CN1332186C - Weak signal detection under complex background and characteristic analysis system - Google Patents

Abstract

The present invention relates to a weak signal detecting and characteristic analyzing system under a complicated background, which is used for automatically detecting and recording environmental noise of the sea and the noise characteristic of sail. The present invention comprises a detection subsystem, a signal processing subsystem and an auxiliary subsystem, wherein the detection subsystem is composed of an amplifier, a gain attenuation network, a program controlling band pass filter, a limiting amplifier, a comparator and a chip microprocessor. The detection of a weak signal is completed by the combination of zero-crossing quantity, a gain value and Nyman-Pearson detection; the signal processing subsystem is composed of a universal peripheral interface, a high-speed numerical processor and a collector and is used for analyzing noise and recording spectrum characteristics; the auxiliary subsystem is composed of a power supply module, a power-fail monitoring module, a parallel interface and a serial interface; the auxiliary subsystem can be connected with a printer and a computer for further research and information storage; the data of the chip microprocessor and the high-speed numerical processor is transmitted to a memory by the universal peripheral interface. The present invention has the advantages of small size, strong detection capability, long service life, etc.

Description

Detection of Weak Signals and characteristic analysis system under the complex background

Technical field

The present invention relates to Detection of Weak Signals and characteristic analysis system under a kind of complex background, be a kind of automatic detection and write down ambient sea noise and the device of the noisiness of navigating, relate to fields such as input, pattern-recognition, can directly apply in the early warning system on duty and fight fuse of underawater ordnance.

Background technology

There are very abundant ocean resources in China; understanding the numerous hydrologic datas of ocean in each marine site and the halobiontic activity in a certain marine site of follow-up study all is very significant for economic development, environmental protection etc., research navigate noisiness to the noise reduction process of boats and ships for the wide adjacent sea safety of defendance China important national defense mission especially then.Therefore it is necessary developing a kind of device that can work and need not manual intervention the long period under water.At present domestic detection system at hydroacoustic noise research has many, corresponding underwater sound checkout equipment is also arranged, for example the B﹠amp of Denmark abroad; The underwater sound equipment of K company is very advanced.But all there are many shortcomings in these systems and product: the big or fixed in shape of (1) equipment volume can't form the system that subsystem is used for particular requirement; (2) cost an arm and a leg, make system cost promote significantly, can finish the performance of native system at present substantially with the A/D device of low-power consumption, but cost too high (about tens times even hundreds of times); (3) full gain and gain unit-sized are fixed, and can not change flexibly according to the actual requirements, use inconvenient.On the other hand detection algorithm too complicated (as Wu Guoqing etc., Underwater Acoustic Transient Signal short-time spectrum form and spectrum correlation method detect, acoustic journal, Vol.25, No.6 510-515) can't use in the real-time detecting system of reality.

Summary of the invention

The objective of the invention is at the deficiencies in the prior art, Detection of Weak Signals and characteristic analysis system under a kind of complex background are provided, be implemented in and survey and write down the ambient sea noise and the noisiness of navigating under the complex background situation for a long time automatically, have that volume is little, detectability reaches advantages such as dependable performance by force.

For realizing such purpose, designed system of the present invention comprises detection subsystem, signal processing subsystem and assistant subsystem.Detection subsystem is made of amplifier, gain reduction network, control tape bandpass filter, limiting amplifier, comparer and chip microprocessor, utilizes zero passage amount and yield value to detect in conjunction with Nyman-Pearson and finishes Detection of weak; Signal processing subsystem is made of universal peripheral interface, high speed numerical processor and collector, analyzes noise and writes down spectral characteristic and the time domain statistical property; Assistant subsystem is made of power module, power down monitoring module, parallel interface and serial line interface, can be connected with printer, computing machine, can do further research and information stores; The data of chip microprocessor and high speed numerical processor are transferred to storer by universal peripheral interface.

