CN112155558B

CN112155558B - Fetal movement signal acquisition and analysis device

Info

Publication number: CN112155558B
Application number: CN202011032075.XA
Authority: CN
Inventors: 贾朋飞; 杨洪波; 张莹莹; 吕甜甜; 蔡黎明; 刘永峰; 郭凯
Original assignee: Yongkang Guoke Rehabilitation Engineering Technology Co ltd
Current assignee: Suzhou Guoke Medical Technology Development Group Co ltd
Priority date: 2020-09-27
Filing date: 2020-09-27
Publication date: 2021-08-06
Anticipated expiration: 2040-09-27
Also published as: CN112155558A

Abstract

The invention discloses a fetal movement signal acquisition and analysis device, which comprises: the signal acquisition module and the signal analysis module are arranged on the abdominal wall of the pregnant woman in a wearable mode; the signal acquisition module comprises N sensors which are arranged on the abdominal wall of the pregnant woman at intervals and used for acquiring vibration signals of the abdominal wall of the pregnant woman so as to acquire N channel data in parallel; the signal analysis module comprises a preprocessing unit, a classification unit and a statistic unit, wherein the preprocessing unit is used for preprocessing the data of the vibration signals acquired by the N sensors in the received signal acquisition module and then transmitting the preprocessed vibration signals to the classification unit. The invention can monitor a plurality of fetal movement evaluation indexes including fetal movement forms (such as knocking, duration, strength and the like), the longest duration time of fetal movement, the ratio of strong fetal movement in the longest duration time, the longest static time of a fetus and the fetal activity, and the indexes quantitatively describe the sensing conditions of the pregnant woman on fetal movement strength, characteristics and duration for the first time, thereby having important medical reference significance.

Description

Fetal movement signal acquisition and analysis device

Technical Field

The invention relates to the technical field of fetal movement detection, in particular to a fetal movement signal acquisition and analysis device.

Background

Fetal movement is the most intuitive feeling of the pregnant woman on the intrauterine health condition of the fetus, and when the fetus suffers intrauterine injury, hypoevolutism or even death, the pregnant woman can be aware of the abnormal change of fetal movement. Guidelines related to fetal activity self-monitoring behaviors mostly confirm the importance of fetal activity self-monitoring, and Chinese 'health guidelines for pre-pregnancy and pregnancy (2018)' recommends pregnant women to start to perform fetal activities for 29-30 weeks¹Most of the foreign guidelinesThe pregnant woman is recommended to start daily monitoring of fetal movements 28 weeks later². Fetal reduction is often the first sign of fetal premature death³2018 Australian/New Zealand clinical practice guidelines indicate that pregnant women should pay more attention to not only fetal movement times, but also fetal movement intensity, fetal movement characteristics and duration⁴. Generally speaking, fetal movement is longer than 5min for the longest duration, and the percentage of strong fetal movement is more than 80% in the longest duration, which means that the fetus is always active during this period and the movement intensity is too high. When fetal movement is abnormally active and strong and the duration is too long, pregnant women and medical care personnel need to take care to prevent the phenomena of umbilical cord wrap, insufficient oxygen supply of fetus and the like. A case-control study in new zealand included 155 cases of stillbirth and 310 cases of live births, and the study showed that sudden fetal movements that became stronger in a short time were associated with an increased risk of stillbirth, which was nearly 7-fold. Sudden frequent and intense fetal movements, particularly the ensuing diminished or diminished fetal movements, are indicative of acute fetal distress, e.g., intense fetal movements failing to relieve, being alerted to further cause fetal death⁶. The longest fetal static time exceeds 50min, and the fetal activity is lower than 10%, which means that the fetal movement frequency is less, and the pregnant woman and the medical care personnel need to pay high attention to the frequent occurrence of the conditions. The frequent fetal movement reduction is closely related to the premature delivery and even the dead fetus phenomenon of the fetus, most dead fetus is accompanied by the frequent fetal movement reduction phenomenon of 3-4 days, and 55 percent of pregnant women experiencing the dead fetus can obviously perceive the frequent fetal movement reduction⁷. Fetal movement reduction is thought to be associated with poor outcomes such as restricted fetal growth, dead fetus, etc., with a 4-fold increase in the risk of a fetal movement reducer developing a dead fetus⁸It is also closely related to the adverse pregnancy outcome such as infection, neurodevelopmental abnormality, maternal-fetal blood transfusion, placental insufficiency, umbilical cord complications and emergency labor, premature delivery, etc. The possibility of poor pregnancy outcome such as brain injury of perinatal infants and neonatal ischemic-hypoxic encephalopathy is increased^9；10. Comprehensively considering the fetal movement times, the longest fetal static time and the fetal activity index, when the fetal movement is less frequently, no matter how the ultrasonic evaluation result is, the pregnant woman has high risk of placental dysfunction⁷Further medicine is neededAnd (6) diagnosis.

Therefore, besides the fetal movement times, the fetal movement monitoring system has important medical reference significance for monitoring more forms of fetal movement indexes (such as the longest fetal static time, fetal activity and the like), but the existing products generally stay at the level of monitoring the fetal movement times only and are difficult to comprehensively reflect the fetal state.

Cited documents:

[1] the Chinese medical society, gynaecological and obstetrical science group, health guide for pre-pregnancy and pregnancy (2018) [ J ] Chinese J. China J. obstetrics and gynecology, 2018,53(1):7-13.

[2] Wen, Zhang static pregnant woman fetal movement self-monitoring behavior research status and guidelines suggest [ J ]. Chinese maternal and child health research, 2017,28(2): 213-.

[3]Jf F.A kick from within-fetal movement counting and the cancelled progress in antenatal care[J].J Perinat Med,2004,32(1):13-24.

[4]Daly L M,Gardener G,Bowring V,et al.Care of pregnant women with decreased fetal movements:Update of a clinical practice guideline for Australia and New Zealand[J].Australian and New Zealand Journal of Obstetrics and Gynaecology,2018,58(4):463-468.

[5]Ryo E,Nishihara K,Matsumoto S,et al.A new method for long-term home monitoring of fetal movement by pregnant women themselves[J].Med Eng Phys,2012,34(5):566-72.

[6]T S,Jmd T,Ea M,et al.Maternal perception of fetal activity and late stillbirth risk:findings from the auckland stillbirth study[J].Birth,2011,38(4):311-316.

[7]Scala C,Bhide A,Familiari A,et al.Number of episodes of reduced fetal movement at term:association with adverse perinatal outcome[J].Am J Obstet Gynecol,2015,213(5):678e1-6.

[8] Wangmeng, Chen Qian, etiology and prevention of stillbirth [ J ] China journal of practical gynecology and obstetrics, 2017,33(11):1121-1125.

[9]M T,H F,S S,et al.Brain damage caused by severe fetomaternal hemorrhage[J].Pediatrics International,2010,52(2):301-304.

[10] Heiming swallow, fetal movement reduction in late pregnancy and neonatal hypoxic ischemic brain injury [ J ]. Chinese J.J.pediatrics, 2015,30(4): 311-.

Disclosure of Invention

The technical problem to be solved by the present invention is to provide a fetal movement signal collecting and analyzing device, aiming at the above-mentioned deficiencies in the prior art.

