Apnea identification method, device and equipment and readable storage medium

Abstract

The invention discloses an apnea identification method, device and equipment and a computer readable storage medium, and relates to the technical field of radar signal processing, and the method comprises the steps: obtaining all frames of echo signals received after a radar signal is transmitted; converting each frame of echo signal to obtain a first complex signal sequence corresponding to each frame of echo signal; extracting a complex signal corresponding to a distance unit where the target object is located from each group of first complex signal sequences, and obtaining a slow-time-dimension second complex signal sequence according to the complex signals extracted from the plurality of groups of first complex signal sequences; complex signal features are extracted according to the second complex signal sequence, and an apnea recognition result of the target object is recognized according to the extracted complex signal features. According to the invention, the apnea identification accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of radar signal processing, and particularly to a method, device, equipment and computer-readable storage medium for apnea recognition. Background Art

[0002] Apnea is a type of obstructive sleep apnea or central sleep apnea that is separated from sleep. During sleep, patients repeatedly experience a stop in the respiratory airflow. If the duration exceeds 10 seconds or the airflow is less than 20% of the normal value, it is considered sleep apnea. Since apnea occurs during sleep and is not easily detectable, if long-term apnea is not discovered and effectively treated, a series of diseases will occur. Therefore, detecting apnea is crucial for human health.

[0003] In related technologies, radar signals are transmitted, and slow-time differential phase signal analysis is performed based on the echo signals reflected by the human body to identify whether there is sleep apnea. However, when analyzing based on differential phase signals, it is easily affected by factors such as direct current and phase wrapping, resulting in incorrect identification, that is, the accuracy of apnea recognition is relatively low. Summary of the Invention

[0004] The main objective of the present invention is to provide a method, device, equipment and computer-readable storage medium for apnea recognition, aiming to propose a solution for recognizing apnea based on complex signals and improving the accuracy of apnea recognition.

[0005] To achieve the above objective, the present invention provides an apnea recognition method, which includes the following steps:

[0006] Obtain each frame of echo signal received after transmitting a radar signal;

[0007] Convert each frame of the echo signal respectively to obtain a first complex signal sequence corresponding to each frame of the echo signal, where the first complex signal sequence includes complex signals corresponding to a preset number of range cells;

[0008] Extract the complex signals corresponding to the range cells where the target object is located from each group of the first complex signal sequences, and obtain a second complex signal sequence in the slow-time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences;

[0009] Extract complex signal features according to the second complex signal sequence, and identify the apnea recognition result of the target object according to the extracted complex signal features, where the complex signal features corresponding to normal breathing behavior and apnea behavior are different.

[0010] Optionally, the step of extracting complex signal features according to the second complex signal sequence and identifying the apnea recognition result of the target object according to the extracted complex signal features includes:

[0011] Extract complex signal features according to the second complex signal sequence, input the extracted complex signal features into a pre-trained classification model for classification, and obtain the apnea recognition result of the target object;

[0012] Among them, the classification model is pre-trained with the complex signal features corresponding to the breathing behavior as the input data and the type of breathing behavior as the classification label. The types of breathing behavior include normal breathing behavior and apnea behavior.

[0013] Optionally, the step of extracting complex signal features according to the second complex signal sequence, inputting the extracted complex signal features into a pre-trained classification model for classification, and obtaining the apnea recognition result of the target object includes:

[0014] Draw a constellation diagram according to the second complex signal sequence, and the two coordinate axes of the constellation diagram are the real part and the imaginary part of the complex signal respectively;

[0015] Use the constellation diagram as the extracted complex signal features and input them into a pre-trained classification model for classification to obtain the apnea recognition result of the target object.

[0016] Optionally, the extracted complex signal features include the standard deviation of the second complex signal sequence. The step of extracting complex signal features according to the second complex signal sequence and identifying the apnea recognition result of the target object according to the extracted complex signal features includes:

[0017] Calculate the standard deviation of the second complex signal sequence;

[0018] If it is detected that the standard deviation is less than the preset threshold, obtain the apnea recognition result that the target object has apnea behavior;

[0019] If it is detected that the standard deviation is greater than or equal to the preset threshold, obtain the apnea recognition result that the target object performs normal breathing behavior.

[0020] Optionally, the step of obtaining the second complex signal sequence of the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences includes:

[0021] On the complex signal sequence composed of the complex signals extracted from each group of the first complex signal sequences, obtain second complex signal sequences of various lengths by adding time windows of different lengths;

[0022] The step of extracting complex signal features according to the second complex signal sequence and identifying the apnea recognition result of the target object according to the extracted complex signal features further includes:

[0023] Determine the standard deviation threshold corresponding to the length of the second complex signal sequence from the preset standard deviation thresholds corresponding to different lengths as the preset threshold.

[0024] Optionally, the step of obtaining the second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences includes:

[0025] After receiving a new frame of the echo signal and extracting a complex signal each time, if the number of complex signals in the third complex signal sequence is equal to the preset quantity, delete the earliest added complex signal in the third complex signal sequence, and add the extracted complex signal to the third complex signal sequence; otherwise, if the number of complex signals in the third complex signal sequence is less than the preset quantity, add the extracted complex signal to the third complex signal sequence, where the number of complex signals in the third complex signal sequence is zero when it is initialized.

[0026] After adding a new complex signal each time to make the number of complex signals in the third complex signal sequence reach the preset quantity, determine a group of the second complex signal sequences according to the complex signals in the third complex signal sequence.

[0027] Optionally, before the step of respectively extracting the complex signals corresponding to the distance units where the target object is located from each group of the first complex signal sequences, it further includes:

[0028] Calculate the amplitudes according to the respective complex signals in the fourth complex signal sequence to obtain the amplitude sequence corresponding to the fourth complex signal sequence, where the amplitude sequence includes the amplitudes corresponding to the multiple distance units respectively, and the fourth complex signal sequence is one of the groups of the first complex signal sequences.