System of the present invention adopts modular construction, mainly by detection subsystem, signal processing subsystem and assistant subsystem are formed, concrete structure is: the signal of input is connected to amplifier through nautical receiving set, amplifier is linked chip microprocessor and controlled by it by the gain reduction network, amplifier output connects the input of control tape bandpass filter, the output of control tape bandpass filter is connected to comparer through limiting amplifier, comparer and limiting amplifier link to each other with chip microprocessor respectively, amplifier, the gain reduction network, the control tape bandpass filter, limiting amplifier, comparer and chip microprocessor constitute detection subsystem.Chip microprocessor is connected with universal peripheral interface in the signal processing subsystem realizes exchanges data, simultaneously the universal peripheral interface realization exchanges data that links to each other with high speed numerical processor.High speed numerical processor is connected with a collector, and the input of collector inserts from the output of control tape bandpass filter, and universal peripheral interface, high speed numerical processor and collector constitute signal processing subsystem.High speed numerical processor connects computing machine by a serial line interface, and universal peripheral interface connects printer by a parallel interface; The power supply of all devices is provided by power module, and power module is connected with the power down monitoring module, and the power down monitoring module is connected with chip microprocessor, and power module, power down monitoring module, parallel interface and serial line interface constitute assistant subsystem; Universal peripheral interface links to each other with storer, and the data of chip microprocessor and high speed numerical processor are transferred to storer by universal peripheral interface.

In detection subsystem of the present invention, gain amplifier network and gain reduction network that the control of utilization low-power consumption chip microprocessor is formed by the multi-stage operational amplifier cascade, thereby reach input signal is amplified monitoring in real time, make the amplitude of signal be controlled in a fixed value near.Near the function whether the detection signal amplitude is controlled in the fixed value utilizes control tape bandpass filter, limiting amplifier and comparer to realize.The situation of change that at this moment just can reflect signal amplitude to the yield value of signal.The process of control signal amplitude gain is to control operational amplifiers at different levels in amplification or the gain network by chip microprocessor with the multidigit binary number.Concrete figure place can be selected according to the precision of practical problems needs.The measuring ability of detection subsystem is realized by microprocessor by detecting software.Hardware detection subsystem and detect more module of value that software constituted system of the present invention jointly.

Be based on detection software of the present invention dualism hypothesis and Nyman-Pearson (Nai Man-Pearson came) criterion design under the certain situation of false-alarm probability, and detection statistic to be the average zero passage of signal in each second count and the amplitude average in each second.Observation signal is the composite signal of ambient sea noise and Ship Radiated-Noise.When no boats and ships occur, ambient sea noise is only arranged in the observation signal, the average zero passage in its each second is counted and the amplitude average in each second is steadily with substantially invariable through smoothing processing.But after boats and ships occurred, owing to the enhancing of boats and ships signal, the average zero passage in each second was counted and the amplitude average in each second can change in the observation signal, and promptly zero passage is counted and diminished along with the close of boats and ships, and amplitude becomes greatly along with the close of boats and ships.Utilize Nyman-Pearson criterion and given false-alarm probability, can be in the hope of the threshold value of judging that boats and ships occur.When all just producing during exceeded threshold, report to the police in continuous three times (three seconds).

Specialized designs of the present invention the assistant subsystem of universal peripheral interface, make system have modular advantage, also reduced simultaneously the power consumption of system.Assistant subsystem mainly contains two functions: a function is the function that cooperates detection subsystem, has designed power module and system's power down monitoring module.Power module provides stable voltage and current for system, the function of power down monitoring module is under the unexpected power-down conditions of generation systems, the notice microprocessor is preserved data and system state, when being subjected to the external interference program in the microprocessor that makes, system takes place to preserve data and system state under " deadlock " situation in addition, and restart microprocessor, work on, guarantee that detection system can not lose efficacy.The another one function of assistant subsystem is to finish the work when native system, when being recovered to the laboratory, the assistant subsystem tool can be connected with computing machine by serial line interface, import the data that write down in the system into computing machine, also can be connected with printer, detection curve be printed by printer by parallel interface.