In order to solve the technical problems, the invention adopts the technical scheme that: a fetal movement signal acquisition and analysis apparatus comprising: the device comprises a signal acquisition module which is arranged on the abdominal wall of a pregnant woman in a wearable manner and a signal analysis module which is in communication connection with the signal acquisition module;

the signal acquisition module comprises N sensors which are arranged on the abdominal wall of the pregnant woman at intervals and used for acquiring vibration signals of the abdominal wall of the pregnant woman so as to acquire N channels of data in parallel;

the signal analysis module comprises a preprocessing unit, a classification unit and a statistic unit, wherein the preprocessing unit is used for preprocessing the received vibration signals acquired by the N sensors in the signal acquisition module and then transmitting the preprocessed vibration signals to the classification unit;

the classification unit is used for classifying the forms of the vibration signals and comprises the following steps:

1) firstly, carrying out data cutting segmentation on data of N channels received in unit monitoring time according to a time line to form a plurality of independent judging units, and then carrying out analysis judgment on the judging units one by one according to the time line through the following steps;

2) the method comprises the steps of firstly distinguishing and judging fetal movement signals and noise signals of vibration signals through multi-channel characteristic analysis, then distinguishing and judging specific forms of the fetal movement signals and the noise signals to form classification results, and then carrying out analysis and judgment of a next judgment unit;

3) repeating the step 2) until the analysis and judgment of all judgment units in unit monitoring time are completed;

wherein the specific forms of the fetal movement signals at least comprise knock-type fetal movements and strong-type fetal movements, and the specific forms of the noise signals at least comprise breathing noise, sneezing noise and body movement noise;

the statistical unit forms a multi-dimensional fetal movement index in unit monitoring time according to the classification result of the classification unit, wherein the multi-dimensional fetal movement index at least comprises fetal movement times, fetal movement forms, maximal duration of fetal movement, strong fetal movement ratio in the maximal duration, maximal static time of a fetus and fetal activity in the unit monitoring time;

wherein the maximum duration of fetal movement is: the total time occupied by the interval with the longest duration in the judging unit interval continuously judged as the fetal movement signal in the unit monitoring time;

the percentage of strong fetal movement within the longest duration is: the percentage of the interval in which fetal movement is judged to be in a strong form in the longest duration interval of fetal movement in unit monitoring time;

the longest static time of the fetus is: the total time occupied by the interval with the longest duration in the judging unit interval continuously judged as the noise signal in the unit monitoring time;

the fetal liveness is as follows: the determination units determined as fetal movement signals per unit of monitoring time account for a percentage of all determination units per unit of monitoring time.

Preferably, the step of preprocessing the data by the preprocessing unit includes: firstly, smoothing is carried out on a vibration signal by adopting a method at least comprising wavelet threshold denoising and Butterworth filtering, and then baseline drift correction is carried out by adopting an asymmetric least square baseline correction method.

Preferably, the step 2) specifically includes:

2-1) carrying out multi-channel feature extraction on a single judgment unit, comprising the following steps:

2-1-1) extracting the Pearson correlation coefficients among the N channels, and then calculating the mean value of the Pearson correlation coefficients of each channel relative to other channels, and recording the mean value as the Pearson correlation coefficient mean value rho_n；

2-1-2) calculate the peak-to-peak quotient for each of the N channels: setting a minimum horizontal unit distance between peak and peak values according to the sampling frequencyAt this distance, the ratio of the maximum peak to the minimum peak of each channel was calculated and recorded as peak division_N；

2-1-3) carrying out spectrum analysis and normalization processing on the N channel signals, and calculating the average value amp of the amplitudes corresponding to different frequencies of the N channel signals_nStatistics of amp_nThe number of channels above the amplitude threshold, denoted count_{(amp＞thresholdAmp)}；

2-1-4) matrix signature analysis: in a single judging unit, forming vibration signal data of N channel signals into a matrix of M x N, wherein M represents the product of sampling frequency f and the time length t of the judging unit, namely M is f x t; for this matrix, mean value of row vectors is represented by meanRowValue, index represents index of row vectors, and the following features are taken:

a) difference between maximum and minimum of row vector mean

r_max-min，r_max-min＝meanRowValue_max-meanRowValue_min，

Wherein, meanRowValue_maxMeans mean maximum value of row vector, meanRowValue_minRepresenting the minimum value of the mean value of the row vectors;

b) index difference index between maximum value of row vector mean and minimum value of row vector mean_max-min，index_max-min＝index_max-index_minTherein, index_maxIndex, representing the maximum value of the mean of the row vectors_minAn index representing the minimum of the row vector means;

2-2) carrying out specific classification on the vibration signal of the judging unit according to the multi-channel characteristics obtained in the step 2-1), wherein the specific classification comprises the following steps:

2-2-1) setting 2 Pearson correlation coefficient judgment threshold values A and B, wherein A is more than 0 and less than B, and averaging the Pearson correlation coefficient mean values rho of all channels_nMinimum value of (1)_minComparing the comparison result with the A and the B, and judging according to the comparison result according to the following different steps;

according to statistical experience, the field grades of the Pearson correlation coefficient values are divided as follows: 0.8-1.0, very strong correlation; 0.6-0.8, strongly correlated; 0.4-0.6, moderately relevant; 0.2-0.4, weakly correlated; 0.0-0.2, very weakly or not; < 0, negative correlation. A, B two threshold values, the number N of channels and the statistical experience need to be considered comprehensively, and the determination and the improvement are carried out;

2-2-2)ρ_minwhen the peak value is greater than B, counting the peak value quotient peak division N of each channel in the N channels;

I. if at least one channel exists, the peak-to-peak quotient is larger than a set threshold value P_TJudging the fetal movement in a knocking mode and outputting a classification result, otherwise, continuously judging according to the following steps, wherein P is more than or equal to 2.0_TLess than or equal to 2.5; according to the experimental results, the threshold value P_TThe method needs to be determined and perfected according to the sensing characteristics (sensitivity, linearity and the like) of a sensing device; in general, 2.0. ltoreq.P_T≤2.5；

II. Statistics count_{(amp＞thresholdAmp}) Number of dominant frequencies count_{(amp＞thresholdAmp)}When the number is J, judging the breathing noise, and outputting a classification result, wherein J is a positive integer; otherwise, continuing to judge according to the following steps;

III, setting a matrix characteristic analysis threshold value if r_max-min> threshold value and index_max-minJudging sneezing noise when the value is less than I f, otherwise judging body movement noise; outputting a classification result, wherein I is a proportionality coefficient, and is more than or equal to 1 and less than or equal to 2.5;

according to the experimental result, the threshold value needs to be determined and perfected according to the sensing characteristics (response curve, linearity and the like) of the sensing device, and the proportionality coefficient I needs to be determined and perfected according to the sample performance, wherein in general, I is more than or equal to 1 and less than or equal to 2.5;

2-2-3)ρ_minif the number of the Pearson correlation coefficients smaller than A in the Pearson correlation coefficients of all the channels is not more than k, judging the channel to be in knocking type fetal movement, otherwise, judging the channel to be in strong type fetal movement, and outputting a classification result; wherein k is greater than N/2;

according to the experimental result, the value of k needs to be determined and perfected according to the number of channels N and the sample performance, generally, k is more than N/2;

2-2-4)A＜ρ_minwhen < B, counting N againThe peak-to-peak quotient of each channel in the channels is peakDivisionN, if the peak-to-peak quotient of at least one channel is larger than a set threshold value P_TIf so, judging according to the following step a), otherwise, judging according to the following step b);

a) if the number of the Pearson correlation coefficients smaller than A in the Pearson correlation coefficients of all the channels is not more than k, judging the channel to be in knocking type fetal movement, otherwise, judging the channel to be in strong type fetal movement, and outputting a classification result; wherein k is greater than N/2;

b) setting a matrix characteristic analysis threshold value if r_max-min> threshold value and index_max-minJudging sneezing noise when the value is less than I f, otherwise judging body movement noise; and outputting a classification result, wherein I is a proportionality coefficient, and I is more than or equal to 1 and less than or equal to 2.5.

