CN116186462A - Respiratory frequency detection method based on air flow sensor and application thereof - Google Patents

Abstract

The invention discloses a respiratory rate detection method based on an air flow sensor and application thereof, and belongs to the technical field of sensing detection. The implementation method of the invention comprises the following steps: (1) performing fixed integration on the curve of the respiratory wave; (2) starting from a starting scanning point, scanning sampling points in sequence, and finding a point closest to the starting point as a scanning end point, wherein the point is preferentially satisfied at or close to a zero point of respiratory waves; (3) according to the amplitude condition of noise near the t axis in the coordinate axis, adding filtering conditions: using the upper threshold and the lower threshold so that signals falling within the threshold range are treated as 0; (4) after dividing the breathing cycle, the average breathing frequency of a certain period of time can be calculated based on the number of the breathing cycles in the certain period of time. The invention realizes the detection of respiratory frequency based on the air flow sensor, and can further realize the period division and frequency extraction of respiratory wave. The invention has the advantages of high detection accuracy and low time complexity.

Description

A respiratory rate detection method based on an air flow sensor and its application

technical field

The invention belongs to the field of sensing technology, and in particular relates to a breathing frequency detection method based on an airflow sensor and an application thereof.

Background technique

Respiratory monitoring instrument is an important device to realize the monitoring of human respiratory health. According to statistics from the Ministry of Health, nearly 300 million people in my country are infected with respiratory diseases every year, and many people have delayed treatment or even lost life.

With the acceleration of the aging trend of my country's society, the demand for health services continues to grow, and the demand for health monitoring is also increasing. In the "Healthy China 2030" Planning Outline, medical and health care has been upgraded to the national strategic level, which emphasizes the need to To develop smart medical care based on innovative technologies, it is proposed to strengthen breakthroughs in key technologies such as precision medicine and smart medical care. Respiratory monitoring sensors and their supporting signal processing and digital information processing are precisely in line with the development trend of smart medicine and precision medicine, have broad market demand and important strategic value, and can also promote the development of medical informatization and intelligence.

At present, there are many types of sensors that can detect the state of human respiration, and it is important to effectively and accurately monitor the physiological characteristics of human respiration, whether it is in the monitoring of clinical physiological indicators related to respiratory diseases or in daily monitoring of respiratory health. meaning. Among these physiological characteristics, parameters such as instantaneous air flow, frequency, tidal volume, and average ventilation volume of breathing are included, among which respiratory frequency is a very important physiological parameter, and tidal volume describes the ventilation volume of a complete breathing cycle. Physiological parameters are closely related to the cycle and frequency of breathing. For the detection of respiratory frequency and respiration rate, many domestic and foreign studies have proposed many methods over the years, and some of them are even very mature methods and technologies. Based on different types of sensors and their working mechanism, there are different signal processing methods to detect respiratory rate. For example, there is a method of detecting the respiratory frequency through a sensor that can detect the physical movement of the human chest; there is also a method of detecting the temperature and humidity changes of the respiratory airflow through a temperature or humidity sensor to detect the respiratory frequency; there is also a method that can detect the heart rate or pulse. A method for inferring respiratory rate using a sensor. These methods are based on different sensing mechanisms, each has its own characteristics, and each has its own advantages and disadvantages for different application scenarios. However, in the field of air flow sensor used for respiratory monitoring, there are not many methods for detecting respiratory frequency. One of the commonly used methods is to divide different respirations by detecting the high and low changes of signal amplitude to achieve the detection of respiratory frequency. the goal of. This method also has some defects, such as weak resistance to noise and random airflow jitter, and the detection accuracy will be affected by airflow jitter and noise. Therefore, this method is often used in conjunction with filters to achieve the purpose of noise reduction, but at the same time it will also increase the time complexity of method execution. Therefore, the present invention explores a new frequency detection method, which is suitable for processing digital sensor signals, and realizes dividing each breathing cycle with high accuracy and efficiency, thereby detecting the breathing frequency.