Signal processing subsystem of the present invention has used the low-power consumption high speed numerical processor, has solved the requirement of signal processing algorithm to arithmetic capability.Confirm to have boats and ships when detection subsystem and occur, information is issued signal processing subsystem, signal processing subsystem will detect data and threshold value stores.Start low-power consumption A/D acquired signal, make fast Fourier transform (FFT), detect line spectrum, whether the general shape of estimating power really occurs so that further confirm boats and ships.

The modular construction that the present invention adopts subsystem to constitute can change flexibly according to requirement in practical systems, has reduced system cost.Detection subsystem is utilized zero passage detection, amplitude detection principle detection signal, not only adopts no A/D and accomplishes the total digitalization processing, and can estimate noise intensity.Advantages such as The actual running results shows, system of the present invention has volume little, low in energy consumption (＜10 milliwatt), detectability is strong and the life-span is long.

Description of drawings

Fig. 1 is a system principle diagram of the present invention.

The statistical average that becomes during Fig. 2 in the background detects schematic diagram.

Embodiment

Below in conjunction with drawings and Examples technical scheme of the present invention is further described.

System architecture theory diagram of the present invention mainly is made of detection subsystem as shown in Figure 1 three parts: amplifier, gain reduction network, control tape bandpass filter, limiting amplifier, comparer, chip microprocessor; Signal processing subsystem: low-power consumption high speed numerical processor and collector, universal peripheral interface; Assistant subsystem: power supply, power down monitoring, serial line interface and parallel interface.Wherein chip microprocessor is a master cpu, and whole tasks of system are finished by its control.The front end nautical receiving set converts acoustical signal to electric signal and imports native system through prime amplifier, and the data of chip microprocessor and high speed numerical processor are transferred to storer by universal peripheral interface.

The concrete connection of system is as follows:

System is divided into detection subsystem, signal processing subsystem and assistant subsystem.The signal of input arrives amplifier through nautical receiving set, amplifier is linked chip microprocessor GMS90C32 and controlled by it by the gain reduction network, amplifier output connects the input of control tape bandpass filter, and the output of control tape bandpass filter is connected to comparer through limiting amplifier.Comparer, limiting amplifier link to each other with chip microprocessor GMS90C32 respectively.Amplifier, gain reduction network, control tape bandpass filter, limiting amplifier, comparer and chip microprocessor GMS90C32 have constituted detection subsystem.Chip microprocessor GMS90C32 is connected with universal peripheral interface ZPSD301A in the signal processing subsystem and realizes exchanges data, and universal peripheral interface ZPSD301A links to each other with high speed numerical processor ADSP2181 simultaneously, the realization exchanges data; High speed numerical processor ADSP2181 is connected with a collector, and the input of collector is the output access from the control tape bandpass filter.ZPSD301A, ADSP2181 and collector have constituted signal processing subsystem.ADSP218 connects a serial line interface, in order to being connected with computing machine; ZPSD301A is connected with a parallel interface, in order to connecing printer; The power supply of all devices has power module to provide; Whether and power module is connected with power down monitoring module (MAX706P), normal with the voltage of monitoring power supply; The monitoring module of power down simultaneously is connected with GMS90C32, will provide warning to GMS90C32 when power work is undesired.Power module, power down monitoring module, parallel interface and serial line interface have constituted assistant subsystem.Universal peripheral interface ZPSD301A links to each other with storer, with the data transmission of chip microprocessor GMS90C32 and high speed numerical processor ADSP2181 to storer.

1, detection subsystem

The main task of this subsystem is to survey the generation of extraordinary noise in the ocean, makes the judgement of being and not being.