Preferably, the signal acquisition module comprises 4 sensors which are arranged on the abdominal wall of the pregnant woman at uniform intervals along the transverse direction and are used for acquiring vibration signals of the abdominal wall of the pregnant woman, so that 4 channels of data are acquired in parallel.

Preferably, the step 2) specifically includes:

2-1-1) extracting the Pearson correlation coefficient among the 4 channels, wherein the Pearson correlation coefficient calculation formula is as follows:

ρ_X，Y＝cov(X，Y)/σ_Xσ_Y＝E((X-μ_X)(Y-μY))/σ_Xσ_Y；

where cov (X, Y) represents the covariance, σ, between two variables X, Y_X，σ_YRepresents the standard deviation of two variables X, Y; the vibration signal data after being preprocessed by the four channels are respectively represented by s0, s1, s2 and s3, so that 6 Pearson correlation coefficients, namely rho, are shared among the four channels_0，1，ρ_0，2，ρ_0，3，ρ_1，2，ρ_1，3，ρ_2，3Calculating the mean value of the pearson correlation coefficient of each channel for the other channels, namely:

ρ₀＝1/3(ρ_0，1+ρ_0，2+ρ_0，3)；

ρ₁＝1/3(ρ_0，1+ρ_1，2+ρ_1，3)；

ρ₂＝1/3(ρ_0，2+ρ_1，2+ρ_2，3)；

ρ₃＝1/3(ρ₀₃+ρ_1，3+ρ_2，3)；

2-1-2) calculate the peak-to-peak quotient for each of the 4 channels: according to the sampling frequency, the minimum horizontal unit distance between the peak values is set, and under the distance, the ratio of the maximum peak value to the minimum peak value in each channel is calculated and is recorded as peak division₀、peakDivision₁、peakDivision₂、peakDivision₃；

2-1-3) carrying out spectrum analysis and normalization processing on the four-channel signals, and calculating the average value amp of the amplitudes corresponding to different frequencies of the 4-channel signals_nStatistics of amp_nThe number of channels above the amplitude threshold, denoted count_{(amp＞thresholdAmp)}；

2-1-4) matrix signature analysis: in a single judging unit, vibration signals of 4 channel signals form an M & ltx & gt 4 matrix, wherein M represents the product of sampling frequency f and the time duration t of the judging unit, namely M & ltx & gt; for this matrix, mean value of row vectors is represented by meanRowValue, index represents index of row vectors, and the following features are taken:

a) difference r between maximum and minimum of row vector mean_max-min，

r_max-min＝meanRowValue_max-meanRowValue_min，

2-2) carrying out specific classification on the vibration signal of the judging unit according to the multi-channel characteristics obtained in the step 3-1), wherein the specific classification comprises the following steps:

2-2-1) setting 2 pearson correlation coefficient decision thresholds a and B, where a is 0.4 and B is 0.6; the mean value of the Pearson's correlation coefficients of all channels is calculated_nMinimum value of (1)_minComparing the comparison result with the A and the B, and judging according to the comparison result according to the following different steps;

2-2-2)(ρ₀、ρ₁、ρ₂、ρ₃)_minwhen the peak value is more than 0.6, counting the peak-to-peak quotient peak division N of each channel in the 4 channels;

I. if at least one channel exists, the peak-to-peak quotient is larger than a set threshold value P_TWhen is, P_TJudging whether the tire is in a knocking form or not, if so, judging whether the tire is in a knocking form or not, and outputting a classification result, otherwise, continuing to judge according to the following steps;

II. Statistics count_{(amp＞thresholdAmp)}If count_{(amp＞thresholdAmp)}When J is equal to 2, judging the breathing noise and outputting a classification result; otherwise, continuing to judge according to the following steps;

III, setting a matrix characteristic analysis threshold value if r_max-min> threshold value and index_max-minJudging sneezing noise when the value is less than I f, otherwise judging body movement noise; outputting a classification result, wherein I is 1.5;

2-2-3)ρ_minif the number of the Pearson correlation coefficients smaller than A in the Pearson correlation coefficients of all the channels is not more than k, judging the channel to be in knocking type fetal movement, otherwise, judging the channel to be in strong type fetal movement, and outputting a classification result; wherein k is 3;

2-2-4)A＜ρ_minwhen the peak-to-peak quotient is less than B, counting the peak-to-peak quotient peak division N of each channel in the N channels, if the peak-to-peak quotient of at least one channel is larger than a set threshold value 2, judging according to the following step a), otherwise, judging according to the following step B);

a) if the number of the Pearson correlation coefficients smaller than A in the Pearson correlation coefficients of all the channels is not more than k, judging the channel to be in knocking type fetal movement, otherwise, judging the channel to be in strong type fetal movement, and outputting a classification result; wherein k is 3;

b) setting a matrix characteristic analysis threshold value if r_max-min> threshold value and index_max-minJudging sneezing noise when the value is less than I f, otherwise judging body movement noise; and outputting a classification result, wherein I is 1.5.

Preferably, the sensor comprises a packaging structure and a sensor body arranged in the packaging structure, and the packaging structure comprises a bottom plate, a cover plate buckled on the bottom plate, and a pressing piece arranged between the bottom plate and the cover plate.

Preferably, the middle part of the bottom plate is provided with a mounting cavity capable of accommodating the sensor body through a plurality of circumferentially arranged limiting barrier strips, and the middle part of the mounting cavity is provided with a first pressing boss;

the inner side surface of the pressing piece is provided with a second pressing boss matched with the first pressing boss, so that the sensor body is pressed between the first pressing boss and the second pressing boss.

Preferably, an annular protrusion is arranged on the bottom plate and surrounds the outer sides of the limit barrier strips, a first annular edge is formed by extending the periphery of the pressing piece outwards, the lower surface of the first annular edge is used for being arranged on the annular protrusion in a matched manner, and a first annular mounting table surface is formed on the upper surface of the first annular edge;

the outside of compressing tightly the piece outwards the arch form be used for with the arc contact surface of pregnant woman's stomach wall contact, the periphery of arc contact surface is formed with second annular installation mesa.

Preferably, the outer periphery of the cover plate extends outwards to form a second annular edge for being in fit contact with the bottom plate outside the annular protrusion, and the cover plate has an annular step platform inside, so that a first annular pressing platform and a second annular pressing platform are formed inside the cover plate, the first annular pressing platform is used for being in fit contact with the first annular mounting platform, and the second annular pressing platform is used for being in fit contact with the second annular mounting platform.

Preferably, the pressing member has a gap in a radial direction with the cover plate, so that the pressing member is in a relaxed state in the radial direction thereof.