Contents of the invention

The main purpose of the present invention is to provide a respiratory frequency detection method and application based on an airflow sensor, which can realize respiratory frequency detection based on the airflow sensor, and can further realize cycle division and frequency extraction of respiratory waves. The invention has the advantages of high detection accuracy and low time complexity.

The present invention is achieved through the following technical solutions:

A method for detecting respiratory frequency based on an air flow sensor disclosed by the present invention comprises the following steps:

① Definitely integrate the curve of the respiratory wave, and set a threshold δ. If the definite integral value of a time series signal satisfies formula (1):

Then this segment signal is divided into a respiratory cycle; wherein, δ in the above formula is a threshold, S(t) is a respiratory wave, which is a function of time t, x is the starting point of scanning, and L is the upper limit length of scanning;

②Start from the initial scanning point, scan the sampling points in order, and find the point closest to the initial point as the end point of the scan, and the point is preferably at or close to the zero point of the respiratory wave;

③According to the amplitude of the noise near the t-axis in the coordinate axis, add filter conditions: use the upper threshold and the lower threshold, so that the signal value falling within the threshold range is treated as 0; among them, thr _h >0, thr _l <0;

④ After dividing the breathing cycle, based on the number of breathing cycles in a certain period of time, the average breathing frequency of this period of time can be calculated. The formula is as follows:

Wherein, R is the respiratory rate, T is a certain period of time, and n is the number of respiratory cycles monitored within this period of time.

Preferably, the timing signal is a discrete digital signal sampled by the ADC, therefore, the discrete form of formula (1) is as follows:

Among them, i represents a certain sampling point, S(i) is a discrete function about sampling point i, and fs is the sampling frequency.

The present invention also discloses a respiratory rate detection method based on an air flow sensor, which specifically includes the following steps:

S1. Read in a piece of discrete time series data to be detected, sampling frequency, upper threshold threshold thr _h for eliminating noise influence, upper threshold thr _l , threshold δ for limiting the upper limit of integration, and create a and to-be-detected An empty array T of the same length as the time series is initialized, and the T is used as a backup array to be detected;

S2. Preprocessing the time-series data: when the time-series data to be detected is not empty, traverse the time-series data to be detected, if the data value S(i)>thr _h at a certain sampling point, then set the value Assign the value to the corresponding position of the T array; if S(i)<thr _l , then assign the value to the corresponding position of the T array; otherwise, assign 0 to the corresponding position of the T array;

S3. Create and initialize an array I for storing the segmentation points of the respiratory cycle; set the upper limit of the duration of each scan to 30s; create and initialize an array A for storing discrete definite integral values, and describe the integral value changing array D;

S4. If the total length of the data is less than 2, end and return 0;

S5. Traverse the array T to be detected, and use the next data point of the first data as the initial scanning data;

S6, start a scan: the upper limit of each time length is 30s, if the time series data to be detected is less than 30s, then the maximum time length of the time series data is the upper limit of the window length of each scan;

S7. Calculate the value of the discrete definite integral of the signal value from the initial sampling point of the scan to the current sampling point every time a sampling point is traversed, and store it in the array D. After looping step 7 to the upper limit of the scanning time Go to the next step;

S8. Read and traverse the array A used to store the discrete definite integral value that stores the scan in twos in turn, calculate their difference and assign it to the array D used to describe the change of the integral value, that is, D[i ]=A[i+1]-A[i];

S9. Record the signal value of the first sampling point i ₀ of this scan as S(i ₀ ); if S(i ₀ )≥0, traverse the array D and detect that D[i+1]≥D[i ], take the sampling point that causes this situation as the temporary cut-off point; if S(i ₀ )<0, traverse the array D until D[i+1]≤D[i] is detected, take the sampling point that causes this situation point is the temporary cut-off point;

S10. Obtain the temporary cut-off point of the previous step, calculate the definite integral value of the signal from the initial sampling point of this scan to the temporary cut-off point, and if it is not greater than the set threshold δ, it is determined that the temporary cut-off point is valid and As the time division point of division respiratory cycle, and be stored in the array I that is used to store the division point of division respiratory cycle, and the next point of this sampling point is used as the starting point of next scan;

S11. Obtain the scan starting point of the previous step, start a new scan, and repeat steps 6-11;

S12, the circular scanning step, repeating steps 6-12, until the data to be detected is scanned;

S13, return the array I for storing the segmentation point of the segmentation respiratory cycle, each value in the array I is the time point of the segmentation respiratory cycle, the time period of each respiratory cycle is obtained by the segmentation point, and the segmentation point The number is the number of respiratory cycles of this segment of the respiratory wave; the average respiratory frequency is obtained by dividing the total duration of the time series data by the number of respiratory cycles, that is, the respiratory frequency detection is realized based on the air flow sensor.

