CN102626307A - Dynamic pulse signal real-time detection system and detection method - Google Patents

Abstract

The invention relates to a dynamic pulse signal real-time detection system and a detection method. A laser photoelectric pulse sensor (1-1) converts a pulse signal into a current signal and outputs the current signal to a pulse signal conditioning circuit (1-2); an I/V (Current/Voltage) conversion circuit (2-1) converts the current signal to a voltage signal, and after being filtered by a high-pass filter (2-2) and a low-pass filter (2-3) and amplified by an amplifier (2-4), the voltage signal is outputted; an A/D (Analog/Digital) acquisition module (3-1) of a DSP (Digital Signal Processor) subsystem is used for sampling and sending data to a memory (3-1); and after the data is filtered by a harmonic notch filter (3-3) and a low-pass filter (3-4), when an audible and visual alarm (3-11) acts under the situation that the pulse frequency is abnormal, a help is turned to a specified object through a GSM module (3-13), and the relevant processing results are transmitted.

Description

Dynamic pulse signal real-time detecting system and detection method

Technical field

The present invention relates to signal detection and processing, relate to collection, pretreatment, the time and frequency domain analysis of pulse signal in particular.Can be used for the real-time detection and the processing of the dynamic pulse signal of human body.

Background technology

In recent years, the continuous change of Along with people's life style, dietary structure becomes commonly encountered diseases such as diseases such as hypertension, coronary heart disease, and sickness rate constantly rises, and particularly cardiovascular disease becomes the principal disease that threatens the human life.For fear of with alleviate of the harm of these diseases to human health; Many algorithms and system arise at the historic moment, like wearable ECG monitor, telemedicine monitoring system or the like; Their appearance has brought Gospel for the patient of cardiovascular disease, has improved the motility and the effectiveness that attend.But because the randomness and the rapidity of cardiovascular disease incidence, the application of these instruments has just had certain limitation.Traditional pulse signal check and analysis need be connected through lead monitored person's pulse signal with monitoring analysis equipment, limited monitored person's activity space, can't realize signals collecting and processing in real time under the detected person in working order.Can be in the generation of very first time perception disease, particularly important for patient's treatment, to this situation, the real-time that improves portable monitoring system just seems very important.

At present, portable pulse detection system mainly accomplishes the collection of signal and handles with simple, and like signal filtering, feature extraction scheduling algorithm; Because its amount of calculation is big; Reasons such as complex algorithm make its application on portable system that certain difficulty arranged, and can't realize real-time monitoring.

Summary of the invention

The real-time detecting system and the detection method that the purpose of this invention is to provide a kind of dynamic pulse signal.

The present invention is dynamic pulse signal real-time detecting system and detection method, and a laser photoelectricity pulse transducer 1-1 who picks up the physiology pulse signal of human body is arranged in the detection system, and laser photoelectricity pulse transducer 1-1 converts the physiology pulse signal of human body into faint current signal and exports to pulse signal modulate circuit 1-2; Pulse signal modulate circuit 1-2 is by I/V change-over circuit 2-1; High pass filter 2-2, low pass filter 2-3, amplifier 2-4 forms; I/V change-over circuit 2-1 converts the current signal of pick off output into voltage signal; Pass through high pass filter 2-2, low pass filter 2-3 filtering again, amplifier 2-4 amplifies, the voltage signal of output 0-3V; The A/D acquisition module 3-1 of DSP subsystem samples to the pulse signal that pulse signal modulate circuit 1-2 picks up, and sample frequency is 400Hz; Then data are delivered to memorizer 3-2 storage; Again the data of storage are carried out real-time filtering; Remove baseline drift and power frequency interference through integral coefficient 50Hz harmonic wave wave trap 3-3; Disturb through integral coefficient low pass filter 3-4 filtering myoelectricity, status displays 3-12 shows pulse frequency, time and date, and audible-visual annunciator 3-11 moves when pulse frequency is unusual; Cry for help to specified object through gsm module 3-13, and will send relevant result.