The invention has the beneficial effects that:

1. the pregnant woman abdominal wall pressure signal acquisition system adopts a design mode of a multi-channel pressure sensor to acquire pregnant woman abdominal wall pressure signals, the number of channels is easy to expand, the extracted pressure signals are complete and comprehensive, various signal characteristics are convenient to distinguish and recognize, and the accuracy and the reliability are good;

2. according to the invention, through a fusion comparison analysis algorithm based on the characteristics of the multi-channel pressure signals, the obvious distinguishing and classification of different fetal movement forms such as knocking, violence and persistence and noises such as body movement, sneezing and laughing can be realized, and the method of the invention is more comprehensive and complete in the judgment of the fetal movement form and the identification of the noises in the process of fetal movement monitoring;

3. besides the fetal movement times, the fetal movement monitoring system can also monitor a plurality of fetal movement evaluation indexes including the fetal movement form (such as knocking, duration, strength and the like), the longest duration time of fetal movement, the strong fetal movement proportion in the longest duration time, the longest static time of a fetus and the fetal activity, and the indexes quantitatively describe the sensing conditions of the pregnant woman on fetal movement strength, characteristics and duration for the first time, and have important medical reference significance;

4. the sensor packaging structure provided by the invention can well collect weak physiological fetal movement signals of the abdominal wall of a pregnant woman.

Drawings

FIG. 1 is a schematic diagram of the 4-channel pressure sensor signal acquisition of the present invention;

FIG. 2 is a waveform diagram illustrating several representative signals in accordance with the present invention;

FIG. 3 is a flow chart of the fetal movement signal collecting and analyzing device of the present invention for signal collection and analysis;

FIG. 4 is a graph showing the effect of pre-processing on a respiratory signal in a resting state according to embodiment 1 of the present invention;

FIG. 5 is a diagram showing the effect of signal preprocessing in the moving tire triggering state in embodiment 1 of the present invention;

fig. 6 is a matrix chart of pressure data of four-channel signals in embodiment 1 of the present invention;

FIG. 7 is a flowchart for specifically classifying fetal movement signals and noise signals in embodiment 1 of the present invention;

fig. 8 is an external structural view of a package structure in embodiment 2 of the present invention;

fig. 9 is an exploded view of a package structure in embodiment 2 of the present invention;

fig. 10 is a schematic structural view of a base plate in embodiment 2 of the present invention;

fig. 11 is a schematic structural view of a pressing member in embodiment 2 of the present invention;

fig. 12 is a schematic structural view of a cover plate in embodiment 2 of the present invention;

fig. 13 is a schematic view of the installation of a sensor body in a package structure in embodiment 2 of the present invention;

description of reference numerals:

1-packaging structure; 2-a sensor body; 3-a bottom plate; 4, a pressing piece; 5, covering a plate; 6-clearance; 30-limiting barrier strips; 31-a mounting cavity; 32-a first compression boss; 33-an annular projection; 34-pin lead-out hole;

40-a second compression boss; 41 — a first annular rim; 42-arc contact surface; 43 — a first annular mounting table; 44-a second annular mounting table;

50-a second annular rim; 51-a first annular compression table; 52-second annular clamping table.

Detailed Description

The present invention is further described in detail below with reference to examples so that those skilled in the art can practice the invention with reference to the description.

It will be understood that terms such as "having," "including," and "comprising," as used herein, do not preclude the presence or addition of one or more other elements or groups thereof.

The fetal movement signal acquisition and analysis device of the embodiment comprises: the device comprises a signal acquisition module which is arranged on the abdominal wall of a pregnant woman in a wearable manner and a signal analysis module which is in communication connection with the signal acquisition module;

the signal acquisition module comprises N sensors which are arranged on the abdominal wall of the pregnant woman at intervals and used for acquiring vibration signals of the abdominal wall of the pregnant woman so as to acquire N channel data in parallel.

The parallel multi-channel pressure sensor is adopted to collect the pressure distribution condition of the abdominal wall of the pregnant woman, so that the development and the maintenance are facilitated. The number of the sensors can be flexibly configured according to the sensing characteristics (such as load area, sensitivity and the like) of the device, and the sensors only need to reflect the pressure trend change in the range of the whole abdominal wall. The phenomenon of signal capturing missing exists when the number of the sensors is too small, and signal acquisition redundancy, large data volume and cost increase are caused when the number of the sensors is too large. In a preferred embodiment, the number of the sensors is at least not lower than 3, and the sensors can cover the left, middle and right 3 areas of the abdominal wall of the pregnant woman. In the following embodiments, a 4-channel pressure sensor is used for illustration (as shown in fig. 1), and S0, S1, S2 and S3 respectively represent 4-channel pressure sensors, which are transversely and uniformly distributed on the abdominal wall of a pregnant woman, so that the distribution mode can reduce the packaging area of the fetal activity monitoring belt to the greatest extent, reduce the skin area in direct contact with the pregnant woman, and can obviously improve the wearing comfort. Under the distribution mode, fetal movement signals acquired from longitudinally far positions of the sensors are slightly weakened, and the problem can be well solved through a data preprocessing means. When no fetal movement exists, S0, S1, S2 and S3 acquire normal abdominal wall breathing signals of the pregnant woman; when the fetal movement is in the form of 'knocking', the triggered fetal movement is located at f0, the fetal movement wave signal can be collected by the sensors S0 and S1, and the signal strength received by the sensors S2 and S3 is very weak; when the fetal movement is in a strong form, a plurality of areas of the abdominal wall of the pregnant woman, such as the areas f1_ a and f1_ b, can be triggered, the fetal movement signals can be acquired by the sensors S0 and S3, and the signal strength received by the sensors S1 and S2 is very weak.

wherein the specific form of the fetal movement signal comprises at least a tapping form of fetal movement and a strong form of fetal movement, and the specific form of the noise signal comprises at least breathing noise, sneezing noise and body movement noise.

In the invention, a batch of multi-channel pressure sensing data is cut and segmented to form a single independent judgment unit, the time span of the single judgment unit needs to be determined by comprehensively considering the fetal movement triggering time and the system sampling frequency, generally speaking, the time span of the single judgment unit can be 2-15 seconds, if the judgment unit in the time span has fetal movement characteristics, the judgment unit is judged to be true, namely a fetal movement signal, and if not, the judgment unit is judged to be false, namely a noise signal. Taking 4-channel sensing data as an example, cutting and segmenting the 4-channel pressure sensing data into a plurality of sequentially connected judging units according to the acquisition time sequence, wherein each judging unit comprises data acquired by 4 channels in the same time segment.

In one embodiment, taking 4-channel sensing data as an example, waveforms of several representative signals are as follows (see fig. 2), and 4 curves in the figure respectively represent pressure data collected by 4 sensing channels. FIG. 2a shows data collected by the 4-channel sensor when fetal movement does not occur, wherein the data of the 4 channels is abdominal wall pressure signals generated by normal respiration of a pregnant woman, and each channel has obvious respiration signal characteristics; FIG. 2b is a waveform of data generated by a fetal movement in the form of a "tap" where the curved path indicated by the arrow has a distinct pressure pulse signature; FIG. 2c is a data waveform generated by fetal movement in the form of "strong" fetal movement that triggers multiple abdominal wall regions, where both curved paths indicated by arrows monitor a significant pressure jump; FIG. 2d shows the data waveform generated by "sneezing" noise, with 4-channel data showing a significant pressure rise, followed by a rapid drop to about 0, and then back to the normal pressure data range with fewer regular peaks and troughs; fig. 2e shows the data waveform generated by the "body movement" noise, and the pressure mean value of 4 channels will obviously rise at the same time and slowly return to the normal value, and there are no regular peaks and troughs. The algorithm provided in some embodiments of the present invention may perform feature fusion on the multi-channel sensing data of each decision unit, and analyze and identify whether a fetal movement signal (represented by two labels of "true" and "false") exists in this single decision unit by combining shallow features (including but not limited to time domain features, frequency domain features, wavelet series, etc.) with similarity algorithms (including but not limited to pearson correlation coefficient, Hausdorff distance, Frechet distance) and AI algorithms (including but not limited to KNN, logistic regression, SVM, etc.), and further determine what form of fetal movement and what form of noise belongs.

