JPH04169883A - Processing signal of sound navigation and ranging - Google Patents

Abstract

PURPOSE:To discriminate objectively a target without the dispersion of a signal interpretation by experience by judging a numerical signal on the basis of the knowledge of a judgment classified by a target and a motion state of the target, and inferring a target signal. CONSTITUTION:A knowledge base control circuit 310 communicates to an inference run circuit 321 when the circuit 310 detects that informations relative to the frequency analysis and amplitude of an input signal are input from a numerical signal processing circuit 200. The circuit 321 grasps that judgment rules relative to the input information are judgment rules 1, 2 and 4, draws these judgment rules in order from a knowledge base circuit 330 and runs on and after 'if' of the judgment rules. Here, the circuit 321 starts a motion state computing circuit 340 because only the rule 4 is an action rule. A track divergence testing circuit 343 evaluates the motion continuity of a target on the basis of the degree of the divergence computed by a Kalman filter tracker circuit 342 and transmits the result to the circuit 321. The circuit 321 recognizes that the probability for the target to be a deep sea submarine can be evaluated applied to the judgment rule 3, draws it from the circuit 330 and runs.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a sonar signal processing method, and in particular, features including frequency components and levels regarding a target signal obtained from numerical processing of an input signal obtained by a passive sonar; This invention relates to a sonar signal processing method for identifying a target by confirming the continuity of the target's barking behavior estimated from the characteristics of the target signal and the consistency with knowledge based on accumulated past experience and experiments.

[Conventional technology]

Traditionally, passive sonar signal processing involves performing numerical signal processing that analyzes the received sound waves and extracts numerical signals of frequency components and levels.The results are displayed, and the operator uses experience and knowledge to determine the target based on this display. was doing the identification.

[Problem to be solved by the invention]

The conventional sonar signal processing method described above only performs numerical processing of input signals, and the process of interpreting and identifying signals by assembling knowledge or estimating the motion state of the target is performed by the operator based on his own experience and knowledge. was.

Therefore, in the process of signal interpretation after numerical signal processing, there are disadvantages in that variations occur due to individual differences in target detection and a lack of objectivity is inevitable.

SUMMARY OF THE INVENTION An object of the present invention is to provide a sonar signal processing method that eliminates the above-mentioned drawbacks, eliminates variations due to individual differences, and enables objective target identification.

[Means to solve the problem]

The sonar signal processing method of the present invention is based on a numerical signal processing means for detecting a numerical signal including frequency components of an input signal obtained by a passive sonar and level information for each predetermined frequency band, and past experience and experiments. In addition to the knowledge of the set target determination processing procedure, it is determined whether or not the numerical signal includes a target signal based on an evaluation scale of the conformity of the numerical signal with the target motion, and the congruence is determined by an operator. Based on the observed target information, the current assumed position of the target including variance is set.

Knowledge signal processing that integrates the target's past position and the assumed position to calculate the current most probable target position and its variance, and determines the degree of divergence of this variance over time as an evaluation measure. and means.

〔Example〕

Next, nine aspects of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention. The configuration of the embodiment shown in FIG. 1 is roughly divided into a receiver 100, a numerical signal processing circuit 200 forming a numerical signal processing means, a knowledge signal processing circuit 300 forming a knowledge signal processing means, and a knowledge signal processing circuit 300 forming a knowledge signal processing means. and a display 400 that displays the identification results subjected to knowledge signal processing.

The numerical processing circuit 200 includes a phasing circuit 201 that performs known phasing processing that eliminates the path difference in input received signals due to the arrangement of the receivers 100 and performs directional synthesis, and a frequency analysis circuit that performs frequency analysis of the input signal. 202, a pitch detection circuit 203, a square integration circuit 204 that outputs level information for each predetermined frequency band, and a band-based level addition circuit 20.
It consists of 5.

Further, the knowledge signal processing circuit 300 includes the knowledge signal processing circuit 3
knowledge base management circuit 310 that controls the exchange of data within 00, and knowledge regarding the output of the numerical signal processing circuit 200 and the determination processing procedure stored in the knowledge database circuit 330 in advance.

An inference circuit 32 that estimates the certainty of the target signal included in the numerical signal based on the match with the target exercise state provided from the exercise state calculation circuit 340 and identifies target information.
0, a knowledge base circuit 330 that stores knowledge regarding processing procedures for target determination based on empirical rules and an experimental period;
It also includes a motion state calculation circuit 340 that receives target information observed by an operator and calculates the motion state of the target by Kalman filtering.

Next, the operation of the ninth embodiment will be explained.

The input signal captured by the wave receiver 100 is subjected to predetermined phasing by a phasing circuit 201, and is then passed through 5 frequency analysis circuits 202 and 2.
A power analysis circuit 204 is provided.