Dynamically the pulse signal real-time detection method the steps include:

(1) utilizes laser type photoelectricity pulse transducer, pick up the dynamic pulse signal of human body;

(2) pulse signal becomes 0-3V voltage through the amplification of pulse signal modulate circuit (1-2) and the conditioning of filtering, supplies DSP subsystem A/D conversion;

(3) the DSP subsystem is sampled to pulse signal with the sample frequency of 400Hz;

(4) in the DSP subsystem, pulse signal is carried out real-time trap and LPF, remove baseline drift, power frequency interference and myoelectricity and disturb;

(5) the real-time wave height of obtaining pulse signal and pulse frequency in the DSP subsystem, and it is carried out the FFT conversion, calculate its frequency spectrum;

(6) judge whether pulse frequency is unusual,, send sound and light alarm, and send distress signal and corresponding signal process result to specified mobile phone through the gsm module in the DSP subsystem if pulse frequency is unusual.

Usefulness of the present invention is the pulse signal of human body that not only can be real-time, can also carry out real-time monitoring to guardianship.It can not influence under monitored person's operate as normal and the active condition, the convenient pulse signal that detects monitored person, and carry out analyzing and processing.

Description of drawings

Fig. 1 is a dynamic pulse signal detection system block diagram of the present invention; Fig. 2 is the structural representation of pulse signal modulate circuit of the present invention; Fig. 3 is the structural representation of DSP subsystem of the present invention; Fig. 4 is a DSP subsystem processes module main program flow chart of the present invention, and Fig. 5 is a DSP subsystem processes module interruption subroutine flow chart.

The specific embodiment

Like Fig. 1, Fig. 2, shown in Figure 3; The present invention is dynamic pulse signal real-time detecting system and detection method, and a laser photoelectricity pulse transducer 1-1 who picks up the physiology pulse signal of human body is arranged in the detection system, and laser photoelectricity pulse transducer 1-1 converts the physiology pulse signal of human body into faint current signal and exports to pulse signal modulate circuit 1-2; Pulse signal modulate circuit 1-2 is by I/V change-over circuit 2-1; High pass filter 2-2, low pass filter 2-3, amplifier 2-4 forms; I/V change-over circuit 2-1 converts the current signal of pick off output into voltage signal; Pass through high pass filter 2-2, low pass filter 2-3 filtering again, amplifier 2-4 amplifies, the voltage signal of output 0-3V; The A/D acquisition module 3-1 of DSP subsystem samples to the pulse signal that pulse signal modulate circuit 1-2 picks up, and sample frequency is 400Hz; Then data are delivered to memorizer 3-2 storage; Again the data of storage are carried out real-time filtering; Remove baseline drift and power frequency interference through integral coefficient 50Hz harmonic wave wave trap 3-3; Disturb through integral coefficient low pass filter 3-4 filtering myoelectricity, status displays 3-12 shows pulse frequency, time and date, and audible-visual annunciator 3-11 moves when pulse frequency is unusual; Cry for help to specified object through gsm module 3-13, and will send relevant result.

Like Fig. 3, Fig. 4, shown in Figure 5, dynamically the pulse signal real-time detection method the steps include:

(1) utilizes laser type photoelectricity pulse transducer, pick up the dynamic pulse signal of human body;

(2) pulse signal becomes 0-3V voltage through the amplification of pulse signal modulate circuit (1-2) and the conditioning of filtering, supplies DSP subsystem A/D conversion;

(3) the DSP subsystem is sampled to pulse signal with the sample frequency of 400Hz;

(4) in the DSP subsystem, pulse signal is carried out real-time trap and LPF, remove baseline drift, power frequency interference and myoelectricity and disturb;

(5) the real-time wave height of obtaining pulse signal and pulse frequency in the DSP subsystem, and it is carried out the FFT conversion, calculate its frequency spectrum;

(6) judge whether pulse frequency is unusual,, send sound and light alarm, and send distress signal and corresponding signal process result to specified mobile phone through the gsm module in the DSP subsystem if pulse frequency is unusual.

Sound and light alarm module 1-4 among Fig. 1, date, clock, pulse frequency and pulse frequency state display module 1-5, the audible-visual annunciator 3-11 among gsm module 1-6 and Fig. 3, status displays 3-12, gsm module 3-13 is an identity element.