[0029] Use the distance unit corresponding to the peak value in the amplitude sequence as the target distance unit corresponding to the fourth complex signal sequence.

[0030] Determine the distance unit where the target object is located according to the target distance units corresponding to one or more groups of the fourth complex signal sequences.

[0031] To achieve the above object, the present invention further provides an apnea recognition device, and the device includes:

[0032] An acquisition module, configured to acquire each frame of echo signal received after transmitting a radar signal;

[0033] A conversion module, configured to perform conversion on each frame of the echo signals respectively to obtain a first complex signal sequence corresponding to each frame of the echo signals, wherein the first complex signal sequence includes complex signals corresponding to a plurality of preset range cells.

[0034] An extraction module, configured to extract the complex signals corresponding to the range cells where the target object is located from each group of the first complex signal sequences respectively, and obtain a second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences.

[0035] An identification module, configured to extract complex signal features according to the second complex signal sequence, and identify a respiratory arrest identification result of the target object according to the extracted complex signal features, wherein the complex signal features corresponding to normal breathing behavior and respiratory arrest behavior are different.

[0036] To achieve the above object, the present invention further provides a respiratory arrest identification device, which includes: a memory, a processor, and a respiratory arrest identification program stored on the memory and executable on the processor. When the respiratory arrest identification program is executed by the processor, the steps of the above-mentioned respiratory arrest identification method are implemented.

[0037] In addition, to achieve the above object, the present invention further proposes a computer-readable storage medium, on which a respiratory arrest identification program is stored. When the respiratory arrest identification program is executed by a processor, the steps of the above-mentioned respiratory arrest identification method are implemented.

[0038] In an embodiment of the present invention, by acquiring each frame of echo signal received after transmitting a radar signal, each frame of echo signal is respectively converted to obtain a first complex signal sequence corresponding to each frame of echo signal. The first complex signal sequence includes complex signals corresponding to a plurality of preset range cells, so as to obtain the basic data for apnea recognition. Then, by respectively extracting the complex signals corresponding to the range cells where the target object is located from each group of first complex signal sequences, a second complex signal sequence in the slow-time dimension is obtained according to the complex signals extracted from multiple groups of first complex signal sequences. Since the second complex signal sequence in the slow-time dimension reflects the characteristics of the breathing behavior of the target object, complex signal features are extracted according to the second complex signal sequence. Moreover, the complex signal features corresponding to normal breathing behavior and apnea behavior are different. Therefore, the apnea recognition result of the target object can be recognized according to the extracted complex signal features. In the related art, when analyzing apnea based on the differential phase signal in the slow-time dimension, it is easily affected by factors such as direct current and phase wrapping, resulting in incorrect recognition. However, in this embodiment, complex signal features are extracted according to the complex signal sequence in the slow-time dimension, and whether there is an apnea behavior is recognized according to the complex signal features, without being affected by factors such as direct current and phase wrapping, thus improving the accuracy of apnea recognition and reducing false detection as much as possible. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic flowchart of an embodiment of the apnea recognition method of the present invention;

[0040] Figure 2 It is a constellation diagram of slow-time complex signal sequences corresponding to normal breathing behavior and abnormal breathing behavior respectively according to an embodiment of the present invention;

[0041] Figure 3 It is a schematic diagram comparing the standard deviations of slow-time complex signal sequences corresponding to normal breathing behavior and abnormal breathing behavior according to an embodiment of the present invention;

[0042] Figure 4 It is a schematic flowchart of an apnea recognition process according to an embodiment of the present invention;

[0043] Figure 5 It is a schematic diagram of functional modules of a preferred embodiment of the apnea recognition device according to an embodiment of the present invention;

[0044] Figure 6 It is a schematic diagram of the structure of the hardware operating environment involved in the solution of an embodiment of the present invention.

[0045] The implementation, functional features, and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] It should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0047] Referring to Figure 1 , Figure 1 is a schematic flowchart of the first embodiment of the apnea recognition method of the present invention.

[0048] The embodiments of the present invention provide an embodiment of the apnea recognition method. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than here. In this embodiment, the execution subject of the apnea recognition method may be an apnea recognition device, and the apnea recognition device may be implemented by devices such as a computer, a smart phone, a server, etc., which are not limited in this embodiment. In this embodiment, for the convenience of description, the following will be described with the recognition device as the execution subject. In this embodiment, the apnea recognition method includes the following steps S10 to S40:

[0049] Step S10, obtain each frame of echo signal received after transmitting the radar signal.

[0050] After the radar signal is transmitted, it is reflected by surrounding objects to form an echo signal and is received. The radar signals are transmitted frame by frame in chronological order, and the echo signals formed by the reflection of each frame of radar signals are also received frame by frame in chronological order.

[0051] In this embodiment, the radar signal is transmitted for the target object, and after being reflected by the target object, an echo signal is formed and received. The target object may be a user who needs to be recognized for the presence of apnea behavior.

[0052] In the specific implementation, there are many ways for the recognition device to obtain each frame of echo signal, which are not limited in this embodiment.

[0053] In a feasible implementation, the recognition device can dynamically obtain each frame of echo signal and dynamically perform apnea recognition based on the obtained each frame of echo signal, that is, it is equivalent to continuously receiving new echo signals and continuously performing apnea recognition according to the new echo signals.

[0054] In a feasible implementation, the radar signal can be transmitted by a radar transmitting device, and the echo signal is received by a radar receiving device. The radar transmitting device and the radar receiving device can be integrated in the recognition device or can be independent of the recognition device and be connected to the recognition device through wired or wireless connection methods.

[0055] In a feasible embodiment, the radar transmitting device may transmit radar pulse signals at a certain pulse emission frequency. For a single pulse, the radar receiving device may sample at a certain sampling frequency to obtain respective echo sampling data within the single pulse. A single-frame echo signal may include echo sampling data within a single pulse or echo sampling data within a continuous plurality of pulses. The frame length may be set as needed and is not limited herein. The echo sampling data may be the original echo data obtained by sampling or the echo data after preprocessing, which is not limited herein.