The statistical unit forms a multi-dimensional fetal movement index in unit monitoring time according to the classification result of the classification unit, wherein the multi-dimensional fetal movement index at least comprises the fetal movement times, the fetal movement form (knocking form fetal movement or strong form fetal movement), the longest duration time of fetal movement, the strong fetal movement ratio in the longest duration time, the longest static time of fetus and the fetal activity in unit monitoring time.

Specifically, wherein:

the maximum duration of fetal movement is as follows: and the total time occupied by the section with the longest duration in the judging unit sections which are continuously judged as the fetal movement signals in the unit monitoring time. Such as: when the measured data is 1 hour in total and a single determination unit is 6s, and a maximum of 10 determination units are continuously determined as "true" in 1 hour, the maximum duration of fetal movement is 60s (6s × 10).

The percentage of strong fetal movement within the longest duration is: percentage of the interval in which fetal movement is judged to be in a strong form in the longest duration interval of fetal movement per unit monitoring time. Such as: the maximum duration of fetal movement was 60s, and of the 10 determination units that were continuously "true", the percentage of strong fetal movement in the maximum duration was 3/10-30% for a total of 3 determination units in which the fetal movement was "strong".

The longest static time of the fetus is: and the total time occupied by the section with the longest duration in the judging unit sections which are continuously judged as the noise signals in the unit monitoring time. Such as: the total length of the measured data is 1 hour, a single judging unit is 6s, and in 1 hour, at most 100 judging units are continuously judged to be false, namely, no fetal movement signal is detected, and the maximum static time of the fetus is 10min (6s is 100).

The fetal liveness is as follows: the determination units determined as fetal movement signals per unit of monitoring time account for a percentage of all determination units per unit of monitoring time. Such as: the total length of the measured data is 1 hour, the total length of the measured data is 600 units to be judged, 150 judging units are judged to be true in 1 hour, and the fetal activity is 150/600-25%.

The fetal movement index can effectively represent the subjective feelings of the pregnant woman on fetal movement strength, characteristics and duration, and the situations of overlong longest fetal movement duration, overhigh strong fetal movement ratio in the longest fetal movement duration, overlong longest fetal static time, overlow fetal activity and the like all need to draw attention of doctors or pregnant women.

Besides the fetal movement times, the fetal movement monitoring system can also monitor a plurality of fetal movement evaluation indexes including fetal movement forms (such as knocking, duration, strength and the like), the longest duration time of fetal movement, the ratio of strong fetal movement in the longest duration time, the longest static time of a fetus and the fetal activity, and the indexes quantitatively describe the sensing conditions of the pregnant woman on fetal movement strength, characteristics and duration for the first time, and have important medical reference significance.

The foregoing is a general idea of the invention, on the basis of which further embodiments are provided below.

Example 1

Referring to fig. 3, the overall steps of the fetal activity signal acquisition and analysis device of the present embodiment for signal acquisition and analysis include:

1. acquiring signals of a multi-channel pressure sensor;

2. the preprocessing unit is used for preprocessing data;

3. the classification unit firstly carries out data cutting segmentation on the vibration signals acquired by each sensor respectively, so that the vibration signals of the parallel N channels form a plurality of independent judgment units;

4. analyzing and judging the single judging units one by one according to time lines through multi-channel feature fusion and comparative analysis: specifically classifying fetal movement signals into knock-type fetal movement or strong-type fetal movement, and specifically classifying noise signals into respiratory noise, sneeze noise or body movement noise;

5. the statistical unit forms a multi-dimensional fetal movement index in unit monitoring time according to the classification result of the classification unit: fetal activity times, fetal activity patterns, maximum duration of fetal activity, percentage of strong fetal activity within the maximum duration, maximum fetal static time, and fetal activity.

In this embodiment, the step of performing data preprocessing by the preprocessing unit includes: firstly, smoothing is carried out on a vibration signal by adopting a method at least comprising wavelet threshold denoising and Butterworth filtering, and then baseline drift correction is carried out by adopting an asymmetric least square baseline correction method.

The hardware acquisition circuit has power frequency interference and high-frequency noise, the pressure sensor has baseline drift due to self characteristics and external environment changes, the noise of the hardware modules is difficult to avoid, the factors have great influence on subsequent feature extraction and model design, and the good pretreatment of the acquired data is a necessary step for designing a good algorithm. The method can obtain better smoothing effect by means of wavelet threshold denoising, Butterworth filtering and the like, an ideal wavelet coefficient denoising threshold value is set, or a proper Butterworth critical frequency is set, the problems of power frequency interference and high-frequency noise existing in the system can be effectively solved, and the parameters are determined and perfected according to the sampling frequency of the system and the trigger frequency of effective signals. For baseline drift, the invention adopts an asymmetric least square smoothing method for correction, and the asymmetric parameter and the smoothness parameter are determined according to the situation. The preprocessing process of the system for typical signals is shown in fig. 4 and 5, wherein fig. 4 shows the preprocessing effect of the breathing signal in a resting state, and fig. 5 shows the preprocessing effect of the signal in the case of fetal movement triggering.

In this embodiment, the analyzing and determining the determination unit through the multi-channel feature analysis specifically includes:

4-1) carrying out multi-channel feature extraction on a single judgment unit, comprising the following steps:

4-1-1) extracting the Pearson correlation coefficient among the 4 channels, wherein the Pearson correlation coefficient calculation formula is as follows:

ρ_X，Y＝cov(X，Y)/σ_Xσ_Y＝E((X-μ_X)(Y-μ_Y))/σ_Xσ_Y；

ρ₀＝1/3(ρ_0，1+ρ_0，2+ρ_0，3)；

ρ₁＝1/3(ρ_0，1+ρ_1，2+ρ_1，3)；

ρ₂＝1/3(ρ_0，2+ρ_1，2+ρ_2，3)；

ρ₃＝1/3(ρ₀₃+ρ_1，3+ρ_2，3)。

wherein the fetal movement signalIs a weak physiological parameter signal, generally speaking, four pressure sensors can not be triggered simultaneously, the Pearson correlation coefficient between four channels is obviously weak, and rho₀，ρ₁，ρ₂，ρ₃The numerical value is smaller; noise signals such as breathing, sneezing, body movement and the like can simultaneously stimulate the four pressure sensors, the waveforms have similar development trends, and correspondingly, the pressure sensors have higher Pearson correlation coefficient rho₀，ρ₁，ρ₂，ρ₃The numerical value is large. The pearson correlation coefficient ρ > -0.6 is a strong correlation, 0.4 ≦ ρ ≦ 0.6 is a moderate correlation, and ρ < 0.4 is a weak or uncorrelated correlation.