Laser photoelectricity pulse transducer 1-1 is made up of a laser diode and an integrated photosensor chip (opt101); In order to gather pulse signal easily; Consider the comfort level of guardianship simultaneously, the vein blood vessel that adopts the wrist back is as surveyed area, and pulse transducer adopts reflective structure.The built-in laser diode of pick off (650nm) is used to provide light source, and integrated photosensor chip is used for the perceived light signal.Because arterial blood changes with arteriopalmus the absorbtivity of light,, realize the fluctuation of finger tip pulse is converted into current signal output so integrated photosensor chip can be converted into the optical signal that changes the current signal of variation.

The modulate circuit of pulse signal is realized the preliminary conditioning of dynamic pulse signal, and through the I/V change-over circuit, the current signal that the photoelectricity pulse transducer is picked up converts voltage signal into earlier; Then, be 0.16Hz second order high pass analog filter through a cut-off frequency, remove baseline drift; The step low-pass analog filter that cut-off frequency is 80Hz reduces myoelectricity and disturbs; Realize the Filtering Processing of dynamic pulse signal; In filtering, through dual-stage amplifier, faint pulse signal is amplified 500 times, convert the voltage signal that can supply AD converter sampling in the DSP into.

As shown in Figure 3, through A/D acquisition module 3-1, the pulse signal that the pulse collection module is picked up to be sampled, sample rate is 400Hz; Then data are delivered to memorizer 3-2 storage; Again the data of storage are carried out real-time filtering, remove baseline drift and power frequency interference, reduce myoelectricity through integral coefficient low pass filter 3-4 and disturb through integral coefficient 50Hz harmonic wave wave trap 3-3.Simultaneously, calculate the crest 3-5 of pulse signal, the trough 3-6 of pulse signal calculates wave height 3-7 according to the Wave crest and wave trough value again; Calculate pulse frequency 3-8, calculate frequency spectrum 3-9, and real-time monitoring pulse frequency unusual 3-10 whether, then; Show pulse frequency, time, date and pulse frequency state 3-12,, sound and light alarm 3-11 will take place if pulse frequency is unusual; Simultaneously, cry for help to specified object, and will send relevant result through gsm module 3-13.

Its kernel processor chip of the detection of pulse Dynamic Signal and processing system is TMS320VC5509 APGE, is a fixed point kernel dsp chip with high processing rate (400MIPS) that TI company produces.(128K * 16bit), (maximum sample rate is 12.5KHz to FLASH on the sheet, gathers voltage range 0V-+3.3V for 4M * 16bit), 2 road 10bitA/D transducers to carry RAM.

The sample frequency that the A/D acquisition module adopts is 400Hz, according to sampling thheorem, meets the demands fully, and can keep the radio-frequency component of pulse signal.

Filter module is mainly accomplished the denoising of pulse signal.Form by integral coefficient 50Hz harmonic wave wave trap and low pass filter.Owing to pulse signal along with heartthrob produces, signal is generally fainter, the pulse signal frequency content mainly is distributed in 0～20Hz, belong to infrasonic wave, and highest frequency is no more than 40Hz.In processes such as the collection of pulse signal, amplification, detection; Many interference that come from the outside are arranged; Wherein being breathed influences the baseline drift of sliding and producing with contact skin, and the power frequency that myoelectricity interference, alternating current 50Hz and the multiple harmonic thereof that the human body potential change produces causes is disturbed particularly general.Only depend on the measure on the signal acquiring system hardware designs can not eliminate these interference fully, and these interference sometimes also can make the signal to noise ratio of pulse collection system descend, even can flood useful pulse signal.So, also need in the DSP subsystem, carry out Real-Time Filtering to pulse signal, disturb and baseline drift for the power frequency in the pulse signal, the present invention adopts 50Hz and harmonic wave wave trap thereof to eliminate.And disturb for myoelectricity, because its frequency range is extensive, can not from primary signal, basic elimination can only reduce its interference through low pass filter to pulse signal.

