the output (output terminal 8) of the adaptive noise removal circuit 3 is minimized. The calculation method is LMS (LEA
A method based on the ST-MEAN-SQUERE algorithm is used.

This method is disclosed in the above document, but it is necessary to set a convergence constant in order to successively update the characteristics of the adaptive filter. This convergence constant is set from the convergence constant input terminal 7c.

In general, when the value of the convergence constant is large, the convergence is fast, but the accuracy of the adaptive filter after convergence cannot be obtained, and the degree of noise removal deteriorates. On the contrary, when the value is small and the convergence is slower than the fluctuation of the propagation state, the update of the adaptive filter cannot catch up with the fluctuation of the propagation state, and the degree of noise removal is deteriorated. Therefore, the convergence constant is determined in consideration of the fluctuation speed of the propagation state. It should be noted that the reference input is not limited to one, and a plurality of reference inputs may be used according to the number of noise sources.

FIG. 5 is a block diagram showing a configuration example of a conventional adaptive noise elimination device using an adaptive noise elimination circuit. 9 _{1 to} 9 _K are sensors, 10 is a beam former, 11 _{1 to} 11 _N are beam outputs, 12
Is an adaptive noise eliminator, 13 is a beam selection circuit, 14 is an interference direction information input terminal, 15 ₁ is a target direction beam output, and 15 ₂ to 15 _M
Is an interference azimuth beam output, 16 is an adaptive noise elimination circuit, which is an adder 16a, an adaptive filter control circuit 16b, and a convergence constant input terminal 16
c and the adaptive filters 16 _{2 to} 16 _M corresponding to the interference direction beam outputs 15 ₂ to 15 _M. Reference numeral 17 is an output terminal of the adaptive noise canceller 12. In Figure 5, the sensor 9 ₁ -
The signal reaching 9 _K is divided into signal components for each azimuth by the beam former 10, but cannot be completely separated, and the other signal, that is, noise is partially mixed. This noise varies with variation in the propagation state of a signal to the sensor 9 ₁ to 9 _K, but can be removed with use of adaptive noise cancellation device 12. Beam selection circuit of the adaptive noise cancellation device 12 13 is selected by the sum interference azimuth information from the beam output 11 ₁ to 11 _N, the target direction beam output 15 ₁ and the interference azimuth beam output 15 _2-15 enter _M terminal 14 Circuit. Further, the adaptive noise elimination circuit 16 outputs the target direction beam which is the primary input.
The noise of the disturbing direction mixed in with 15 ₁ is the reference input (M
-1) A circuit for removing the interference direction beam outputs 15 ₂ to 15 _M to obtain a signal from only the target direction as an output.

(Problems to be Solved by the Invention) However, in the adaptive noise eliminator having the above configuration, it is necessary to set the convergence constant based on the fluctuation speed of the propagation state. (1) When the fluctuation speed is unknown, (2) When the fluctuating speed is not constant and changes, there is a problem that sufficient noise removal cannot be expected.

The present invention provides an adaptive noise eliminator capable of stably obtaining a noise eliminating effect as described above (1) when the fluctuation speed is unknown and (2) when the fluctuation speed is not constant but changes. With the goal.

(Means for Solving the Problems) The main point of the present invention is that, in the adaptive noise elimination device, the adaptive noise elimination circuit using the adaptive filter controllable by changing the convergence constant is duplicated, and one output is (A) means for calculating the input power, (b) means for storing the value of the calculated power at a predetermined time, (c) means for comparing the calculated power with the value of the power in the storage means, (D) The convergence constant is determined according to the result of the comparison means,
Inputting to a convergence constant control circuit comprising: (e) means for storing the output value of the output convergence constant at a predetermined time, and changing the convergence constant in a predetermined step size, In each case, the output power at the predetermined time is compared with the current output power to find an optimum convergence constant, and thereby the adaptive filter of the other adaptive noise elimination circuit is controlled.

(Operation) According to the present invention, the convergence constant control circuit can obtain the optimum convergence constant by trial and error, and since the adaptive noise elimination circuit is duplicated, it is not unstable and optimal noise elimination is performed. be able to.

(Example) Figure 1 is a block diagram showing an embodiment of the present invention, 9 ₁
9 _K is a sensor, 10 is a beam former, 11 _{1 to} 11 _N are beam outputs, 12 is an adaptive noise canceller, 13 is a beam selection circuit, 14 is a disturbing direction information input terminal, 15 ₁ is a target direction beam output, 15 ₂ ~ 15 _M is an interference direction beam output, 16 is an adaptive noise elimination circuit, 16a is an adder, 16b is an adaptive filter control circuit,
16c is a convergence constant input terminal, 16 _{2 to} 16 _M are adaptive filters, 17
Is an output terminal, 18 is a convergence constant control circuit, and 19 is the same adaptive noise elimination circuit as 16.

In FIG. 1, the signal reaching the sensor is divided into signal components for each direction by the beam former, but it cannot be completely separated, and the other signal, that is, noise, is partially mixed. The beam selection circuit in the adaptive noise eliminator is a circuit that selects the target azimuth beam output and the disturbing azimuth beam output from the beam output based on the azimuth information.