The frequency analysis circuit 202 performs frequency analysis of the input signal using a known method such as FFT, and the pitch detection circuit 20
3 determines the pitch regarding the analysis frequency and outputs it as frequency information regarding the input signal.

The square integration circuit 204 calculates the level (amplitude) by taking the integral value of the input signal squared, and this level is added for each level of each predetermined frequency analysis band in the band-specific level addition circuit 205, and the 9 frequencies are added. The cumulative value of the level for each analysis band is output as one novel information of the input signal. These frequency information and level information are supplied to the knowledge signal processing circuit 300 as numerical signals of input signals. This numerical signal is 9For example, if the input signal includes a target signal and it includes pitch information existing in a certain frequency band,
The addition level in that frequency band increases with time, exhibiting a so-called cumulative accumulation state, and exhibiting the maximum amplitude for each frequency band. Conventionally, an operator has observed this numerical signal through the display 400 and confirmed and identified the presence of a target in the input signal based on experience and knowledge.

The knowledge signal processing circuit 300 inputs the numerical signal output from the numerical signal processing circuit 200 to the knowledge base management circuit 310.

The knowledge base management circuit 310 is the knowledge signal processing circuit 300
It has a function to control smooth transfer of internal data. The input signal is passed unconditionally to the inference circuit 320 via the knowledge base management circuit 310. The inference circuit 320 always keeps the inference execution circuit 321 aware of what kind of knowledge the knowledge base circuit 330 has regarding the given information.

Necessary knowledge is extracted from the empirical rule storage circuit 331 and the experiment side storage circuit 332 of the knowledge base circuit 330 via the knowledge base management circuit 310.

FIG. 2 is a diagram illustrating the judgment rules stored in the knowledge base circuit 330 of FIG. 1. The knowledge base circuit 330 includes an empirical rule storage circuit 331 and an experimental side storage circuit 332, each of which is set based on experience and experiment. Contents related to four determination rules as knowledge indicating processing procedures for target determination are stored in a program format.

Judgment rule 1 shown in FIG. 2 is a rule regarding the process procedure for target judgment based on the pitch as frequency information of the input signal.If the pitch P of the input signal is determined, then if P≧n, then 1 + P<n If so, it indicates that it is processed as O. Here, n is a target regarding the pitch P, in this case a scale for determining the suitability of a deep-sea submersible, and is set in advance by experience or experiment. If P≧n, then 1 means that if the pitch P takes a value exceeding n, it is determined that the input signal having that pitch is generated by a deep-sea submersible. Conversely, if P<n, 0 means that it is determined that the input signal is not typical of a deep-sea submersible as far as the pitch is concerned.

Such a determination rule regarding pitch is set based on the fact that if the pitch is greater than a certain level, that is, unless the frequency is below a certain level, the target is not a low frequency component emitted by a screw or the like.

In addition, 9 decision rule 2 is a decision rule based on the level (amplitude) information of the input signal, and is a decision rule for determining whether the frequency region in which the frequency band F indicating the maximum amplitude of the input signal exists is between fl and 12. +fl and f2 are set in advance based on actual results, estimates, etc., corresponding to the characteristics of the spontaneous signal of the deep-sea submersible.

In addition, 9 judgment rule 3 is the pitch P of the input signal.

When the band F indicating the maximum amplitude is all 1, or when the wake divergence degree indicating the degree of convergence and divergence of the wake dispersion is less than a predetermined value Q, it can be considered that there is not much divergence of the wake. , the processing rule is to determine that what the input signal indicates is a deep-sea submersible.

Further, 9-determination rule 4 provides knowledge regarding the motion analysis of the target, and is a broadly defined determination rule for calculating the degree of wake divergence among the conditions necessary for the determination of 9-determination rule 4.

Judgment rules 1 to 4 are based on the type of target,
Empirical rule memory circuit 3 based on past operational results, experience, etc.
31 or the experimental storage circuit 332, but which judgment rule is assigned to which storage circuit is not necessarily fixed.

The allocation will be changed from time to time depending on the operational purpose and goals.

The knowledge base management circuit 310 is a numerical signal processing circuit 200
When it detects that information regarding the frequency analysis and amplitude of the input signal has been input from.

This is notified to the inference execution circuit 321.

Since the inference execution circuit 321 knows that the decision rules related to the input information are decision rules 1.2 and 4 in FIG. 2, it sequentially applies these decision rules via the knowledge base management circuit 310. From the knowledge base circuit 330, the steps after "if" of the decision rule 9 are executed. Here, only judgment rule 4 is a broad judgment rule regarding 9 actions rather than factual knowledge, and the inference execution circuit 321 uses the knowledge base to indicate the maximum amplitude for each frequency band. Conventionally, an operator has observed this numerical signal through the display 400 and confirmed and identified the presence of a target in the input signal based on experience and knowledge.