The present invention adopts two to have linear-phase filter that identical traffic postpones and subtract each other and obtain band elimination filter, promptly from an all-pass network, deducts a bandpass network.Bandpass network can use the comb filter of an integral coefficient to realize that all-pass network only has delay action, plays the Phase delay effect that the zona pectinata open network brings of offsetting.The transfer function and the frequency response of band elimination filter are respectively:

H _BS(z)＝H _AP(z)-H _BP(z) (1)

$H_{\mathrm{BS}}\left(z\right)=z^{-(r-k)*m/2}-{(\frac{1}{q}*\frac{1-z^{-r}}{1-z^{-k}})}^m---\left(2\right)$

$H_{\mathrm{BS}}\left(e^{\mathrm{jw}}\right)=e^{-j(r-k)*m*w/2}-{(\frac{1}{q}*\frac{\sin (\mathrm{rw}/2)}{\sin (\mathrm{kw}/2)})}^m*e^{-j(r-k)*m*w/2}---\left(3\right)$

M, r, k, q is integer,

First is all-pass network in the following formula, and back one is zona pectinata pass filter network.In formula (2) and the formula (3), (fs=400Hz is sample frequency to k=fs/fp.Fp=50Hz is stop-band frequency), can try to achieve k=8; Select suitable r and m after k value is confirmed, then can confirm the characteristic of passband, stopband and the intermediate zone of whole filter, get m=2 here; R=160 then has q=r/k=160/8=20, and q is the gain factor of comb filter.Then formula (2) is:

$H_{\mathrm{BS}}\left(z\right)=z^{-152}-{(\frac{1}{20}*\frac{1-z^{-160}}{1-z^{-8}})}^2---\left(4\right)$

Obviously, two pure time-delay is 152 before and after the formula (4), and the output equation that can be tried to achieve this band elimination filter by formula (1) is:

y _BS(n)＝y _AP(n)-y _BP(n) (5)

Y wherein _AP(n)=x (n-152) is (5-1)

$y_{\mathrm{BP}}\left(n\right)=2y_{\mathrm{BP}}(n-8)-y_{\mathrm{BP}}(n-16)+\frac{1}{400}(x\left(n\right)-2x(n-160)+x(n-320))---(5-2)$

The design of low pass filter:

Owing to after handling through notch filter, still exist comparatively serious myoelectricity to disturb in the pulse signal, and its frequency range is extensive, can not be from primary signal basic elimination can only reduce its interference to pulse signal through low pass filter.For this reason, adopt the integral coefficient low pass filter of pole zero cancellation method simplicity of design.The transfer function and the frequency response of this wave filter are following:

$H_{\mathrm{LP}}\left(z\right)=\frac{1-z^{-M}}{1-z^{-1}}=1+z^{-1}+\·\·\·+z^{-M}---\left(6\right)$

$H_{\mathrm{LP}}\left(e^{\mathrm{j\ω}}\right)=e^{\frac{-j(M-1)\mathrm{\ω}}{2}}\·\frac{\sin \left(\frac{\mathrm{\ωM}}{2}\right)}{\sin \left(\frac{\mathrm{\ω}}{2}\right)}---\left(7\right)$

Can find out that by formula (7) this low pass filter has M zero point and limit.Wherein, be evenly distributed on the unit circle M zero point, and a limit is arranged at the z=1 place, is used for offsetting first zero point, and other limit all concentrates on the initial point place, thereby has low-pass characteristic.Whole unit circle angle of circumference span is 0～2 π, and the actual samples frequency is fs, and its corresponding circular frequency is 0～2 π fs.In order to make low-pass cut-off frequencies less relatively, suppose that it is 2 π f (0)/fs that amplitude-versus-frequency curve is reduced to zero frequency for the first time, the number at all zero points is M=fs/f (0).Because the sample frequency that the DSP subsystem adopts is 400Hz, and must guarantee that M is an integer, gets f (0)=66.67Hz, obtains M=6.Then transfer function is:

$H_{\mathrm{Lp}}\left(z\right)=\frac{1-z^{-6}}{6\·(1-z^{-1})}---\left(8\right)$

The output difference equation that can be tried to achieve this low pass filter by formula (8) is:

$y_{\mathrm{LP}}(n-1)=y_{\mathrm{LP}}(n-1)+\frac{1}{6}\·(x\left(n\right)-x(n-6))---\left(9\right)$