[0056] Step S20: Convert each frame of the echo signals respectively to obtain respective first complex signal sequences corresponding to each frame of the echo signals, where the first complex signal sequence includes complex signals corresponding to a preset plurality of range cells.

[0057] Taking a single-frame echo signal as an example, the recognition device may convert the frame of the echo signal to obtain a complex signal sequence corresponding to the frame of the echo signal (hereinafter referred to as the first complex signal sequence). That is, each frame of the echo signal can be converted to obtain a corresponding first complex signal sequence.

[0058] There are many ways to convert the echo signal to obtain the first complex signal sequence, which is not limited in this embodiment. For example, the fast Fourier transform method can be used to convert the echo signal to obtain the first complex signal sequence.

[0059] It should be noted that the first complex signal sequence is a sequence composed of complex signals obtained by performing a time-domain to frequency-domain conversion on the echo signal in the range dimension (or also referred to as the fast time dimension), which includes complex signals corresponding to a preset plurality of range cells. The range cell corresponds to a frequency point in the frequency domain. The number of range cells, that is, the number of frequency points, can be set in advance. The complex signal includes a real part and an imaginary part.

[0060] In the specific implementation manner, when the single-frame echo signal includes the echo sampling data within a single pulse, the single-frame echo signal can be directly converted to obtain the first complex signal sequence; when the single-frame echo signal includes the echo sampling data within a continuous plurality of pulses, the echo sampling data within a single pulse is arranged in the order of sampling time. The echo sampling data at the same position within a plurality of pulses can be averaged to obtain the average echo sampling data, and the plurality of average echo sampling data is converted to obtain the first complex signal sequence. Suppose that the echo sampling data within a single pulse includes N sampling points, and the single-frame echo signal includes the echo sampling data within continuous M pulses. Then, the echo sampling data at the i-th sampling point within M pulses can be averaged to obtain the i-th average echo sampling data, where i = 1, 2,..., N. After obtaining N average echo sampling data, the N average echo sampling data is converted to obtain the first complex signal sequence.

[0061] Step S30: Respectively extract the complex signals corresponding to the range bins where the target object is located from each group of the first complex signal sequences, and obtain the second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences.

[0062] It should be noted that the range bin where the target object is located reflects the distance between the target object and the radar transceiver. The signal reflecting the breathing behavior of the target object detected by the radar is reflected in the complex signal corresponding to this range bin. Therefore, by extracting the complex signal corresponding to the range bin where the target object is located from the first complex signal sequence, this complex signal can reflect the breathing behavior of the target object.

[0063] In this embodiment, the recognition device respectively extracts the complex signals corresponding to the range bins where the target object is located from each group of the first complex signal sequences, and forms the complex signals extracted from multiple groups of the first complex signal sequences into a group of complex signal sequences (hereinafter referred to as the second complex signal sequence for distinction). It should be noted that the second complex signal sequence is a complex signal sequence in the slow time dimension, and the slow time dimension is relative to the fast time dimension.

[0064] The number of complex signals included in the second complex signal sequence, that is, the number of groups of the first complex signal sequences based on which the second complex signal sequence is extracted, is not limited in this embodiment and can be set in advance according to needs. The time interval for taking a group of the second complex signal sequences is also not limited in this embodiment and can also be set in advance according to needs.

[0065] In a feasible implementation, the second complex signal sequence can be obtained by adding a time window to the complex signal sequence composed of the complex signals extracted from each group of the first complex signal sequences. The complex signals extracted from each group of the first complex signal sequences can be regarded as a group of complex signal sequences in the slow time dimension. As time increases, that is, as new echo signals are continuously received, this complex signal sequence grows. By adding a time window of a certain length to this complex signal sequence, a group of second complex signal sequences can be obtained. Sliding this time window in the sequence growth direction can continuously obtain new second complex signal sequences of this length. The step size of the time window sliding can be set as needed. For example, it can be slid once after each new frame or multiple frames of echo signals are received and the corresponding complex signals are extracted.

[0066] For example, in a feasible implementation, in the case of sliding once after each new frame of echo signal is received and the corresponding complex signal is extracted, the step of obtaining the second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences in step S30 includes S301 to S302:

[0067] Step S301, after each new frame of the echo signal is received and a complex signal is extracted, if the number of complex signals in the third complex signal sequence is equal to the preset quantity, the earliest added complex signal in the third complex signal sequence is deleted, and the extracted complex signal is added to the third complex signal sequence. Otherwise, if the number of complex signals in the third complex signal sequence is less than the preset quantity, the extracted complex signal is added to the third complex signal sequence, where the number of complex signals in the third complex signal sequence is zero when it is initialized.

[0068] Before the first apnea recognition is performed after the recognition device is powered on, a third complex signal sequence can be initialized. When initialized, the third complex signal sequence is empty, that is, the number of complex signals in it is zero. The preset quantity can be set in advance as needed and is not limited in this implementation.

[0069] Step S302, after each new complex signal is added and the number of complex signals in the third complex signal sequence reaches the preset quantity, a group of the second complex signal sequences is determined according to the complex signals in the third complex signal sequence.

[0070] Exemplarily, it is assumed that the preset number is set to 30; each time a new frame of echo signal is received and a set of first complex signal sequences is obtained through conversion, after extracting the complex signal corresponding to the range cell where the target object is located based on this set of first complex signal sequences, if the number of complex signals in the third complex signal sequence is less than 30, it indicates that apnea recognition has not been performed before. In this case, the extracted complex signal is directly added to the third complex signal sequence; if the number of complex signals in the third complex signal sequence is equal to 30, it indicates that apnea recognition has been performed before. In this case, the earliest added complex signal in the third complex signal sequence is deleted, and the latest extracted complex signal is added to the third complex signal sequence. At this time, the number of complex signals in the third complex signal sequence reaches 30 again. Then, a new apnea recognition is performed based on the complex signals in the current third complex signal sequence, that is, a set of second complex signal sequences is determined according to the complex signals in the current second complex signal sequence, and an apnea recognition is performed according to the second complex signal sequence; thus, each time a new frame of echo signal is received, an apnea recognition is performed, which is equivalent to that the length of the time window is fixed at 30 frames, and the sliding step of the time window slides once after each new frame of echo signal is received and the corresponding complex signal is extracted.