4-1-2) calculate the peak-to-peak quotient for each of the 4 channels: according to the sampling frequency, the minimum horizontal unit distance between the peak values is set, and under the distance, the ratio of the maximum peak value to the minimum peak value in each channel is calculated and is recorded as peak division₀、peakDivision₁、peakDivision₂、peakDivision₃；

4-1-3) carrying out spectrum analysis and normalization processing on the four-channel signals, and calculating the average value amp of the amplitudes corresponding to different frequencies of the 4-channel signals_nStatistics of amp_nThe number of channels above the amplitude threshold, denoted count_{(amp＞thresholdAmp)}(ii) a In general, the amplitude threshold after the normalization process is about 0.2. The periodic regular respiratory signal is represented as a cosine signal, the main frequency number count_{(amp＞thresholdAmp}) And 2, the two frequencies respectively correspond to the constant term and the cosine frequency.

4-1-4) matrix signature analysis: in a single judging unit, vibration signals of 4 channel signals form an M & ltx & gt 4 matrix, wherein M represents the product of sampling frequency f and the time duration t of the judging unit, namely M & ltx & gt; such as: the sampling frequency f is 20Hz and the duration of a single decision unit is 10s, the pressure data of the four-channel signal form a 200 x 4 matrix, see fig. 6.

For this matrix, mean value of row vectors is represented by meanRowValue, index represents index of row vectors, and the following features are taken:

a) difference between maximum and minimum of row vector meanr_max-min，

r_max-min＝meanRowValue_max-meanRowValue_min，

wherein when r is_max-minGreater than a certain threshold value and index difference index_max-minSneeze signals can be effectively distinguished < 1.5f, and in general, the threshold value may be set to 3 times the breathing amplitude.

4-2) carrying out specific classification on the vibration signal of the judging unit according to the multi-channel characteristics obtained in the step 4-1) on a fetal movement signal and a noise signal, and referring to fig. 7, the method comprises the following steps:

4-2-1) setting 2 pearson correlation coefficient decision thresholds a and B, where a is 0.4 and B is 0.6; the mean value of the Pearson's correlation coefficients of all channels is calculated_nMinimum value of (1)_minComparing the comparison result with the A and the B, and judging according to the comparison result according to the following different steps; according to statistical experience, the field grades of the Pearson correlation coefficient values are divided as follows: 0.8-1.0, very strong correlation; 0.6-0.8, strongly correlated; 0.4-0.6, moderately relevant; 0.2-0.4, weakly correlated; 0.0-0.2, very weakly or not; < 0, negative correlation. A, B two threshold values, the number N of channels and the statistical experience need to be considered comprehensively, and the determination and the improvement are carried out; in the embodiment, a is 0.4, and B is 0.6;

4-2-2)(ρ₀、ρ₁、ρ₂、ρ₃)_minwhen the peak value is more than 0.6, counting the peak-to-peak quotient peak division N of each channel in the 4 channels;

I. if at least one channel exists, the peak-to-peak quotient is larger than a set threshold value P_TWhen the temperature of the water is higher than the set temperature,P_Tjudging whether the tire is in a knocking form or not, if so, judging whether the tire is in a knocking form or not, and outputting a classification result, otherwise, continuing to judge according to the following steps; according to the experimental results, the threshold value P_TThe method needs to be determined and perfected according to the sensing characteristics (sensitivity, linearity and the like) of a sensing device; in general, 2.0 < ═ P_T2.5; in this example, P is selected_T＝2；

II. Statistics count_{(amp＞thresholdAmp}) Number of dominant frequencies count_{(amp＞thresholdAmp)}When J is equal to 2, judging the breathing noise and outputting a classification result; otherwise, continuing to judge according to the following steps; according to a frequency domain analysis formula, a respiratory signal is expressed as a cosine waveform, and the number J of main frequencies is 2;

III, setting a matrix characteristic analysis threshold value if r_max-min> threshold value and index_max-minJudging sneezing noise when the value is less than I f, otherwise judging body movement noise; outputting a classification result, wherein I is 1.5; according to the experimental result, the threshold value needs to be determined and perfected according to the sensing characteristics (response curve, linearity and the like) of the sensing device, and the proportionality coefficient I needs to be determined and perfected according to the sample performance, wherein in general, I is more than or equal to 1 and less than or equal to 2.5; in the embodiment, 1.5 is selected;

4-2-3)ρ_minif the number of the Pearson correlation coefficients smaller than A in the Pearson correlation coefficients of all the channels is not more than k, judging the channel to be in knocking type fetal movement, otherwise, judging the channel to be in strong type fetal movement, and outputting a classification result; wherein k is 3; according to the experimental result, the value of k needs to be determined and perfected according to the number of channels N and the sample performance, generally, k is more than N/2; in the embodiment, k is 3;

4-2-4)A＜ρ_minwhen the peak-to-peak quotient is less than B, counting the peak-to-peak quotient peak division N of each channel in the N channels, if the peak-to-peak quotient of at least one channel is larger than a set threshold value 2, judging according to the following step a), otherwise, judging according to the following step B);

In the present embodiment, the signal is classified into 5 categories such as noise (breathing), noise (body movement), noise (sneezing), fetal movement (knocking), and fetal movement (strong) by the above algorithm, and the 5 categories are respectively represented by labels 0 (breathing), -1 (body movement), -2 (sneezing), 1 (knocking), and 2 (strong), as shown in fig. 7. In this embodiment, the performance of the classification algorithm is also evaluated, and the method adopted is as follows: calculating Precision indexes and Recall rate Recall indexes of each class, and multiplying the Precision indexes and the Recall rate Recall indexes of the whole classification model by the weight ratio of the class in the total sample number to obtain the Precision indexes and the Recall rate Recall indexes of the whole classification model, so as to deduce the F1 value of the whole classification model, thereby judging the performance of the algorithm: the higher the value of F1, the more desirable the classification model.

Rate of accuracy

Recall rate

Precision＝precision₀*w₀+precision_-2*w_-2+precision_-1*w_-1+precision₁*w₁+precision₂*w₂ (3)

Recall＝recall₀*w₀+recall_-2*w_-2+recall_-1*w_-1+recall₁*w₁+recall₂*w₂ (4)

The above equations (1) and (2) respectively represent algorithms of accuracy and recall, TP represents a positive case of correct prediction, FP represents a positive case of incorrect prediction, and FN represents a negative case of incorrect prediction. The significance of the accuracy rate is: the correct predicted positive (TP) accounts for the proportion of all predicted positive (TP + FP), and the recall ratio has the following meaning: correctly predicted as the proportion of positive (TP) to all positive instances of the sample (TP + FN);

(3) and (4) Precision and Recall calculation formulas of the design algorithm of the invention, w₀Represents the weight of label 0 (breath) in all samples, w_-2、w_-1、w₁、w₂And so on; precision₀Indicating a precision rate, recall, identified as tag 0 (breath)₀Indicating a recall rate identified as tag 0 (breath), and so on for other categories;

(5) representing the calculation formula of the algorithm F1 value of the invention.

And (4) evaluation results:

the fetal movement monitoring data of a pregnant woman are randomly extracted, 1498 groups of data of 30 weeks and 2 days of pregnancy, 32 weeks and 7 days of pregnancy and 36 weeks and 4 days of pregnancy of the pregnant woman are taken as examples, the statistics of classification results of 5 types of signals of the algorithm model are shown in the following table 1, the horizontal axis represents a real type, and the vertical axis represents a prediction type. The following description takes the label 0 (breath) as an example: the test set actually has 949 total tags 0 (breaths), the first row and the first column representing: a total of 880 cases actually labeled 0 (breaths) and predicted labeled 0 (breaths); the first row and the second column represent: actually label 0 (breathing) and predicted to be 0 cases of label-2 (sneezing); the first row and the third column indicate: actually label 0 (breathing), predicted to be 0 cases of label-1 (body movement); the first row and the fourth column represent: actually label 0 (breath), predicted to be 50 cases of label 1 (tap); the first row and the fifth column represent: there are 19 cases that are actually labeled 0 (breathing) and predicted to be labeled 2 (strong).