The detection of wave height:

Wave height is meant the difference of pulse wave crest and adjacent trough amplitude, calculate wave height, at first will detect pulse wave crest and trough.Because the data buffering sector width that is provided with in the DSP subsystem is 1024 points, can only store 1024 pulse data, we only need obtain pulse wave in this 1024 point data crest (trough) gets final product.Because pulse wave crest (trough) is a maximum (minima) in its neighborhood, simultaneously, for fear of the influence of tidal wave and dicrotic wave, through a large amount of pulse waves are analyzed repeatedly, the optimum width that can obtain this neighborhood is 100 points.So; In the data buffer zone, getting 100 width is a fixed window, is made as W [j]; Then this window is fixed in the data buffer zone; Let data flow cross this window, obtain the maximum (minima) of data in this window, judge according to pulse signal crest (trough) judgment condition whether this maximum (minima) is crest (trough) again.If the pulse signal data sequence is X [i], wherein i [0,1023] can obtain crest X [i in the crest neighborhood by last surface analysis _p] judgment condition be:

$\left\{\begin{array}{c}X\left[i_p\right]=\max \left\{W\right[j\left]\right\},j\∈[k+50,k+150],k\∈\left[\mathrm{0,873}\right]\\ X\left[i_p\right]\≥X\left[i\right],i\∈[i_p-50,i_p+50]\end{array}\right.---(11-1)$

Max{W [j] wherein } be data maximum in the fixed window, i _pBe crest location, owing to be worth the restriction of width neighborhood, i most _pValue can not can not determine the value upper and lower bound of j thus greater than 1153 less than 50.The value of j is determining the position of fixed window in the data buffer zone, decides in case the j value is got, and so just can confirm in the data buffer zone that a length is 100 fixed window.And the value of k is exactly in order to limit the value of j, it to be met the demands, and k can be the arbitrary integer in its domain of definition.Similar with the crest judgment condition, can get trough X [i _t] judgment condition be:

$\left\{\begin{array}{c}X\left[i_t\right]=\min \left\{W\right[j\left]\right\},j\∈[k+50,k+150],k\∈\left[\mathrm{0,873}\right]\\ X\left[i_t\right]\≤X\left[i\right],i\∈[i_t-50,i_t+50]\end{array}\right.---(11-2)$

I wherein _tBe wave trough position.

The adjacent peaks and the trough amplitude difference of being tried to achieve by top method just can obtain the waveform height, and promptly the waveform height of one-period is that the crest amplitude in the cycle deducts the most contiguous trough amplitude in its front and obtains thus.

The calculating of pulse frequency:

Pulse frequency is meant the number of times of per minute pulse, and is consistent with heart rate under the normal condition, and normal adults pulse frequency under rest state is 60-100 time/minute.And pulse frequency has the fluctuation of certain limit under the influence that receives such as factors such as age, sex, builds.And the difference of pulse frequency, its corresponding pulse condition is also different, is reflecting physiology and pathological characters that human body is different, so in clinical diagnosis, be necessary pulse frequency is detected in real time.

The method that detects pulse frequency is: crest (trough) value that detects adjacent two pulse waves earlier.Then, obtain the number of data points (being made as n) between two pulse crests (trough), obtain this pairing time of n point, be the cycle of pulse signal, be made as T by sample rate.At last, can know by the definition of pulse frequency, with 60/T just can calculate real-time pulse frequency (be made as P time/min).The method of concrete detection pulse frequency is following:

(1) the pulse signal peak value detects and confirms peak.Detect two consecutive periods crests (trough) and position thereof with formula (11-1), formula (11-2), and obtain the n that counts between the crest (trough).