[0071] In the specific implementation manner, the range cell where the target object is located can be obtained in advance, and there are many ways to obtain it, which are not limited in this embodiment. For example, in a feasible implementation manner, before step S30, steps S50 to S70 are further included:

[0072] Step S50, calculating the amplitude according to each complex signal in the fourth complex signal sequence to obtain the amplitude sequence corresponding to the fourth complex signal sequence, where the amplitude sequence includes the amplitudes corresponding to the multiple range cells respectively, and the fourth complex signal sequence is one of the groups of the first complex signal sequences.

[0073] One or more groups of first complex signal sequences can be used to determine the range cell where the target object is located. Which group or groups of first complex signal sequences are selected to determine the range cell where the target object is located is not limited in this implementation manner. For example, each group of first complex signal sequences obtained within a certain period of time after the recognition device is powered on can be used to determine the range cell where the target object is located, and then the second complex signal sequence is extracted based on the range cell where the target object is located determined by the first complex signal sequences obtained thereafter, and then apnea recognition is performed.

[0074] In this implementation manner, the first complex signal sequence used to determine the range cell where the target object is located is called the fourth complex signal sequence for distinction.

[0075] Taking a set of fourth complex signal sequences as an example, the recognition device can calculate the amplitude according to each complex signal in the fourth complex signal sequence, and each complex signal corresponding to each distance unit can calculate an amplitude, so a set of amplitude sequences can be obtained, including the amplitudes corresponding to each distance unit. It should be noted that there are many ways to calculate the amplitude according to the complex signal, which is not limited in this embodiment.

[0076] Step S60: taking the distance unit corresponding to the peak value in the amplitude sequence as the target distance unit corresponding to the fourth complex signal sequence.

[0077] Step S70: determining the distance unit where the target object is located according to one or more groups of target distance units corresponding to the fourth complex signal sequences.

[0078] In the case where the distance unit where the target object is located is determined according to a set of fourth complex signal sequences, the recognition device can use the distance unit corresponding to the peak value in the amplitude sequence corresponding to the set of fourth complex signal sequences as the distance unit where the target object is located. In the case where the distance unit where the target object is located is determined according to multiple sets of fourth complex signal sequences, the distance unit corresponding to the peak value in the amplitude sequence corresponding to the fourth complex signal sequence is called the target distance unit corresponding to the set of fourth complex signal sequences, so as to distinguish them. Then, the recognition device can obtain the target distance units corresponding to the multiple sets of fourth complex signal sequences respectively, and determine the distance unit where the target object is located according to the multiple target distance units. The principle of determining the distance unit where the target object is located according to the multiple target distance units is that the target object is generally identified when apnea is stationary, such as when apnea is identified in a sleeping state, and the distance between the target object and the radar transceiver is unchanged, so the target distance units determined based on the multiple sets of fourth complex signal sequences should all be the same, that is, when multiple target distance units are the same, the target distance unit can be determined as the distance unit where the target object is located. Based on this principle, different implementation methods can be adopted according to actual conditions. For example, if multiple target distance units are different or have large differences, multiple groups of fourth complex signal sequences can be re-acquired to detect the distance units where the target objects are located again; for another example, there may be some errors in the actual situation, so the error tolerance range can also be set as needed. For example, only a few of the multiple target distance units are different from the vast majority of other target distance units, and the differences are not large, then the vast majority of the same target distance units can be determined as the distance units where the target objects are located.

[0079] It should be noted that the method of determining the range cell where the target object is located based on a set of fourth complex signal sequences and the method of determining the range cell where the target object is located based on multiple sets of fourth complex signal sequences each have their own advantages and disadvantages, and can be selected according to the needs in the actual scenario.

[0080] Step S40: Extract complex signal features according to the second complex signal sequence, and identify the apnea recognition result of the target object according to the extracted complex signal features, where the complex signal features corresponding to normal breathing behavior and apnea behavior are different.

[0081] After obtaining one or more sets of second complex signal sequences, the recognition device can extract complex signal features according to the one or more sets of second complex signal sequences. In this embodiment, the specific data form of the complex signal features is not limited, as long as it can distinguish between normal breathing behavior and apnea behavior. By obtaining the second complex signal sequences by the above-mentioned steps S10 - S30 for the user under normal breathing behavior and apnea behavior respectively, and comparing the second complex signal sequences of the two, it is found that there are significant differences between the second complex signal sequences of the two. For example, the second complex signal sequences are plotted in the form of a constellation diagram, as Figure 2 shown. The abscissa is the real part of the complex signal, and the ordinate is the imaginary part of the complex signal. The constellation diagram on the left is for normal breathing behavior, and the constellation diagram on the right is for apnea behavior (note: the scales of the abscissa and ordinate of the two constellation diagrams on the left and right are different). It can be seen that there are obvious differences in the distribution characteristics between the constellation diagram of normal breathing behavior and the constellation diagram of apnea behavior. Therefore, in this embodiment, by extracting complex signal features from the second complex signal sequence, since the complex signal features of normal breathing behavior and apnea behavior are different, it is possible to identify and determine whether the target object has apnea behavior according to the complex signal features extracted from the second complex signal sequence.

[0082] Moreover, in the related art, apnea analysis based on the differential phase signal in the slow time dimension is easily affected by factors such as direct current and phase wrapping, resulting in incorrect recognition. However, in this embodiment, complex signal features are extracted according to the complex signal sequence in the slow time dimension, and whether there is apnea behavior is identified according to the complex signal features, which will not be affected by factors such as direct current and phase wrapping and lead to incorrect recognition, thus improving the accuracy of apnea recognition.