Consider the needs of the output and the digital processing of nautical receiving set, total amplification quantity of having determined amplifier is 80dB, CPU is by low six control attenuation networks of gain control word, control accuracy is 1.25dB, the overall attenuation of attenuation network is 78.75dB so, reaches the purpose of control channel full gain thus.The control tape bandpass filter is by high two controls of gain control word, its band connection frequency with only with frequency can according to research object different and respectively the setting, the present invention experiment to as if ship noise, therefore the needs according to research are set to 100-800Hz and two kinds of wave filters of 10-10000Hz respectively with two frequency ranges; The former serves zero passage detection, and the latter serves limit band energy measuring (amplitude detection).Comparer among the figure is a voltage threshold, is used for gain control.Signal thresholding therewith compares, and comparative result is a series of pulses, and the T2 (timer/counter) that gives CPU does second numeration, and T0 (timer/counter) provides interrupt a second.One pulse number double threshold is set again in software, and with the pulse number remembered and last Lower Threshold relatively, its result adjusts the foundation of attenuation network as CPU, and corresponding increase and decrease is done in the gain control word adjust, and makes passage output keep constant output average.Meanwhile CPU writes down this gain control word.The output signal of programmable filter will be amplified by shaping again, become a series of pulses, pass to the T1 (timer/counter) of CPU, carry out the zero crossing numeration of each second.

The function of chip microprocessor CPU in the detection subsystem:

(1) adjusting the gain control word makes output stable.

(2) CPU handles a second interruption, reads the zero passage in a second and counts and the gain control number.

(3) respectively to zero passage X and the amplitude fading amount Q average treatment that takes statistics of counting, estimate average, the variance of background zero crossing and amplitude relative quantity, and produce automatic threshold; Judgement in real time then is if having one to find that signal occurs, and will start high speed numerical processor ADSP2181 among both.

(4) monitoring supply voltage is done power down protection and is handled.When supply voltage during less than predefined warning value, power down monitoring module MAX706P sends to CPU and interrupts application, and CPU will in time preserve the parameter that current state and acquisition process are crossed.

(5) finish dealing with after, enter dormant state automatically.According to calculating, system CPU of the present invention has time of 95% to be in dormant state, guarantees that like this CPU average power consumption is very low.

(6) storage of control data drives printer, with the serial communication management of microcomputer, carries out exchanges data with high speed numerical processor ADSP2181.

Realize the software algorithm of detection in this subsystem, promptly the pre-service judging process is united realization by zero passage detection and amplitude detection, as long as there is a kind of detection to find signal, system is with the enabling signal processing subsystem.Certainly, enabling signal processing subsystem under which kind of situation can be set in program as required.

It is that a kind of counting based on zero passage in the unit interval of statistics limit tape input signal waveform changes the method that detects feeble signal that zero cross signal detects.If the appearance of a signal has caused the variation of power spectrum composition, then the unit interval zero passage average of counting also changes.The pass of average A that the unit interval zero passage of stable Gaussian random noise is counted and power spectrum density W (f) is:

$A$ $=2\·$ $[{\∫}_0^{\∞}{f}^{2}W\left(f\right)\mathrm{df}/{\∫}_0^{\∞}W\left(f\right)\mathrm{df}{]}^{1/2}---(1)$

Be with Gaussian noise to add the situation of sinusoidal component (line spectrum component) to limit, the average of counting according to the unit interval zero passage of Rice formula derivation is:

$A=2$ $\·[\frac{\frac{1}{3}(f_L^2+f_L{f}_A+f_H^2)+{\mathrm{rf}}_s^2}{1+r}{]}^{1/2}---(2)$

F wherein _HBe upper cut-off frequency frequency, f _LBe passband lower frequency limit, f _sBe the frequency of line spectrum component, r is the power ratio of noise with continuous spectrum in the line spectrum component is with limit.In the radiated noise of navigating, there are many line spectrum components that produce by thruster, so its appearance variation that will cause zero passage to count.The assembly average that the zero passage of common marine background noise is counted changes slowly, and Gaussian distributed, so its variation can be considered the variation of parameter (average and variance).The continuous parameter estimation approach can obtain zero crossing number average value the time become automatic threshold, and detect the significant change that zero passage that thruster causes is counted.The outstanding advantage of zero passage detection is that antijamming capability is strong.