Beneficial effect:

1. In the scheme provided by the present invention, the upper limit threshold and the lower limit threshold are used, so that the signal value falling within the threshold range is treated as 0, so as to eliminate and shield the signal of noise and random airflow jitter, and prevent it from being regarded as a normal signal and then Influence cycle division. Therefore, this solution is highly resistant to noise and random airflow jitter, and the detection accuracy is less affected by airflow jitter and noise.

2. Because the respiratory rate detection method proposed by the present invention is based on the digital sensor signal for direct processing, it can be realized in a digital circuit or software programming mode more conveniently, without complex circuits, so that it can be put into practice quickly application.

3. In the solution provided by the present invention, the method of calculating integrals is used to divide the period instead of using the conventional method of finding zero points, which can make the positive signal value and negative signal value of the divided period more symmetrical, so that the period The division is more reliable, and each breathing cycle can be divided with higher precision and efficiency. At the same time, it has been statistically verified that the scheme actually has high detection accuracy and low time complexity.

4. The solution disclosed in the present invention can be applied to period division and frequency calculation of most periodic signals because of its excellent performance in processing and period division of periodic signals.

Description of drawings

Below in conjunction with accompanying drawing and embodiment the present invention is further described:

Fig. 1 is the curve example diagram of the respiratory wave mentioned in the present invention;

Fig. 2 is the signal diagram of the airflow jitter and noise part of the breath wave mentioned in the present invention;

Fig. 3 is a method flowchart of the present invention;

Fig. 4 is the result figure that the method of the present invention implements and divides based on the part of the respiratory cycle of the sample test;

Fig. 5 is a schematic structural diagram of a respiratory monitoring system provided by the implementation method of the present invention;

Fig. 6 is an example diagram of a user terminal of a human breathing monitoring system applied to the implementation method of the present invention;

Fig. 7 is an error distribution diagram of detecting respiratory frequency by the method of the present invention during implementation.

Among them, the signal of the airflow jitter and noise part of the respiratory wave mentioned in Figure 2 is in the part framed by the dotted line. It can be seen that compared with other normal signals, the amplitude of the airflow jitter and noise clutter is obviously small, and the frequency is There are highs and lows.

The sample mentioned in Fig. 4 is the real human body respiratory data that is actually collected by the breathing monitoring system that the present invention implements, and 4 pieces of figures shown in the figure are exactly realized with software programming (programming language is Python language) according to the method, to sample Part of the data in is the result of performing respiratory cycle division. The vertical dotted line, that is, the dotted line crossing the t-axis through the respiratory cycle division point t=t _i (i=0,1,2...), can be clearly seen These dotted lines divide the breathing cycles one by one. The three horizontal straight lines respectively represent the upper threshold thr _h =0.15, the t-axis (time axis) and the lower threshold thr _l =-0.15 from top to bottom.

Detailed ways

In order to better illustrate the purpose and advantages of the present invention, the content of the invention will be further described below in conjunction with the accompanying drawings and examples.

As shown in Figure 1, taking the respiratory wave as an example, in the timing signal showing periodic changes, it can be found that the low-frequency wave with obvious amplitude has a certain periodicity, and each cycle of the signal corresponds to one time in this sample breathe. However, it can also be found that when the breathing speeds up or slows down, the vibration frequency of the breath wave will speed up or slow down, corresponding to the different lengths of each breathing cycle. In the actual application and implementation process, the length of the timing signal to be detected is usually relatively long, and may include many breathing cycles with large differences in length. If the timing signal of the respiratory wave is only processed by Fourier transform or other similar methods, it is not suitable for dividing the respiratory cycle and obtaining the respiratory frequency within the detection range, and it is too suitable for this application scenario, so consider other implementation methods.