[0083] Based on the above first embodiment, a second embodiment of the apnea recognition method of the present invention is proposed. In this embodiment, the step S40 includes S401 - S402:

[0084] Step S401: Extract complex signal features according to the second complex signal sequence, and input the extracted complex signal features into a pre-trained classification model for classification to obtain the apnea recognition result of the target object. Among them, the classification model is pre-trained with the complex signal features corresponding to the breathing behavior as the input data and the type of breathing behavior as the classification label. The types of breathing behavior include normal breathing behavior and apnea behavior.

[0085] In this embodiment, a classification model can be used to classify based on complex signal features. Utilizing the powerful classification ability of the classification model, the apnea recognition result of the target object can be determined more accurately based on complex signal features, thereby further improving the accuracy of apnea recognition.

[0086] The structure of the classification model can be implemented using the classification model structure in related technologies, and no limitation is imposed here.

[0087] The classification model can be pre-trained so that the classification model can accurately classify the apnea recognition result based on complex signal features. The training process of the classification model can be carried out in the recognition device, or after being trained in other devices, the trained classification model can be deployed in the recognition device. The training process can use the complex signal features corresponding to the breathing behavior as the input data and the type of this breathing behavior as the classification label. The complex signal features corresponding to this breathing behavior are the complex signal features obtained by using the methods such as steps S10 - S40 when the user performs this breathing behavior. The type of this breathing behavior is known, and the types of breathing behavior can include normal breathing behavior and apnea behavior. For example, in a feasible implementation manner, the methods such as steps S10 - S40 can be used to collect multiple groups of second complex signal sequences for one or more users in normal breathing behavior, extract multiple complex signal features based on the multiple groups of second complex signal sequences, and each complex signal feature and the classification label of "normal breathing behavior" form a sample data to obtain multiple sample data (referred to as "positive sample data" for distinction); the methods such as steps S10 - S40 can be used to collect multiple groups of second complex signal sequences for one or more users with apnea behavior, extract multiple complex signal features based on the multiple groups of second complex signal sequences, and each complex signal feature and the classification label of "apnea behavior" form a sample data to obtain multiple sample data (referred to as "negative sample data" for distinction); multiple positive sample data and multiple negative sample data are combined to form a training data set, and the training data set is used to train the classification model. The training method of the classification model can refer to the training method of machine learning models in related technologies, which will not be elaborated here.

[0088] Based on the trained classification model, the recognition device can input the complex signal features obtained for the target object in the manner as in steps S10 to S40 into the classification model for classification to obtain the apnea recognition result of the target object. In a specific implementation manner, the complex signal features input into the classification model can be, for example, the second complex signal sequence itself, the standard deviation of the second complex signal sequence, or a standard deviation sequence composed of the standard deviations of multiple groups of the second complex signal sequences, etc., which are not limited in this embodiment.

[0089] In a feasible implementation manner, step S401 includes S4011 to S4012:

[0090] Step S4011, draw a constellation diagram according to the second complex signal sequence, where the two coordinate axes of the constellation diagram are the real part and the imaginary part of the complex signal respectively.

[0091] As Figure 2 shown, the distribution characteristics of the constellation diagrams of the complex signal sequences corresponding to normal breathing behavior and apnea behavior are significantly different. In this embodiment, the second complex signal sequence can be drawn into a constellation diagram, and the constellation diagram is used as the extracted complex signal feature.

[0092] Step S4022, input the constellation diagram as the extracted complex signal feature into a pre-trained classification model for classification to obtain the apnea recognition result of the target object.

[0093] The structure of the classification model can be implemented by adopting the structure of an image classification model in related technologies. That is, the classification model can be an image classification model. Input the constellation diagram into the pre-trained image classification model for image classification to obtain the apnea recognition result of the target object.

[0094] Based on the above first and / or second embodiments, a third embodiment of the apnea recognition method of the present invention is proposed. In this embodiment, step S40 includes S402 to S404:

[0095] Step S402, calculate the standard deviation of the second complex signal sequence.

[0096] As Figure 2 shown, in the constellation diagrams of the complex signal sequences corresponding to normal breathing behavior and apnea behavior, it can be clearly seen that the complex signals in the constellation diagram corresponding to normal breathing behavior are concentrated, while the complex signals in the constellation diagram corresponding to apnea behavior are quite different. The standard deviations of the two show an obvious distinction. As Figure 3 shown, it can also be seen that there are significant differences in the standard deviations of the complex signal sequences corresponding to normal breathing behavior and apnea behavior. Figure 3The standard deviation of the middle arc length is the standard deviation calculated based on the second complex signal sequence. In this embodiment, the standard deviation of the second complex signal sequence can be calculated, and the standard deviation can be used as the extracted complex signal feature.

[0097] Step S403, if it is detected that the standard deviation is less than the preset threshold, a apnea recognition result that the target object has an apnea behavior is obtained.

[0098] Step S404, if it is detected that the standard deviation is greater than or equal to the preset threshold, a apnea recognition result that the target object has a normal breathing behavior is obtained.

[0099] In advance, the standard deviation threshold of the complex signal sequence corresponding to the normal breathing behavior and the apnea behavior can be obtained through experimental tests and pre-configured in the recognition device.

[0100] After the recognition device calculates the standard deviation of the second complex signal sequence, it can obtain the pre-configured standard deviation threshold, that is, the preset threshold, and compare the standard deviation with the preset threshold to determine the apnea recognition result of the target object according to the comparison result.

[0101] In a feasible embodiment, the apnea recognition can be performed according to the Figure 4 shown process. Among them, data input refers to obtaining the echo signal; target detection is to determine the range cell where the target object is located; extracting the slow-time complex signal is to extract the complex signal corresponding to the range cell where the target object is located from each group of first complex signal sequences, and obtain the second complex signal sequence based on the complex signals extracted from multiple groups of first complex signal sequences; calculating the constellation diagram standard deviation is to calculate the standard deviation of the second complex signal sequence.