According to the evaluation method, the accuracy of the algorithm classification model calculated after weight processing is as follows: 0.920, the recall ratio is: 0.912, F1 values: 0.916. the fetal movement judgment belongs to weak physiological signal monitoring, noise interference is more, the classification method achieves the performance while considering fetal movement form judgment and noise type identification, and the excellent performance of the classification method is fully demonstrated.

TABLE 1

Actual/predicted	Breath 0	Sneezing-2	Body movement-1	Knocking 1	Intensity 2
						Breath 0	880	0	0	50	19
Sneezing-2	0	16	1	0	0
						Body movement-1	3	0	98	9	7
Knocking 1	22	0	0	294	10
						Intensity 2	1	0	3	7	78

Example 2

This embodiment still provides a concrete sensor on above basis, in this embodiment, the sensor includes packaging structure and sets up sensor body in the packaging structure, packaging structure includes the bottom plate, the lock is in apron on the bottom plate and setting are in compress tightly the piece between bottom plate and the apron.

In this embodiment, piezoelectric sensor is selected to the sensor body, in order not to influence the sensing characteristic of device, avoids buckling, avoids sticky as far as possible when encapsulating the sensor, and the packaging structure that this embodiment provided can satisfy this requirement, and can gather the weak physiology fetal movement signal of pregnant woman's stomach wall well.

In the embodiment, the middle part of the bottom plate is provided with an installation cavity capable of accommodating the sensor body through a plurality of circumferentially arranged limiting barrier strips, and the middle part of the installation cavity is provided with a first pressing boss;

the inner side surface of the pressing piece is provided with a second pressing boss matched with the first pressing boss, so that the sensor body is pressed between the first pressing boss and the second pressing boss. The boss type compression fit mode can ensure that the sensor is in good contact, and is beneficial to stably and accurately transmitting piezoelectric signals.

The bottom plate is provided with a plurality of limiting barrier strips, the limiting barrier strips are arranged on the bottom plate, the outer sides of the limiting barrier strips are provided with annular bulges, the periphery of the pressing piece extends outwards to form a first annular edge, the lower surface of the first annular edge is used for being arranged on the annular bulges in a matched mode, and the upper surface of the first annular edge forms a first annular mounting table top;

the outer side of the pressing piece protrudes outwards to form an arc-shaped contact surface which is used for being in contact with the abdominal wall of a pregnant woman, and a second annular mounting table top is formed on the periphery of the arc-shaped contact surface; adopt the arc structure with pregnant woman's stomach wall contact surface, the direct amazing piezoelectric sensor of child fluctuation of being convenient for promotes sensor signal acquisition efficiency, and in addition, the wearing travelling comfort of arc contact surface is better.

The periphery of the cover plate extends outwards to form a second annular edge which is used for being in matched contact with the bottom plate outside the annular bulge, an annular step table is arranged inside the cover plate, so that a first annular pressing table face and a second annular pressing table face are formed inside the cover plate, the first annular pressing table face is used for being in matched contact with the first annular mounting table face, and the second annular pressing table face is used for being in matched contact with the second annular mounting table face.

And a gap is formed between the pressing piece and the cover plate in the radial direction, so that the pressing piece is in a loose state in the radial direction and a weak fetal movement signal is favorably transmitted.

Wherein, two pin leading-out holes are further formed in the bottom plate so as to lead out a lead of the sensor body. In a preferred embodiment, the transmission between the lead of the sensor body and the external circuit can be flexible flat cable or fabric lead, and the circuit has the characteristics of invisibility, flattening, flexibility and the like. Packaging structure can integrate in products such as present support binder, child prison area or elasticity are weaved, also can show comfort, the gas permeability that promotes the long-time wearing of pregnant woman with the circuit encapsulation in the silica gel material.

In this embodiment, the 4-position limiting barrier strips form a mounting cavity of a square area to be provided with a square sensor body. The limiting barrier strips can ensure that the effective sensing area of the sensor body is positioned in the mounting cavity, and the position of the sensor body along the axial direction of the bottom plate is kept fixed through limitation; the sensor body is compressed between the first compression boss and the second compression boss, so that the sensor body is kept fixed along the radial position of the bottom plate.

In the embodiment, when the sensor body is fixedly installed in the packaging structure, the sensor body is firstly arranged on the first pressing boss of the base plate and is just clamped in the installation cavity formed by the limit barrier strip; then the pressing piece is buckled on the bottom plate, the second pressing boss presses the sensor body, and the first annular edge is arranged on the annular bulge in a matched mode; then the cover plate is buckled on the bottom plate, so that the second annular edge is in contact with the periphery of the bottom plate, the first annular pressing table top is arranged on the first annular mounting table top in a matched mode, the second annular pressing table top is arranged on the second annular mounting table top in a matched mode, and then the cover plate and the bottom plate are fixed together. When the pregnant woman abdominal wall is worn, the pregnant woman abdominal wall is in direct contact with the arc contact surface of the pressing piece, vibration generated by the abdominal wall is transmitted to the sensor body through the pressing piece, and therefore the acquisition of vibration signals is achieved.

While embodiments of the invention have been disclosed above, it is not limited to the applications listed in the description and the embodiments, which are fully applicable in all kinds of fields of application of the invention, and further modifications may readily be effected by those skilled in the art, so that the invention is not limited to the specific details without departing from the general concept defined by the claims and the scope of equivalents.

Claims

1. A fetal movement signal acquisition and analysis device, comprising: the device comprises a signal acquisition module which is arranged on the abdominal wall of a pregnant woman in a wearable manner and a signal analysis module which is in communication connection with the signal acquisition module;

the fetal liveness is as follows: the percentage of the judging units judged as fetal movement signals in the unit monitoring time to all the judging units in the unit monitoring time;

the step 2) specifically comprises the following steps:

2-1-2) calculate the peak-to-peak quotient for each of the N channels: according to the sampling frequency, the minimum horizontal unit distance between the peak values is set, and under the distance, the ratio of the maximum peak value to the minimum peak value in each channel is calculated and is recorded as peak division_N；

2-1-3) carrying out spectrum analysis and normalization processing on the N channel signals, and calculating the average value amp of the amplitudes corresponding to different frequencies of the N channel signals_nStatistics of amp_nThe number of channels above the amplitude threshold, denoted count_{(amp>thresholdAmp)}；

a) difference between maximum and minimum of row vector mean

r_max-min，r_max-min＝meanRowValue_max-meanRowValue_min，

b) index difference index between maximum value of row vector mean and minimum value of row vector mean_max-min,index_max-min＝index_max-index_minTherein, index_maxIndex, representing the maximum value of the mean of the row vectors_minAn index representing the minimum of the row vector means;

2-2-2)ρ_minwhen the peak-to-peak quotient is greater than B, counting the peak-to-peak quotient peak division of each channel in the N channels_N；