Further, the adaptive noise removing circuit is a circuit for removing the noise in the disturbing direction mixed in the beam output of the target direction, and a signal from only the target direction can be obtained as an output. The convergence constant control circuit 18 is
A circuit for controlling the convergence constant in combination with the convergence constant control adaptive noise elimination circuit 19, which changes the convergence constant to be sent to the convergence constant control adaptive noise elimination circuit 19 to change the convergence constant control adaptive noise elimination circuit 19 Is a circuit that selects the convergence coefficient that minimizes the output power of and sends the convergence coefficient to the adaptive noise elimination circuit 16.

Since the convergence constant control circuit 18 constantly calculates and outputs the convergence constant, the output of the convergence constant control adaptive noise elimination circuit 19 becomes unstable and cannot be used as it is as feedback.

FIG. 2 is a detailed block diagram of the convergence constant control circuit 18,
20 is the output input terminal of the adaptive noise elimination circuit for convergence constant control, 21
Is a square integration circuit, 22 is a power storage circuit, 23 is a power comparison circuit, 24 is a convergence constant changing circuit, 25 is a convergence constant storage circuit, 26
Is an adaptive noise elimination circuit for convergence constant control
27 is a convergence constant output terminal.

The power obtained by the square integration circuit 21 is P ₁ , the power stored in the power storage circuit 22 is P ₂ , and the convergence constant changing circuit is
If the convergence constant in 24 is μ ₁ , the convergence constant stored in the convergence constant storage circuit 25 is μ ₂ , and the step size for obtaining the optimum convergence constant is Δμ, then the convergence constant control circuit 18
It operates like the flow chart shown.

Step 31 Set the values of 1 and μ ₂ to 0.

Step 32 Make the value of P ₂ equal to the value of P ₁ .

Step 33 Add Δμ to the value of μ ₁ .

Step 34: Compare the values of P ₁ and P _{2, and if} the value of P ₂ is large, return to 32.

Step 35 Subtract Δμ from the value of μ ₁ .

Step 36 Make the value of μ ₂ equal to the value of μ ₁ .

Step 37 Subtract Δμ from the value of μ ₁ .

Step 38 Compare the values of P ₁ and P ₂ and go to 40 if the value of P ₂ is small.

Step 39 Make the value of P ₂ equal to the value of P ₁ and go to 37.

Step 40 Add Δμ to the value of μ ₁ .

Step 41 Equalize the value of μ _{2 with} the value of μ ₁ and go to 33.

In the figure,: = is an assignment symbol.

The convergence constant μ ₂ obtained in step 36 or 41 of the convergence constant control circuit 18 in this way is the optimum convergence constant for the propagation state at that time. After that, noise is removed while always calculating the optimum convergence constant.

(Effect of the Invention) As described above, according to the present invention, since the convergence constant control circuit is provided, it is possible to use the optimum convergence constant corresponding to the propagation state. Therefore, optimal noise removal can be expected even when (1) the fluctuation speed of propagation is unknown and (2) when the fluctuation speed of propagation is not constant and changes.

[Brief description of drawings]

1 is a block diagram of an embodiment of the present invention, FIG. 2 is a block diagram of a convergence constant control circuit of the present invention, FIG. 3 is a flow chart for explaining the operation of the convergence constant control circuit shown in FIG. FIG. 4 is a block diagram of an adaptive noise elimination circuit, and FIG. 5 is a block diagram showing a configuration example of a conventional adaptive noise elimination device. 1 ... Signal source, 2 ... Noise source, 3 ... Adaptive noise elimination circuit, 4 ... Primary input terminal, 5 ... Reference input terminal, 6 ...
Adaptive filter, 7a ... Adder, 7b ... Adaptive filter control circuit, 7c ... Convergence constant input terminal, 8 ... Output terminal, 9 ₁ ...
9 _K ...... sensor, 10 ...... beamformer, 11 ₁ ~11 _N
...... Beam output, 12 …… Adaptive noise eliminator, 13 …… Beam selection circuit, 14 …… Interference direction information input terminal, 15 ₂ to 15 _M
…… Interference direction beam output, 16 …… Adaptive noise elimination circuit, 16
a …… adder, 16b …… adaptive filter control circuit, 16c ……
Convergence constant input terminal, 16 _{2 to} 16 _M ... Adaptive filter, 17 ...
Output terminal, 18 ... Convergence constant control circuit, 19 ... Adaptive noise elimination circuit for convergence constant control, 19a ... Adder, 19b ... Adaptive filter control circuit, 19c ... Convergence constant input terminal, 19 _{2 to} 19
_M: Adaptive filter, 20: Adaptive noise elimination circuit output input terminal for convergence constant control, 21 ... Square integration circuit, 22 ... Power storage circuit, 23 ... Power comparison circuit, 24 ... Convergence constant changing circuit , 25 ...... Convergence constant memory circuit, 26 ...... Convergence constant control adaptive noise elimination circuit Convergence constant output terminal, 27 ...... Convergence constant output terminal.

Claims (1)

[Claims] 1. An adaptive noise elimination circuit using an adaptive filter controllable by changing a convergence constant is duplicated, and one output is (a) means for calculating input power, and (b) the calculation. Means for storing the value of the power at a predetermined time; (c) means for comparing the calculated power with the value of the power in the storage means; (d) determining a convergence constant according to the result of the comparing means,
Inputting to a convergence constant control circuit comprising: (e) means for storing the output value of the output convergence constant at a predetermined time, and changing the convergence constant in a predetermined step size, An adaptive noise elimination device characterized in that the output power at the predetermined time is compared with the current output power each time to find an optimum convergence constant, and thereby controls the adaptive filter of the other adaptive noise elimination circuit.