The practical optimal detector of limit band energy measuring is the quadratic detection integrator.To the Testing of Feeble Signals of low signal-to-noise ratio, the output signal-to-noise ratio loss that replaces quadratic detection therefore to cause with linear detection is 0.57dB, but computing is greatly simplified.Detection subsystem is exactly to carry out linear amplitude digital signal detection.The full gain DR of the whole passage of system (dB) is certain, and attenuation network is the six bit W that provided by chip microprocessor _i(each attenuation unit is 1.25dB) that controls, so gain reduction amount Q is:

Q＝W _N·DR/2 ¹+W _N-1·DR/2 ²+…+W ₁·DR/2 ^N (3)

Since output is constant, the variation of Q has reflected the dynamic change of signal amplitude, and is six bit binary number amounts, is easy to be no A/D with CPU and is connected, and carries out digitizing and detects.If during greater than band limited noise bandwidth reciprocal, can think uncorrelated between the Q value sample value to the sample time spaced far of Q, thereby the statistics that sampling can be made signal amplitude to Q detects.

It is when finishing the ship noise detection that the present invention limits the outstanding advantage of band energy measuring, can utilize the Q average statistical to provide the valuation (when the output average of given nautical receiving set Sensitirity va1ue, pretreatment system gain maximum and setting) of Ship Radiated-Noise intensity.

Since simultaneously with the gain control weight coefficient W of band limited noise amplitude detection and unit interval zero passage count A detect input noise the time, frequency domain characteristic variation, they are binary number, therefore can handle with the similar detection scheme, save resource and the time of CPU.Be detection scheme below:

The detected segment length of stochastic process X (t) can represent with multi-C vector X for the sample value sequence of N:

X＝[x ₁，x ₂，…，x _N] ^T

x ₁, x ₂..., x _NSeparate.Suppose that the average when it signal do not occur is A ₀, the signal hourly value occurring is A ₁, variance is σ ², then Xiang Liang conditional probability density is:

$P$ $\left(X\right|H_0)\frac{1}{({2\mathrm{\π\σ}}^2{)}^{N/2}}\exp [-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\frac{\left(X\right(i)-A_0{)}^2}{{2\mathrm{\σ}}^2}]---\left(4\right)$

$P$ $\left(X\right|H_1)=\frac{1}{(2\mathrm{\π}{\mathrm{\σ}}^2{)}^{N/2}}\exp [-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\frac{(X\left(i\right)-A_1{)}^2}{{2\mathrm{\σ}}^2}]---\left(5\right)$

H wherein ₀Signal, H do not appear in expression ₁The expression signal occurs; Try to achieve the likelihood ratio function

$\mathrm{\λ}\left(X\right)=P\left(X\right|H_1)/P\left(X\right|H_0)=\exp [-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\frac{\left(X\right(i)-A_1{)}^2}{{2\mathrm{\σ}}^2}+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\frac{(X\left(i\right)-A_0{)}^2}{{2\mathrm{\σ}}^2}]$

$=\exp [\frac{A_1-A_0}{{\mathrm{\σ}}^2}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}X\left(i\right)-\frac{N(A_1^2-A_0^2)}{{2\mathrm{\σ}}^2}]---\left(6\right)$

$\ln \left(\mathrm{\λ}\right(X\left)\right)=\frac{A_1-A_0}{{\mathrm{\σ}}^2}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}X\left(i\right)-\frac{N(A_1^2-A_0^2)}{{2\mathrm{\σ}}^2}---\left(7\right)$

Suppose a thresholding λ T, have

$\mathrm{\&lambda;}\left(X\right)=\exp [\frac{A_1-A_0}{{\mathrm{\&sigma;}}^2}\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}X\left(i\right)-\frac{N(A_1^2-A_0^2)}{{2\mathrm{\&sigma;}}^2}]\begin{array}{c}{H}_1\\ >\\ <\\ H_0\end{array}{\mathrm{\&lambda;}}_T---\left(8\right)$

In order to reduce the variance of statistical test amount, the statistical test amount is done average treatment, then average is constant, and variance is σ ²/ N, have this moment: $\tilde{X}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}X\left(i\right),$ , be referred to as short sample parameter and estimate.Can obtain