I. If at least one channel exists, the peak-to-peak quotient is larger than a set threshold value P_TJudging the fetal movement in a knocking mode and outputting a classification result, otherwise, continuously judging according to the following steps, wherein P is more than or equal to 2.0_T≤2.5；

II. Statistics count_{(amp>thresholdAmp)}Number of dominant frequencies count_{(amp>thresholdAmp)}When the number is J, judging the breathing noise, and outputting a classification result, wherein J is a positive integer; otherwise, continuing to judge according to the following steps;

III, setting a matrix characteristic analysis threshold value if r_max-min> threshold value and index_max-min<Judging sneezing noise when I f is detected, otherwise, judging body movement noise; outputting a classification result, wherein I is a proportionality coefficient, and is more than or equal to 1 and less than or equal to 2.5;

2-2-3)ρ_minif the number of the Pearson correlation coefficients smaller than A in the Pearson correlation coefficients of all the channels is not more than k, judging the channel to be in knocking type fetal movement, otherwise, judging the channel to be in strong type fetal movement, and outputting a classification result; wherein k is>N/2；

2-2-4)A＜ρ_minWhen the peak value is less than B, the peak-to-peak quotient peak division of each channel in the N channels is counted_NIf at least one channel exists, the peak-to-peak quotient is larger than a set threshold value P_TIf so, judging according to the following step a), otherwise, judging according to the following step b);

a) if the number of the Pearson correlation coefficients smaller than A in the Pearson correlation coefficients of all the channels is not more than k, judging the channel to be in knocking type fetal movement, otherwise, judging the channel to be in strong type fetal movement, and outputting a classification result; wherein k > N/2;

b) setting a matrix characteristic analysis threshold value if r_max-min> threshold value and index_max-min<Judging sneezing noise when I f is detected, otherwise, judging body movement noise; and outputting a classification result, wherein I is a proportionality coefficient, and I is more than or equal to 1 and less than or equal to 2.5.

2. The fetal activity signal acquisition and analysis device of claim 1, wherein the step of preprocessing the data by the preprocessing unit comprises: firstly, smoothing is carried out on a vibration signal by adopting a method at least comprising wavelet threshold denoising and Butterworth filtering, and then baseline drift correction is carried out by adopting an asymmetric least square baseline correction method.

3. The fetal activity signal acquisition and analysis device according to claim 2, wherein the signal acquisition module comprises 4 sensors which are arranged on the abdominal wall of the pregnant woman at uniform intervals along the transverse direction and are used for acquiring vibration signals of the abdominal wall of the pregnant woman, so that 4 channels of data can be acquired in parallel.

4. The fetal movement signal acquisition and analysis device according to claim 3, wherein the step 2) comprises:

ρ_X,Y＝cov(X,Y)/σ_Xσ_Y＝E((X-μ_X)(Y-μ_Y))/σ_Xσ_Y；

where cov (X, Y) represents the covariance, σ, between two variables X, Y_X，σ_YRepresents the standard deviation of two variables X, Y; the total number of 6 Pearson correlation coefficients, namely rho, among the four channels_0,1，ρ_0,2，ρ_0,3，ρ_1,2，ρ_1,3，ρ_2,3Calculating the mean value of the pearson correlation coefficient of each channel for the other channels, namely:

ρ₀＝1/3(ρ_0,1+ρ_0,2+ρ_0,3)；

ρ₁＝1/3(ρ_0,1+ρ_1,2+ρ_1,3)；

ρ₂＝1/3(ρ_0,2+ρ_1,2+ρ_2,3)；

ρ₃＝1/3(ρ_0,3+ρ_1,3+ρ_2,3)；

2-1-3) carrying out spectrum analysis and normalization processing on the four-channel signals, and calculating the average value amp of the amplitudes corresponding to different frequencies of the 4-channel signals_nStatistics of amp_nThe number of channels above the amplitude threshold, denoted count_{(amp>thresholdAmp)}；

a) difference r between maximum and minimum of row vector mean_max-min，

r_max-min＝meanRowValue_max-meanRowValue_min，

2-2-1) setting 2 pearson correlation coefficient decision thresholds a and B, where a is 0.4 and B is 0.6; the mean value rho of the Pearson correlation coefficients of all channels₀、ρ₁、ρ₂、ρ₃Minimum value of (p)₀、ρ₁、ρ₂、ρ₃)_minComparing the comparison result with the A and the B, and judging according to the comparison result according to the following different steps;

2-2-2)(ρ₀、ρ₁、ρ₂、ρ₃)_minwhen the peak-to-peak quotient is greater than 0.6, the peak-to-peak quotient peak division of each channel in 4 channels is counted_N；

II. Statistics count_{(amp>thresholdAmp)}If count_{(amp>thresholdAmp)}When J is equal to 2, judging the breathing noise and outputting a classification result; otherwise, continuing to judge according to the following steps;

III, setting a matrix characteristic analysis threshold value if r_max-min> threshold value and index_max-min<Judging sneezing noise when I f is detected, otherwise, judging body movement noise; outputting a classification result, wherein I is 1.5;

2-2-3)(ρ₀、ρ₁、ρ₂、ρ₃)_minif < A, if all channels have Pearson's phaseIn the correlation coefficients, judging that knocking type fetal movement is performed when the number of Pearson correlation coefficients smaller than A is not larger than k, otherwise judging that strong type fetal movement is performed, and outputting a classification result; wherein k is 3;

2-2-4)A＜(ρ₀、ρ₁、ρ₂、ρ₃)_minwhen the peak value is less than B, the peak-to-peak quotient peak division of each channel in the N channels is counted_NIf the quotient of the peak value and the peak value of at least one channel is larger than a set threshold value 2, judging according to the following step a), otherwise, judging according to the following step b);

b) setting a matrix characteristic analysis threshold value if r_max-min> threshold value and index_max-min<Judging sneezing noise when I f is detected, otherwise, judging body movement noise; and outputting a classification result, wherein I is 1.5.

5. The fetal activity signal acquisition and analysis device according to any one of claims 1-4, wherein the sensor comprises a packaging structure and a sensor body arranged in the packaging structure, and the packaging structure comprises a bottom plate, a cover plate buckled on the bottom plate and a pressing piece arranged between the bottom plate and the cover plate.

6. The fetal movement signal acquisition and analysis device according to claim 5, wherein a mounting cavity capable of accommodating the sensor body is formed in the middle of the bottom plate through a plurality of circumferentially arranged limiting barrier strips, and a first pressing boss is arranged in the middle of the mounting cavity;

7. The fetal movement signal acquisition and analysis device according to claim 6, wherein an annular protrusion is arranged on the bottom plate around the outer sides of the limit barrier strips, a first annular edge is formed on the outer periphery of the pressing piece in an outward extending manner, the lower surface of the first annular edge is used for being arranged on the annular protrusion in a matching manner, and a first annular mounting table surface is formed on the upper surface of the first annular edge;

8. The fetal activity signal collecting and analyzing apparatus of claim 7, wherein the cover plate has an outer periphery outwardly extending to form a second annular rim for mating contact with the base plate outside the annular protrusion, and the cover plate has an annular step formed therein to form a first annular pressing land and a second annular pressing land in the cover plate, the first annular pressing land being for mating contact with the first annular mounting land and the second annular pressing land being for mating contact with the second annular mounting land.

9. The fetal movement signal acquisition and analysis device of claim 8, wherein the compressing member has a gap with the cover plate in a radial direction so that the compressing member is in a relaxed state in the radial direction thereof.