(2) confirm the cycle T of pulse signal, calculate the pulse frequency value.Sample frequency fs=400Hz, then the sampling period is 1/fs, so the cycle of pulse signal is T=n/fs.According to definition, can try to achieve pulse frequency:

P＝60/T＝(60×fs)/n。

The frequency spectrum of pulse signal calculates:

Though fairly simple directly perceived with time domain method research pulse information, can not meet the demands fully in aspect researchs such as doing pulse condition and arteries and veins gesture with it, certain limitation is arranged.Pulse signal has self unique frequency domain character, and these frequency domain characters are containing abundant pulse information, and the amplitude spectrum of obtaining pulse signal is very necessary.Signal is carried out discrete Fourier transform (DFT), and its basic transformation formula is:

$X\left(k\right)=\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}x\left(n\right)e^{\frac{-\mathrm{jkn}2\mathrm{\π}}{N}}=\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}x\left(n\right)W_N^{\mathrm{kn}}---\left(12\right)$

Wherein N is 2 power;

x (n) is an input signal sequence, and X (k) is the later output sequence of DFT conversion.Can find out that by formula (12) whole DFT computing needs 4N altogether ²Inferior real multiplications and 2N (2N-1) sub-addition, when N was very big, operand was very big.So the design adopts the butterfly computation shortcut calculation of DIT-FFT, the transformation for mula after the simplification is:

$X\left(k\right)=\underset{p=0}{\overset{\frac{N}{2}-1}{\mathrm{\Σ}}}x\left(2p\right)W_{N/2}^{\mathrm{pk}}+W_N^k\underset{p=0}{\overset{\frac{N}{2}-1}{\mathrm{\Σ}}}x(2p+1)W_{N/2}^{\mathrm{pk}}=X_e\left(k\right)+W_N^k{X}_o\left(k\right)---\left(13\right)$

Wherein, N=128, p, k are integer, and X (k) is the later output sequence of FFT conversion, and x (2p) is an input signal even number point sequence, X _e(k) be even number point sequence FFT, x (2p+1) is an input signal odd point sequence, X _o(k) be the odd point sequence FFT.In the formula (13), N/2 the even number point that every summation is respectively an original series and the DFT of N/2 odd point.Compare DFT, amount of calculation reduces greatly.Can use formula (13) 128 pulse data in the DSP subsystem data buffer zone are carried out the FFT conversion, obtain its spectrogram.

Fig. 4, Fig. 5 are the program circuit of DSP subsystem processes module of the present invention.The intervalometer that adopts dsp processor to carry is controlled the A/D converter image data through the mode of regularly interrupting, and data of every collection are just accomplished a date processing in the timing interruption subroutine.Fig. 4 is a main program flow chart.Mastery routine is accomplished system initialization, is calculated frequency spectrum, the video data result of pulse signal.Fig. 5 is an A/D conversion timing interruption subroutine flow chart.Take timer interrupt mode control A/D conversion, sample rate is 400Hz, each regularly end, and intervalometer 0 just can interrupt to system's application, and after system responses was interrupted, interruption subroutine brought into operation.Interruption subroutine finish the work for: start A/D converter and begin translation data, read and store data, the pulse data of data relief area is carried out trap, low pass and envelope filtering; Detect the peak value and the valley of pulse signal, calculate wave height and pulse frequency, and judge whether pulse frequency is unusual; If pulse frequency is unusual; Send sound and light alarm, and cry for help to appointed object, send relevant result through gsm module.

Claims (6)

1. dynamic pulse signal real-time detecting system; A laser photoelectricity pulse transducer (1-1) that picks up the physiology pulse signal of human body is arranged; It is characterized in that laser photoelectricity pulse transducer (1-1) converts the physiology pulse signal of human body into faint current signal and exports to pulse signal modulate circuit (1-2), pulse signal modulate circuit (1-2) is by I/V change-over circuit (2-1), high pass filter (2-2); Low pass filter (2-3); Amplifier (2-4) is formed, and I/V change-over circuit (2-1) converts the current signal of pick off output into voltage signal, passes through high pass filter (2-2), low pass filter (2-3) filtering again; Amplifier (2-4) amplifies, the voltage signal of output 0-3V; The A/D acquisition module (3-1) of DSP subsystem is sampled to the pulse signal that pulse signal modulate circuit (1-2) picks up, and sample frequency is 400Hz; Then data are delivered to memorizer (3-2) storage; Again the data of storage are carried out real-time filtering; Remove baseline drift and power frequency interference through integral coefficient 50Hz harmonic wave wave trap (3-3); Disturb through integral coefficient low pass filter (3-4) filtering myoelectricity, status displays (3-12) shows pulse frequency, time and date, and audible-visual annunciator (3-11) moves when pulse frequency is unusual; Cry for help to specified object through gsm module (3-13), and will send relevant result.