The knowledge signal processing circuit 300 inputs the numerical signal output from the numerical signal processing circuit 200 to the knowledge base management circuit 310.

The knowledge base management circuit 310 is the knowledge signal processing circuit 300
It has a function to control smooth transfer of internal data. The input signal is passed unconditionally to the inference circuit 320 via the knowledge base management circuit 310. The inference circuit 320 always keeps the inference execution circuit 321 aware of what kind of knowledge the knowledge base circuit 330 has regarding the given information.

Necessary knowledge is extracted from the empirical rule storage circuit 331 and the experiment sub-storage circuit 332 of the knowledge base circuit 330 via the knowledge base management circuit 310.

FIG. 2 is a diagram illustrating the judgment rules stored in the knowledge base circuit 330 of FIG. 1. The knowledge base circuit 330 includes an empirical rule storage circuit 331 and an experimental sub-memory circuit 332, each of which is set based on experience and experiment. Contents related to four determination rules as knowledge indicating processing procedures for target determination are stored in a program format.

Judgment rule 1 shown in FIG. 2 is a rule regarding the process procedure for target judgment based on the pitch as frequency information of the input signal.If the pitch P of the input signal is determined, then if P≧n, then 1 + P<n If so, it indicates that it is treated as 0. Here, n is a target regarding the pitch P, in this case a scale for determining the suitability of a deep-sea submersible, and is set by preliminary experience or experiment. If P≧n, then 1 means that if the pitch P takes a value exceeding n, it is determined that the input signal having that pitch is generated by a deep-sea submersible. Conversely, if P<n, 0 means that it is determined that the input signal is not typical of a deep-sea submersible as far as the pitch is concerned.

Such a determination rule regarding pitch is set based on the fact that if the pitch is a hole above a certain level, that is, unless the frequency is below a certain level, the target is not a low frequency component emitted by a screw or the like.

Is Kei's level of manual labor a rule? This is a judgment rule based on the frequency band F indicating the maximum amplitude of the input signal, and the judgment rule is whether the frequency band F indicating the maximum amplitude of the input signal exists, and the judgment rule is whether or not the frequency band F is between fl and 12. + fl and f2 are deep sea It is set in advance based on actual results, estimates, etc., corresponding to the characteristics of the submersible's spontaneous signal.

In addition, 9 judgment rule 3 is the pitch P of the input signal.

When the band F indicating the maximum amplitude is all 1, or the wake divergence degree indicating the degree of convergence 9 divergence of the wake dispersion is less than a predetermined value Q, and it can be considered that there is not much divergence of the wake. , the processing rule is to determine that what the input signal indicates is a deep-sea submersible.

Furthermore, 2nd judgment rule 4 provides knowledge regarding the motion analysis of the target, and is a broadly defined judgment rule for calculating the degree of wake divergence among the conditions necessary for the judgment of 1st judgment rule 4.

Judgment rules 1 to 4 are based on the type of target,
Past operation actual m, l! The judgment rules are distributed and stored in the heuristic rule storage circuit 331 or the heuristic rule storage circuit 332 based on experiments, etc., but which judgment rule is assigned to which storage circuit is not necessarily fixed.

The allocation will be changed from time to time depending on the operational purpose and goals.

The knowledge base management circuit 310 is a numerical signal processing circuit 200
When it detects that information regarding the frequency analysis and amplitude of the input signal has been input from.

This is notified to the inference execution circuit 321.

Since the inference execution circuit 321 knows that the decision rules related to the input information are decision rules 1.2 and 4 in FIG. 2, it sequentially applies these decision rules via the knowledge base management circuit 310. From the knowledge base circuit 330, the steps after "if" of the decision rule 9 are executed. Here, only judgment rule 4 is a broad judgment rule regarding 9 actions, not knowledge about facts, and the inference execution circuit 321 is based on the knowledge base shown in FIG.

Claims (1)

[Claims] Knowledge of numerical signal processing means for detecting numerical signals including frequency components of input signals obtained by passive sonar and level information for each predetermined frequency band, and target determination processing procedures set based on past experience and experiments. In addition to this, it is inferred whether or not the numerical signal includes a target signal using the matching of the numerical signal with the target motion as an evaluation criterion, and the matching includes variance based on target information observed by the operator. After setting the current hypothetical position of the target, the past position of the target and the hypothetical position are integrated to calculate the current most probable target position and its variance, and the variance of this variance over time is calculated. 1. A sonar signal processing method, comprising knowledge signal processing means for determining a degree of divergence as an evaluation measure.