The method is statistically verified to have high detection accuracy and low time complexity. This method can not only overcome the deficiencies of the prior art, but also has a certain degree of technical mobility. It can not only deal with the period division and frequency extraction of the respiratory wave mentioned in the present invention, but also migrate to other periods and frequency that need to extract periodic wave signals. frequency in the technical field. Notice that when the breathing wave is vibrating, in each breathing cycle, because the total amount of gas inhaled and exhaled is approximately equal, in a breathing cycle, the curve of the breathing wave is definitely integrated, even if the integral result is not It must be 0, and it will be a relatively small number. A threshold δ can be set for this. If the definite integral value of a time series signal over a period of time satisfies formula (1):

Then the signal segment can be divided into one respiratory cycle. Wherein, δ in the above formula is a threshold value, which is a highest upper limit describing that the above integral value satisfies this condition. In practical application, δ can set its value according to specific signal conditions. S(t) is the respiratory wave, which is a function of time t; x is the starting point of the scan, and L is the upper limit length of the scan. In fact, this method is implemented by programming, so the timing signal used is not a continuous analog signal, but a discrete digital signal after being sampled by the ADC. Therefore, the discrete form of formula (1) is as follows:

Among them, i represents a certain sampling point, S(i) is a discrete function about sampling point i, and fs is the sampling frequency. The above two equations are one of the sufficient conditions for the method to divide the different periods of the respiratory wave, which are referred to as the sufficient condition (1) below. But only this sufficient condition is not enough, because the integral value of multiple periods may also satisfy the above formula, so other sufficient conditions are needed to ensure the correctness of period division.

Because in this application scenario of respiratory waves, the signal value at the beginning point of each respiratory cycle must be close to 0 (otherwise it should not be counted as a complete respiratory cycle), so another sufficient condition can be constructed based on this feature, that is, sufficient The construction idea of condition (2) is to start from the initial scanning point, scan the sampling points in order, and find the point that satisfies the sufficient condition (1) and is the closest point to the initial point as the end point of the scan, and this point is preferentially satisfied in the or close to the zero point of the respiratory wave. And in the specific implementation of this condition in this method, the innovation is not to directly seek the point i where the absolute value of S(i) is equal to 0 or less than a value close to 0, but by judging at this point and other points i Whether the integral value at the two points before and after has increased or decreased.

The above two sufficient conditions can basically ensure that each fluctuation period is divided. However, in practical applications, taking respiratory waves as an example, signals with high frequency and extremely low amplitude can be found everywhere in the signal curve. These are actually due to the use of more sensitive airflow sensors, and the airflow "jitters" when people breathe. "What is produced is the noise and clutter generated by the airflow. The noise in the normal signal has little effect, but if the noise occurs near the signal value of zero, it corresponds to the part of the breathing stop in the breathing process, as shown in Figure 3, if it is still constructed using the above two sufficient conditions method, this part of the noise signal will also be divided into separate breathing cycles, which is obviously not in line with the actual situation. To solve this contradiction, the conventional method is to use filters, such as Fourier filters or Butterworth low-pass filters, to filter out high-frequency noise, but this will first increase the time complexity of the method's operation, and secondly, it is not necessarily Ensure that the filter can clean up the noise: because some clutter in the breathing wave is more of a low-amplitude characteristic, and its frequency is not all much higher than the human breathing frequency, there may be some low-frequency noises whose frequency is close to or even It is lower than the upper limit of human breathing frequency, obviously this part of noise and clutter cannot be completely filtered out by the filter, but their existence greatly affects the correctness of breathing cycle division and breathing frequency extraction. This is to consider other methods, such as starting from the amplitude characteristics of the noise, and setting filter conditions to eliminate the influence of noise.