2. dynamic pulse signal real-time detection method the steps include:

(1) utilizes laser type photoelectricity pulse transducer, pick up the dynamic pulse signal of human body;

(2) pulse signal becomes 0-3V voltage through the amplification of pulse signal modulate circuit (1-2) and the conditioning of filtering, supplies DSP subsystem A/D conversion;

(3) the DSP subsystem is sampled to pulse signal with the sample frequency of 400Hz;

(4) in the DSP subsystem, pulse signal is carried out real-time trap and LPF, remove baseline drift, power frequency interference and myoelectricity and disturb;

(5) the real-time wave height of obtaining pulse signal and pulse frequency in the DSP subsystem, and it is carried out the FFT conversion, calculate its frequency spectrum;

(6) judge whether pulse frequency is unusual,, send sound and light alarm, and send distress signal and corresponding signal process result to specified mobile phone through the gsm module in the DSP subsystem if pulse frequency is unusual.

3. dynamic pulse signal real-time detection method according to claim 2 is characterized in that adopting two to have linear-phase filter that identical traffic postpones and subtract each other and obtain band elimination filter, promptly from an all-pass network, deducts a bandpass network.

4. dynamic pulse signal real-time detection method according to claim 2 is characterized in that the pulse signal data sequence is X [i], i [0,1023] wherein, crest X [i in the crest neighborhood _p] judgment condition be:

$\left\{\begin{array}{c}X\left[i_p\right]=\max \left\{W\right[j\left]\right\},j\∈[k+50,k+150],k\∈\left[\mathrm{0,873}\right]\\ X\left[i_p\right]\≥X\left[i\right],i\∈[i_p-50,i_p+50]\end{array}\right.$

Max{W [j] wherein } be data maximum in the fixed window, i _pBe crest location, 0≤i _p≤1023, k is the arbitrary integer in [0,873], mainly confirms the position of detection window, and j is the arbitrary integer in the window.

Trough X [i _t] judgment condition be:

$\left\{\begin{array}{c}X\left[i_t\right]=\min \left\{W\right[j\left]\right\},j\∈[k+50,k+150],k\∈\left[\mathrm{0,873}\right]\\ X\left[i_t\right]\≤X\left[i\right],i\∈[i_t-50,i_t+50]\end{array}\right.$

I wherein _tBe the valley position.

The waveform height of one-period interior crest amplitude of cycle for this reason deducts the most contiguous trough amplitude in its front.

5. dynamic pulse signal real-time detection method according to claim 2 is characterized in that pulse frequency confirms that step is:

(1) the pulse signal peak value detects and confirms peak: with the described crest X of claim 3 [i _p] judgment condition, trough X [i _t] judgment condition detect two consecutive periods crests or trough and position thereof, and obtain the n that counts between crest or the trough;

(2) confirm the cycle T of pulse signal, calculate pulse frequency value: sample frequency fs=400Hz, the sampling period is 1/fs, and the cycle of pulse signal is T=n/fs, pulse frequency P=60/T=(60 * fs)/n.

6. dynamic pulse signal real-time detection method according to claim 2 is characterized in that adopting following formula that 128 pulse data in the data buffer zone of DSP subsystem are carried out the FFT conversion, obtains its spectrogram:

$X\left(k\right)=\underset{p=0}{\overset{\frac{N}{2}-1}{\mathrm{\Σ}}}x\left(2p\right)W_{N/2}^{\mathrm{pk}}+W_N^k\underset{p=0}{\overset{\frac{N}{2}-1}{\mathrm{\Σ}}}x(2p+1)W_{N/2}^{\mathrm{pk}}=X_e\left(k\right)+W_N^k{X}_o\left(k\right)$

Wherein, N=128, p, k are integer, and X (k) is the later output sequence of FFT conversion, and x (2p) is an input signal even number point sequence, X _e(k) be even number point sequence FFT, x (2p+1) is an input signal odd point sequence, X _o(k) be the odd point sequence FFT.

Images