$\tilde{X}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}X\left(i\right)\begin{array}{c}{H}_1\\ >\\ <\\ H_0\end{array}\frac{A_1+A_0}{2}+\frac{{\mathrm{\&sigma;}}^2\ln {\mathrm{\&lambda;}}_T}{N(A_1-A_0)}={\tilde{X}}_c---\left(9\right)$

Use the Neyman-Pearson criterion here, suppose the false-alarm probability P of detecting device _fBe α, then have

$P_f={\&Integral;}_{{\tilde{x}}_c}^{\&infin;}P\left(\tilde{X}\right|H_0)d\tilde{X}=\mathrm{\&alpha;}---\left(10\right)$

Because observation and statistic amount X becomes when by ambient sea noise being over time, therefore to the statistic of neighbourhood noise

The real parameter estimation of carrying out when needing.Suppose to handle the mean variance that estimates the observation and statistic amount,, obtain the confidence interval of mean of observation and statistic amount according to the false-alarm probability that detecting device can be allowed by statistical average.For

, its average is

, variance is σ ²/ N; When N is big, observation and statistic amount Θ

$\mathrm{\&Theta;}=\frac{\tilde{X}-\tilde{A}}{\mathrm{\&sigma;}/\sqrt{N}}---\left(11\right)$

Average is 0, and variance is 1, and we are referred to as long sample parameter and estimate.According to the false-alarm probability α that detecting device can be allowed, obtain statistic confidence interval of mean 100 (1-α) %, definition Θ _αFor being the statistic under the α condition in false-alarm probability,, should have again according to gauss of distribution function

P(Θ≤Θ _α)＝1-α (12)

Obtain being limited on the fiducial interval

Because σ ²Be unknown, so it need be by estimating to obtain S ²Approach, n is a sample length here, and the statistic of sample is like this

${\mathrm{\&Theta;}}_{\tilde{x}}=(\tilde{x}-\tilde{A})/s\sqrt{N}---\left(14\right)$

The t that obeys the n-1 degree of freedom distributes.Be limited on the fiducial interval

${\mathrm{\&omega;}}_{\mathrm{wp}}=\widetilde{\mathrm{\&mu;}}+\frac{s}{\sqrt{n}}\&CenterDot;t_{\mathrm{\&alpha;}}---\left(15\right)$

Wherein Be statistic

Average, t _αBe to be that α and degree of freedom are the value of finding in the t distribution table under the n-1 condition in false-alarm probability.

According to the Neyman-Pearson criterion, under the situation of given false-alarm probability, calculate likelihood ratio and set thresholding λ _TOptimal value be AT, make the detection probability maximum.Though actual background right and wrong stably, thereby average A ₀Become when being, but the variation of marine background noise is slow fortunately, needs a few minutes even tens minutes just can see the significant change of its parameter (average, variance) sometimes, thus can do judgement according to following formula, but different values should be arranged to zero passage detection A or energy measuring W.

Become input computing block diagram in the background when Fig. 2 provides,, estimate to describe marine background A average A with long sample parameter with the variation that short sample parameter Estimation of Ship noise causes ₀And variances sigma ²Valuation.The interval A that calculates of utilization ₀Fiducial interval, and produce the real-time automatic threshold that likelihood ratio detects.

2, assistant subsystem

Parallel interface is connected with printer, plotting apparatus, and CPU can unify zero passage is printed by family curve, each analysis of spectrum curve and some characteristic parameters by family curve, energy.Serial port is used for and the microcomputer communication.

In addition in this subsystem, by power down monitoring module MAX706P, the real-time monitoring power voltage of CPU, when supply voltage is lower than setting value, CPU will reset automatically, and guarantee can not cause because of the voltage fluctuation in the external world CPU to crash.The power supply of system is made of with more corresponding peripheral components MAX639.This power supply input power requirement to external world very low (only needing 5-11V) just can provide system desired several voltages.