[0102] In a feasible embodiment, the step S30 of obtaining the second complex signal sequence in the slow-time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences includes S303, and step S40 further includes S405.

[0103] Step S303, on the complex signal sequence composed of the complex signals extracted from each group of the first complex signal sequences, multiple-length second complex signal sequences are obtained by adding time windows of different lengths.

[0104] Since the specific manifestations of the apnea behaviors of different users may be different, in order to cover different users or different apnea characteristics shown by the same user at different times, multiple different-length second complex signal sequences can be obtained, and the apnea recognition is respectively performed based on the second complex signal sequences of different lengths to improve the recognition accuracy of the apnea behavior of the user.

[0105] If apnea behavior is identified based on the second complex signal sequence of any length among multiple different lengths of second complex signal sequences, it can be determined that the target object has apnea behavior.

[0106] The way to obtain multiple different lengths of second complex signal sequences can be to obtain them by adding time windows of different lengths. The complex signals extracted from each group of first complex signal sequences can be regarded as a group of complex signal sequences. As time increases, that is, as new echo signals are continuously received, this complex signal sequence grows. By adding a time window of a certain length to this complex signal sequence, a group of second complex signal sequences can be obtained. Sliding this time window in the sequence growth direction can continuously obtain new second complex signal sequences of this length. In the same way, by adding multiple time windows of different lengths simultaneously, second complex signal sequences of different lengths can be obtained.

[0107] Step S405: Determine the standard deviation threshold corresponding to the length of the second complex signal sequence from the preset standard deviation thresholds corresponding to different lengths as the preset threshold.

[0108] Since the time window lengths are different, that is, since the lengths of the second complex signal sequences are different, the corresponding standard deviation thresholds will also be different. The standard deviation thresholds corresponding to time windows of different lengths can be pre-tested through experiments. For a certain second complex signal sequence, when performing apnea recognition based on this second complex signal sequence, the standard deviation threshold corresponding to the length of this second complex signal sequence can be obtained as the preset threshold and compared with the standard deviation of this second complex signal sequence. If the standard deviation is less than this preset threshold, it is determined that the target object has apnea behavior; otherwise, it is determined that the target object is performing normal breathing behavior.

[0109] In addition, an embodiment of the present invention also proposes an apnea recognition device. Refer to Figure 5 , the device includes:

[0110] An acquisition module 10, configured to acquire each frame of echo signal received after transmitting a radar signal;

[0111] A conversion module 20, configured to respectively convert each frame of the echo signal to obtain a first complex signal sequence corresponding to each frame of the echo signal, where the first complex signal sequence includes complex signals corresponding to a preset plurality of range cells;

[0112] An extraction module 30, configured to respectively extract the complex signals corresponding to the range cell where the target object is located from each group of the first complex signal sequences, and obtain a second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences;

[0113] An identification module 40, configured to extract complex signal features according to the second complex signal sequence, and identify an apnea identification result of the target object according to the extracted complex signal features, wherein the complex signal features corresponding to normal breathing behavior and apnea behavior are different.

[0114] In a feasible implementation manner, the identification module 40 is further configured to:

[0115] Extract complex signal features according to the second complex signal sequence, input the extracted complex signal features into a pre-trained classification model for classification, and obtain the apnea identification result of the target object;

[0116] Wherein, the classification model is pre-trained with the complex signal features corresponding to breathing behavior as input data and the type of breathing behavior as classification labels, and the types of breathing behavior include normal breathing behavior and apnea behavior.

[0117] In a feasible implementation manner, the identification module 40 is further configured to:

[0118] Draw a constellation diagram according to the second complex signal sequence, where the two coordinate axes of the constellation diagram are the real part and the imaginary part of the complex signal;

[0119] Use the constellation diagram as the extracted complex signal features and input them into a pre-trained classification model for classification to obtain the apnea identification result of the target object.

[0120] In a feasible implementation manner, the extracted complex signal features include the standard deviation of the second complex signal sequence, and the identification module 40 is further configured to:

[0121] Calculate the standard deviation of the second complex signal sequence;

[0122] If it is detected that the standard deviation is less than a preset threshold, obtain an apnea identification result indicating that the target object has apnea behavior;

[0123] If it is detected that the standard deviation is greater than or equal to the preset threshold, obtain an apnea identification result indicating that the target object is performing normal breathing behavior.

[0124] In a feasible implementation manner, the extraction module 30 is further configured to:

[0125] On the complex signal sequence composed of complex signals extracted from each group of the first complex signal sequences, obtain second complex signal sequences of various lengths by adding time windows of different lengths;

[0126] The identification module 40 is further configured to:

[0127] Determine the standard deviation threshold corresponding to the length of the second complex signal sequence from the preset standard deviation thresholds corresponding to different lengths as the preset threshold.

[0128] In a feasible implementation, the extraction module 30 is further configured to:

[0129] After receiving a new frame of the echo signal and extracting a complex signal each time, if the number of complex signals in the third complex signal sequence is equal to the preset quantity, delete the earliest added complex signal in the third complex signal sequence, and add the extracted complex signal to the third complex signal sequence; otherwise, if the number of complex signals in the third complex signal sequence is less than the preset quantity, add the extracted complex signal to the third complex signal sequence, where the number of complex signals in the third complex signal sequence is zero when it is initialized.

[0130] After adding a new complex signal each time to make the number of complex signals in the third complex signal sequence reach the preset quantity, determine a set of the second complex signal sequences according to the complex signals in the third complex signal sequence.

[0131] In a feasible implementation, the apnea recognition device further includes:

[0132] A calculation module, configured to calculate the amplitude according to each complex signal in the fourth complex signal sequence to obtain the amplitude sequence corresponding to the fourth complex signal sequence, where the amplitude sequence includes the amplitudes corresponding to the multiple range cells, and the fourth complex signal sequence is one of the groups of the first complex signal sequences;

[0133] A determination module, configured to use the range cell corresponding to the peak value in the amplitude sequence as the target range cell corresponding to the fourth complex signal sequence; determine the range cell where the target object is located according to the target range cells corresponding to one or more groups of the fourth complex signal sequences.