According to the amplitude of the noise near the t-axis in the coordinate axis, add a filter condition: use the upper threshold (threshold high, thr _h ) and the lower threshold (threshold low, thr _l ), so that the signal that falls within the threshold range is treated as 0. Among them, thr _h >0, thr _l <0, there is no specific fixed value for the upper and lower limits, because the amplitude of the noise band may vary in different scenarios, but it is statistically determined according to specific analysis or sampling experiments, and its The value should satisfy the signal in the noise band basically falls within the upper and lower limits, that is: satisfy the condition of thr _l <S _noise (t)<thr _h . With the above conditions, the method is constructed with this method, and the effect is good after the implementation verification, and the time complexity of the method is relatively low in theory.

After dividing the breathing cycle, based on the number of breathing cycles in a certain period of time, the average breathing frequency during this period can be calculated, the formula is as follows:

Among them, R is the respiratory rate, T is a certain period of time, and n is the number of respiratory cycles monitored during this period, and the unit is "breaths per minute" (BPM) or "Hertz" (Hz). First of all, the object of the present invention is aimed at an application of a high-performance air flow sensor in a human respiratory monitoring system. In order to realize the monitoring of human respiration, it is necessary to extract the physiological information of human respiration from the respiration signal, that is, the respiration wave mentioned above, including the instantaneous air flow, frequency, tidal volume, and average ventilation of respiration, among which the respiration frequency It is a very important physiological parameter, and tidal volume is a physiological parameter describing the ventilation volume of a complete breathing cycle, and they are closely related to the breathing cycle and frequency, so the present invention solves how to obtain each complete breathing cycle, and Based on this the respiratory rate is obtained.

In order to implement the method of the present invention, it is necessary to build an electronic system for respiratory monitoring based on a high-performance airflow sensor, and its system structure is shown in FIG. 5 . The system uses a high-performance, low-power, and small-volume airflow sensor chip as the core device to build a functional circuit. The circuit functions include signal reading, amplification, analog signal conversion to digital signal, Bluetooth signal transmission and reception. The airflow sensor chip is affected by the airflow to generate electronic signals, and the analog signal is converted into a digital signal through the functional circuit and the data is sent to the smart terminal (computer or smart phone) through the Bluetooth module, and the terminal software performs data processing and respiratory cycle division 1. Respiratory frequency calculation, that is, the realization carrier of the method is software.

For the setting of each threshold above, the specific reference value can be set through statistical sampling experiments. For example, in the implementation of the present invention, the respiratory wave data with a total duration of more than 810s are collected, summarized for sampling statistics, and determine the threshold for eliminating noise influence by finding the statistical average value and the maximum and minimum values of the amplitude of the noise part The value of the upper and lower limits of . After the implementation of the present invention, it is set as thr _h =0.15 and thr _l =-0.15 through multiple experiments, and the unit of the data is the unit of the signal, that is, volt (V). Regarding δ in the sufficient condition (1) above, it is not set as a fixed constant in this implementation, but by a proportional factor and the product of the integral value of the absolute value of each segment of the respiratory cycle (ie, the curve and the surface area enclosed by the time axis), the scale factor is set to 5% in the implementation of multiple experiments, so that the threshold will be more flexible for the division of respiratory cycles of respiratory waves of different intensities, but in practical applications It is also possible to set it as a fixed constant through experiments, because in different application scenarios, the parameter value should be set according to the actual data. The present invention mainly provides the above-mentioned train of thought and method.

Combined with the setting of the threshold value during implementation, the execution process of the method during implementation is as follows:

1. Read in the time series data array S(i) to be detected, i=0,1,2,...,L, i is a certain sampling time point, L is the maximum length of the array; sampling frequency fs, and thr _h = 0.15, thr _l =-0.15 threshold δ=5%; create and initialize backup array T.

2. Data preprocessing: when -0.15≤S(i)≤0.15, T(i)=0; otherwise T(i)=S(i).

3. Create and initialize the array I used to store the segmentation points of the respiratory cycle; set the upper limit of the duration of each scan to 30s; create and initialize the array A used to store the discrete definite integral value, and to describe the integral value Change the array D.

4. If the total data length L<2, end and return 0.

5. Traverse the array T to be detected, and take the next point i=1 after the starting point i=0 as the starting point of scanning.