3, signal processing subsystem

After zero passage detection or energy measuring discovery ship noise signal and the appearance of biological group noise, trigger high speed numerical processor ADSP2181 immediately, make ADSP2181 break away from dormant state, start the A/D collector, acquired signal, and utilize these data to do various analyses.The groundwork of high speed numerical processor ADSP2181 is to finish in subsequently each one second:

(1) starts back realization, the instantaneous spectrum of calculating noise and eight average power spectras with fft algorithm.

(2) note each constantly zero passage count and amplitude gain control number.

(3) carry out exchanges data and deposit data in storer by universal peripheral interface ZPSD301A and CPU.

(4) enter dormant state automatically after finishing the work, the working time of ADSP2181 can shorten like this, and average power consumption is reduced to significantly less than the 1 milliwatt order of magnitude.

The termination of these work was a null value and equal consistent moment of value when reporting to the police of range value constantly.

When chip microprocessor GMS90C32 startup high speed numerical processor device ADSP2181, the power spectrum that its main work is signal calculated, line spectrum detection, the calculating of statistic and the storage of data.

Software algorithm in the signal processing subsystem is to adopt basic 2 fft algorithms to realize power spectrumanalysis, and eight spectrums of doing at 2048 are average, effectively analyze frequency range 800Hz, and frequency resolution is 1Hz.The ADSP2181 chip realizes that be 1.07 milliseconds FFT operation time of 1024 fixed points, and be 3 milliseconds the operation time of Floating FFT, thereby can realize real-time calculating.Averaging spectrum is done line spectrum detect, and preserve the position of line spectrum and the amplitude of line spectrum.

Claims (3)

1, Detection of Weak Signals and characteristic analysis system under a kind of complex background, it is characterized in that by detection subsystem, signal processing subsystem and assistant subsystem are formed, the signal of input is connected to amplifier through nautical receiving set, amplifier is linked chip microprocessor and controlled by it by the gain reduction network, amplifier output connects the input of control tape bandpass filter, the output of control tape bandpass filter is connected to comparer through limiting amplifier, comparer and limiting amplifier link to each other with chip microprocessor respectively, amplifier, the gain reduction network, the control tape bandpass filter, limiting amplifier, comparer and chip microprocessor constitute detection subsystem; Chip microprocessor is connected with universal peripheral interface in the signal processing subsystem realizes exchanges data, simultaneously the universal peripheral interface realization exchanges data that links to each other with high speed numerical processor; High speed numerical processor is connected with a collector, and the input of collector inserts from the output of control tape bandpass filter, and universal peripheral interface, high speed numerical processor and collector constitute signal processing subsystem; High speed numerical processor connects computing machine by a serial line interface, and universal peripheral interface connects printer by a parallel interface; The power supply of all devices is provided by power module, and power module is connected with the power down monitoring module, and the power down monitoring module is connected with chip microprocessor, and power module, power down monitoring module, parallel interface and serial line interface constitute assistant subsystem; Universal peripheral interface links to each other with storer, and the data of chip microprocessor and high speed numerical processor are transferred to storer by universal peripheral interface.

2, according to Detection of Weak Signals and characteristic analysis system under the complex background of claim 1, the measuring ability that it is characterized in that described detection subsystem is realized by chip microprocessor by detecting software, the observation and statistic amount is that the average zero passage of signal in each second counted and the amplitude average in each second, observation signal is the composite signal of ambient sea noise and Ship Radiated-Noise, utilize Nyman-Pearson criterion and given false-alarm probability, handle the mean variance that estimates the observation and statistic amount by statistical average, and then obtain observation and statistic amount confidence interval of mean, the upper limit of fiducial interval as the threshold value of judging that boats and ships occur, is just produced when exceeded threshold and reports to the police.

3, according to Detection of Weak Signals and characteristic analysis system under the complex background of claim 1, it is characterized in that of high two controls of described control tape bandpass filter by the gain control word, its band connection frequency and only be with frequency to be set to 100-800Hz and 10-10000Hz respectively is respectively applied for zero passage detection and amplitude detection.