[0134] In addition, an embodiment of the present invention further provides an apnea recognition device, as Figure 6 shown, Figure 6 is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment solution of the present invention. It should be noted that the apnea recognition device in the embodiment of the present invention may be a device such as a computer, a smart phone, a server, etc., and no specific limitation is made here.

[0135] Such as Figure 6As shown in the figure, the apnea recognition device may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display) and an input unit such as a keyboard (Keyboard). Optionally, the user interface 1003 may further include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0136] Those skilled in the art can understand that Figure 6 the device structure shown in the figure does not constitute a limitation on the apnea recognition device, and it may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0137] As Figure 6 shown, the memory 1005, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an apnea recognition program. The operating system is a program that manages and controls the hardware and software resources of the device and supports the operation of the apnea recognition program and other software or programs. In Figure 6 the device shown in the figure, the user interface 1003 is mainly used for data communication with the client; the network interface 1004 is mainly used to establish a communication connection with the server; and the processor 1001 may be used to call the apnea recognition program stored in the memory 1005 and perform the following operations:

[0138] Obtain each frame of echo signal received after transmitting a radar signal;

[0139] Convert each frame of the echo signal respectively to obtain a first complex signal sequence corresponding to each frame of the echo signal. Among them, the first complex signal sequence includes complex signals corresponding to a preset number of range cells;

[0140] Extract the complex signal corresponding to the range cell where the target object is located from each group of the first complex signal sequences, and obtain a second complex signal sequence in the slow-time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences;

[0141] Extract complex signal features from the second complex signal sequence, and identify the apnea recognition result of the target object according to the extracted complex signal features, where the complex signal features corresponding to normal breathing behavior and apnea behavior are different.

[0142] In a feasible implementation manner, the operations of extracting complex signal features from the second complex signal sequence and identifying the apnea recognition result of the target object according to the extracted complex signal features include:

[0143] Extract complex signal features from the second complex signal sequence, input the extracted complex signal features into a pre-trained classification model for classification, and obtain the apnea recognition result of the target object;

[0144] Among them, the classification model is pre-trained with the complex signal features corresponding to breathing behavior as input data and the type of breathing behavior as classification labels. The types of breathing behavior include normal breathing behavior and apnea behavior.

[0145] In a feasible implementation manner, the operations of extracting complex signal features from the second complex signal sequence, inputting the extracted complex signal features into a pre-trained classification model for classification, and obtaining the apnea recognition result of the target object include:

[0146] Draw a constellation diagram according to the second complex signal sequence, where the two coordinate axes of the constellation diagram are the real part and the imaginary part of the complex signal;

[0147] Use the constellation diagram as the extracted complex signal features and input them into a pre-trained classification model for classification to obtain the apnea recognition result of the target object.

[0148] In a feasible implementation manner, the extracted complex signal features include the standard deviation of the second complex signal sequence. The operations of extracting complex signal features from the second complex signal sequence and identifying the apnea recognition result of the target object according to the extracted complex signal features include:

[0149] Calculate the standard deviation of the second complex signal sequence;

[0150] If it is detected that the standard deviation is less than a preset threshold, obtain the apnea recognition result that the target object has apnea behavior;

[0151] If it is detected that the standard deviation is greater than or equal to the preset threshold, obtain the apnea recognition result that the target object performs normal breathing behavior.

[0152] In a feasible implementation manner, the operation of obtaining the second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences includes:

[0153] On the complex signal sequence composed of the complex signals extracted from each group of the first complex signal sequences, by adding time windows of different lengths, multiple lengths of second complex signal sequences are obtained;

[0154] The operation of extracting complex signal features according to the second complex signal sequence and identifying the apnea recognition result of the target object according to the extracted complex signal features further includes:

[0155] From the standard deviation thresholds corresponding to different preset lengths, determine the standard deviation threshold corresponding to the length of the second complex signal sequence as the preset threshold.

[0156] In a feasible implementation manner, the operation of obtaining the second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences includes:

[0157] After each new frame of the echo signal is received and a complex signal is extracted, if the number of complex signals in the third complex signal sequence is equal to the preset quantity, delete the earliest added complex signal in the third complex signal sequence, and add the extracted complex signal to the third complex signal sequence; otherwise, if the number of complex signals in the third complex signal sequence is less than the preset quantity, add the extracted complex signal to the third complex signal sequence, where the number of complex signals in the third complex signal sequence is zero when it is initialized;

[0158] After each new complex signal is added to make the number of complex signals in the third complex signal sequence reach the preset quantity, determine a group of the second complex signal sequences according to the complex signals in the third complex signal sequence.

[0159] In a feasible implementation manner, before the operation of respectively extracting the complex signals corresponding to the distance units where the target object is located from each group of the first complex signal sequences, the processor 1001 can also be used to call the apnea recognition program stored in the memory 1005 and perform the following operations:

[0160] Calculate the amplitude according to each complex signal in the fourth complex signal sequence to obtain the amplitude sequence corresponding to the fourth complex signal sequence, where the amplitude sequence includes the amplitudes respectively corresponding to the multiple distance units, and the fourth complex signal sequence is one of the groups of the first complex signal sequences;

[0161] Use the range cell corresponding to the peak value in the amplitude sequence as the target range cell corresponding to the fourth complex signal sequence;

[0162] Determine the range cell where the target object is located according to the target range cell corresponding to one or more groups of the fourth complex signal sequences.

[0163] In addition, an embodiment of the present invention also provides a computer-readable storage medium, on which a apnea recognition program is stored. When the apnea recognition program is executed by a processor, the steps of the apnea recognition method described below are implemented.