6. Start a scan: the upper limit of each scan is 30s. If the duration of the time series data to be detected is less than 30s, the maximum duration of the time series data is the upper limit of the window length of each scan.

7. Every time a sampling point is traversed, the value of the discrete definite integral of the signal value from the initial sampling point of the scan to the current sampling point is calculated, and stored in the array D, and this step is repeated until the upper limit of the scanning time Go to the next step;

8. Read and traverse the array A used to store the discrete definite integral value that stores the scan two by two in turn, calculate their difference and assign it to the array D used to describe the change of the integral value, that is, D[i ]=A[i+1]-A[i].

9. Record the signal value of the first sampling point i ₀ of this scan as S(i ₀ ); if S(i ₀ )≥0, traverse the array D and detect that D[i+1]≥D[i ], take the sampling point that causes this situation as the temporary cut-off point; if S(i ₀ )<0, traverse the array D until D[i+1]≤D[i] is detected, take the sampling point that causes this situation point is the temporary cut-off point;

10. Obtain the cut-off point i=k of the previous step, and calculate the discrete definite integral value J ⁺ of all positive points and the discrete value of negative points in the signal values from the initial sampling point of this scan to the temporary cut-off point respectively Definite integral value J ^- . Since δ=5%, if it satisfies: ||J ⁺ |-|J-||<5%||J ⁺ |+|J-||, then it is determined that the temporary cut-off point is valid and used as a division point for dividing the respiratory cycle , its k is stored in the array I, and the next point i=k+1 of the sampling point is taken as the starting point of the next scan.

11. Obtain the scan starting point of the previous step, start a new scan, and repeat steps 6-11.

12. Cycle the scanning steps and repeat steps 6-12 until the data to be detected is scanned.

13. Return the array I used to store the segmentation points for dividing the breathing cycle. Each value in the array I is the time point of dividing the breathing cycle. Check the length of the array to get the number of breathing cycles C. Divide C by the total duration to get the total Respiratory rate RR.

In addition, as shown in Figure 6, this method is applied to the detection effect displayed by the user end (mobile app) of the system. Regarding the performance and effect of detection, the present invention has carried out actual application test. The test process is as follows: 1. The tester wears a breathing mask and connects to the respiratory monitoring system. 2. The tester starts to breathe, and counts the number of breaths by himself as the actual number of breaths. 3. When the test is finished, check and record the number of respirations detected by the user terminal (app) using the method of the present invention.

The test was carried out on 21 people, and the test time for each test ranged from 10 seconds to several minutes, and the number of breaths (number of cycles) also varied, as shown in Table 1.

Table 1 method practical application test

For each test, the difference ΔRR is obtained by subtracting the _actual respiratory rate RR detected by the method from the actual respiratory rate RR, that _is , ΔRR= _RRactual -RR _detection , and then _{the actual} RR is the horizontal axis and ΔRR is the vertical axis. The error distribution diagram of the detection method for respiratory frequency is shown in Figure 7.

Statistics based on Table 1, the maximum absolute error is -0.013Hz, the conversion unit is -0.78BPM; the average absolute error is the arithmetic mean of all ΔRR is 0.00118Hz, the conversion unit is 0.0708BPM . From this embodiment, it can be seen that the execution efficiency of the method is high.

The above are only preferred embodiments of the application, and are not intended to limit the application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the application shall be included in the protection of the application. within range. Other structures and principles are the same as those of the prior art, and will not be repeated here.

Claims (3)

1. A respiratory rate detection method based on an air flow sensor is characterized in that: comprises the steps of,

(1) performing fixed integration on the curve of the respiratory wave, setting a threshold delta, and if the fixed integration value of a period of time sequence signal over a period of time satisfies the formula (1):

dividing the segment of the signal into a breathing cycle; wherein delta in the above formula is a threshold value, S (t) is a respiratory wave, and is a function of time t, x is a starting point of scanning, and L is an upper limit length of scanning;

(2) scanning sampling points sequentially from a starting scanning point, wherein a point closest to the starting point is to be found as an end point of scanning, and the point is preferentially satisfied at or close to a zero point of respiratory waves;

(3) according to the amplitude condition of noise near the t axis in the coordinate axis, adding filtering conditions: using the upper threshold and the lower threshold so that signals falling within the threshold range are treated as 0; wherein thr _h >0，thr _l <0；

(4) After dividing the breathing cycle, based on the number of the breathing cycles in a certain period of time, the average breathing frequency in the period of time can be calculated, and the formula is as follows:

wherein R is the respiratory rate, T is a certain period of time, and n is the number of respiratory cycles monitored in the period of time.