[0164] For each embodiment of the apnea recognition device and computer-readable storage medium of the present invention, reference can be made to each embodiment of the apnea recognition method of the present invention, which will not be elaborated here.

[0165] It should be noted that in this article, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including that element.

[0166] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments.

[0167] Through the description of the above embodiments, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions for causing a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0168] The above are only the preferred embodiments of the present invention, and do not limit the protection scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied to other related technical fields, are equally included in the protection scope of the present invention.

Claims

1. A method for apnea recognition, characterized in that, The apnea recognition method includes the following steps: Obtain each frame of echo signal received after transmitting a radar signal; Convert each frame of the echo signal respectively to obtain a first complex signal sequence corresponding to each frame of the echo signal, where the first complex signal sequence includes complex signals corresponding to a preset number of range cells; Extract the complex signals corresponding to the range cells where the target object is located from each group of the first complex signal sequences respectively, and obtain a second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences; Extract complex signal features according to the second complex signal sequence, and recognize the apnea recognition result of the target object according to the extracted complex signal features, where the complex signal features corresponding to normal breathing behavior and apnea behavior are different; 2. The method for apnea recognition according to claim 1, characterized in that, The step of extracting complex signal features according to the second complex signal sequence and recognizing the apnea recognition result of the target object according to the extracted complex signal features includes: Extract complex signal features according to the second complex signal sequence, input the extracted complex signal features into a pre-trained classification model for classification, and obtain the apnea recognition result of the target object; Wherein, the classification model is pre-trained with the complex signal features corresponding to breathing behavior as input data and the type of breathing behavior as classification labels, and the types of breathing behavior include normal breathing behavior and apnea behavior; 3. The method for apnea recognition according to claim 2, characterized in that, The step of extracting complex signal features according to the second complex signal sequence, inputting the extracted complex signal features into a pre-trained classification model for classification, and obtaining the apnea recognition result of the target object includes: Draw a constellation diagram according to the second complex signal sequence, where the two coordinate axes of the constellation diagram are the real part and the imaginary part of the complex signal respectively; Input the constellation diagram as the extracted complex signal features into a pre-trained classification model for classification, and obtain the apnea recognition result of the target object; 4. The method for apnea recognition according to claim 1, characterized in that, The extracted complex signal features include the standard deviation of the second complex signal sequence, and the step of extracting complex signal features according to the second complex signal sequence and recognizing the apnea recognition result of the target object according to the extracted complex signal features includes: Calculate the standard deviation of the second complex signal sequence; If it is detected that the standard deviation is less than a preset threshold, obtain the apnea recognition result that the target object has apnea behavior; If it is detected that the standard deviation is greater than or equal to the preset threshold, obtain the apnea recognition result that the target object performs normal breathing behavior; 5. The method for apnea recognition according to claim 4, characterized in that, The step of obtaining a second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences includes: On the complex signal sequence composed of the complex signals extracted from each group of the first complex signal sequences, obtain second complex signal sequences of various lengths by adding time windows of different lengths; The step of extracting complex signal features according to the second complex signal sequence and identifying the apnea recognition result of the target object according to the extracted complex signal features further includes: Determining a standard deviation threshold corresponding to the length of the second complex signal sequence from preset standard deviation thresholds corresponding to different lengths as the preset threshold.

6. The method for apnea recognition according to claim 1, characterized in that, The step of obtaining the second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences includes: After receiving a new frame of the echo signal and extracting a complex signal each time, if the number of complex signals in the third complex signal sequence is equal to a preset number, deleting the earliest added complex signal in the third complex signal sequence and adding the extracted complex signal to the third complex signal sequence; otherwise, if the number of complex signals in the third complex signal sequence is less than the preset number, adding the extracted complex signal to the third complex signal sequence, where the number of complex signals in the third complex signal sequence is zero at initialization. After each addition of a complex signal makes the number of complex signals in the third complex signal sequence reach the preset number, determining a group of the second complex signal sequences according to the complex signals in the third complex signal sequence.

7. The method for apnea recognition according to any one of claims 1 to 6, characterized in that, Before the step of respectively extracting the complex signals corresponding to the distance units where the target object is located from each group of the first complex signal sequences, it further includes: Calculating the amplitude according to each complex signal in the fourth complex signal sequence to obtain an amplitude sequence corresponding to the fourth complex signal sequence, where the amplitude sequence includes the amplitudes corresponding to the multiple distance units respectively, and the fourth complex signal sequence is one of the groups of the first complex signal sequences. Taking the distance unit corresponding to the peak value in the amplitude sequence as the target distance unit corresponding to the fourth complex signal sequence. Determining the distance unit where the target object is located according to the target distance unit corresponding to one or more groups of the fourth complex signal sequences.

8. An apnea recognition device, characterized in that, The apnea recognition device includes: An acquisition module, configured to acquire each frame of echo signal received after transmitting a radar signal; A conversion module, configured to respectively convert each frame of the echo signal to obtain a first complex signal sequence corresponding to each frame of the echo signal, where the first complex signal sequence includes complex signals corresponding to preset multiple distance units respectively; An extraction module, configured to respectively extract the complex signals corresponding to the distance units where the target object is located from each group of the first complex signal sequences, and obtain a second complex signal sequence in the slow time dimension according to the complex signals extracted from multiple groups of the first complex signal sequences; A recognition module, configured to extract complex signal features according to the second complex signal sequence and identify the apnea recognition result of the target object according to the extracted complex signal features, where the complex signal features corresponding to normal breathing behavior and apnea behavior are different.

9. An apnea recognition device, characterized in that, The apnea recognition device includes: a memory, a processor, and an apnea recognition program stored on the memory and executable on the processor. When the apnea recognition program is executed by the processor, it implements the steps of the apnea recognition method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, An apnea recognition program is stored on the computer-readable storage medium. When the apnea recognition program is executed by a processor, it implements the steps of the apnea recognition method according to any one of claims 1 to 7.

Images