2. The method of claim 1, wherein: the timing signal is a discrete digital signal sampled by the ADC, and therefore, the discrete form of equation (1) is as follows:

where i represents a certain sampling point, S (i) is a discrete function with respect to the sampling point i, and fs is the sampling frequency.

3. Use of a respiratory rate detection method based on an air flow sensor, employing an algorithm according to any one of claims 1 or 2, characterized in that: in particular comprising the following steps of the method,

s1, reading in a section of discrete time series data to be detected, sampling frequency and upper threshold limit thr for eliminating noise influence _h Upper threshold thr _l A threshold delta used for limiting an integral upper limit value, and creating and initializing a null array T with the same length as the time sequence to be detected, wherein the T is used as an array to be detected for backup;

s2, preprocessing time sequence data: traversing the time sequence data to be detected under the condition that the time sequence data to be detected is not empty, if the data value S (i) at a certain sampling point>thr _h Assigning the value to the corresponding position of the T array; if S (i)<thr _l Assigning the value to the corresponding position of the T array; otherwise, assigning 0 to the corresponding position of the T array;

s3, creating and initializing an array I for storing the segmentation points for segmenting the respiratory cycle; setting the upper limit of the time length of each scanning to be 30s; creating and initializing an array A for storing discrete constant integration values and an array D for describing the variation of the integration values;

s4, if the total length of the data is smaller than 2, ending and returning to 0;

s5, traversing the array T to be detected, and taking the next data point of the first data as initial scanning data;

s6, starting one-time scanning: the upper limit of each time length is 30s, if the time length of the time sequence data to be detected is less than 30s, the maximum time length of the time sequence data is the upper limit of the window length of each scanning;

s7, calculating the discrete fixed integral value of the signal value from the initial sampling point to the current sampling point of the scanning every time when one sampling point is traversed, storing the discrete fixed integral value into an array D, and entering the next step after the step 7 is cycled to the upper limit of the duration of the scanning;

s8, reading and traversing the array A which stores the discrete constant value of the scanning in pairs, calculating the difference value of the array A and the array D which describes the change of the integral value, namely, di=Ai+1-Ai;

s9, recording the first sampling point i of the scanning ₀ The signal value of (i) is S (i ₀ ) The method comprises the steps of carrying out a first treatment on the surface of the If S (i) ₀ ) If not less than 0, traversing the group D, and detecting D [ i+1]]≥D[i]Taking the sampling point which causes the situation as a temporary cut-off point; if S (i) ₀ )<0, traversing the array D to detect Di+1]≤D[i]Taking the sampling point which causes the situation as a temporary cut-off point;

s10, acquiring a temporary cut-off point of the previous step, calculating a fixed integral value of a signal from a starting sampling point to the temporary cut-off point of the current scanning, if the fixed integral value is not more than a set threshold delta, judging that the temporary cut-off point is effective and is used as a time division point for dividing the respiratory cycle, storing the time division point into an array I for storing the division point for dividing the respiratory cycle, and taking the next point of the sampling point as a starting point of the next scanning;

s11, obtaining a scanning starting point of the previous step, starting a new scanning, and repeating the steps 6-11;

s12, circularly scanning, and repeating the steps 6-12 until the data to be detected are scanned;

s13, returning an array I for storing the division points for dividing the respiratory cycle, wherein each value in the array I is the time point for dividing the respiratory cycle, the time period of each respiratory cycle is obtained through the division points, and meanwhile, the number of the division points is the respiratory cycle number of the respiratory wave of the section; dividing the total duration of the time sequence data by the number of the respiratory cycles to obtain the average respiratory frequency, namely realizing respiratory frequency detection based on the air flow sensor.

Images