Variable step size self-adaptive noise reduction algorithm, device, medium and equipment for auscultation

Abstract

The invention discloses a variable step size self-adaptive noise reduction algorithm and device for auscultation, a medium and equipment. According to the algorithm, the states of friction, separation, re-auscultation and the like are set according to the conditions that the stethoscope moves on the body surface or leaves the body surface; in the auscultation process, the state of the stethoscope is judged online, and whether the weight of the adaptive filter is updated or not is determined according to the state of the stethoscope; in a normal auscultation state, a robust normalized least mean square algorithm is adopted to adaptively update the weight of the filter. According to the use characteristics of the stethoscope and the influence of factors such as friction sound on the environment sound noise reduction effect, the step length of weight updating of the self-adaptive filter is changed in a targeted mode, the convergence speed can be increased, steady-state weight imbalance is reduced, and therefore the good environment sound noise reduction effect is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical measurement and signal processing, and in particular to a variable step-size adaptive noise reduction algorithm, device, medium and equipment for auscultation. Background Art

[0002] Body sound signals such as heart sounds, respiratory sounds, and bowel sounds are weak, which makes body sound auscultation very susceptible to interference from environmental noise. Dual-channel variable step-size adaptive filtering is applied to environmental sound noise reduction during auscultation, which can significantly suppress environmental noise and effectively improve the signal-to-noise ratio of body sound signals.

[0003] The environmental noise reduction during the auscultation process has certain particularities, mainly manifested in: the duration of a single auscultation process is not long, and the weights of the adaptive filter need to converge quickly; in addition, doctors often move the stethoscope to listen to the body sound signals of multiple parts in turn; in this process, high-amplitude friction noise may be generated due to the movement of the stethoscope on the body surface, or the main channel may only collect environmental noise signals due to the stethoscope leaving the body surface for a short time. The way that friction noise enters the main and secondary channels is not exactly the same as the way that environmental noise enters the main and secondary channels, so the correlation between the friction noise collected by the main and secondary channels and the correlation between the environmental noise collected by the main and secondary channels are not exactly the same; in other words, the adaptive filter that can suppress friction noise may not be able to suppress environmental noise; therefore, if the stethoscope continues to iteratively update the weights of the environmental noise adaptive filter while the body surface is moving, it is very likely to reduce the environmental noise reduction performance. Similarly, after the stethoscope leaves the body surface, the correlation between the environmental noise collected by the main and auxiliary channels is also very different from the correlation between the environmental noise collected by the main and auxiliary channels when the stethoscope is attached to the body surface; at this time, if the weights of the environmental noise adaptive filter are updated, the environmental sound noise reduction performance will also be reduced.

[0004] In view of the particularity of the above-mentioned auscultation environmental sound noise reduction, the following variable step size principle can be adopted: after the transmission characteristics of the environmental noise channel mutate, a large step size is taken to increase the filter weight convergence speed; and after the weight converges, the step size is reduced to reduce the misalignment. The sudden change in the transmission characteristics of the environmental noise channel refers to the change in the correlation between the environmental noise collected by the main and auxiliary channels due to the stethoscope moving from the body surface to another auscultation position or leaving the body surface and contacting the body surface again for auscultation. However, there are two issues to be resolved in the specific implementation of this principle: ① How to determine whether the transmission characteristics of the environmental noise channel have mutated; ② How to change the step size after the mutation occurs. Summary of the invention

[0005] In order to overcome the shortcomings and deficiencies in the prior art, the purpose of the present invention is to provide a variable step-size adaptive noise reduction algorithm, device, medium and equipment for auscultation; the algorithm targets the influence of the stethoscope moving on or leaving the body surface on the noise reduction effect of ambient sound during auscultation, and determines the rules for updating the filter weights with variable step-size according to the state of the stethoscope, so as to obtain a noise reduction effect with fast convergence and low offset.

[0006] In order to achieve the above object, the present invention is implemented by the following technical solution: a variable step-size adaptive noise reduction algorithm for auscultation, characterized by comprising the following steps:

[0007] Step 1, call the "initialization" subroutine S.

[0008] Step 2: With sampling frequency F s The noisy body sound signal d(k) of the main channel and the environmental noise signal n(k) of the secondary channel at the kth moment are collected respectively.

[0009] Step 3: Call the "status judgment" subroutine ST.

[0010] Step 4: Determine the corresponding operation based on the current status. The rules are as follows:

[0011] If the out-of-body mark S L =1 or friction mark S F =1, then set S R =0, and jump to step 3;

[0012] If you start auscultation, mark S C =1; then take δ(K)=δ0, and set S C =0, S R =1;

[0013] If the normal state flag S R =1, then the step size threshold is changed Where the memory factor α∈(0,1) and the side channel noise vector is n(k)=[n(k),n(k-1),…,n(k-M+1)] T .

[0014] Step 5: Calculate the main channel noise estimate y(k):

[0015]

[0016] Step 6: Calculate the error signal e(k):

[0017] e(k)=d(k)-y(k).

[0018] Step 7: Update the adaptive filter weights w(k+1,j).

[0019] Step 8, take e(k) as the filtered result at the current moment and output e(k); determine whether the adaptive noise reduction termination condition is met: if the adaptive noise reduction termination indicator variable is true, then end the adaptive noise reduction algorithm, otherwise k=k+1, jump to step 2 to calculate the error signal e(k) at the next moment.

[0020] Preferably, the "initialization" subroutine S of step 1 includes the following steps S1 and S2:

[0021] Step S1: Set the normal state flag S R =0, out-of-body mark S L =0, friction mark S F =0, start auscultation mark S C =0;

[0022] In step S2, set the current time sequence number k = 0, the frame sequence number h = 0, the variable step threshold δ(k) = δ0, and the adaptive filter weight w(0,j) = 0, j∈{0,…,M-1}, where δ0 varies depending on the stethoscope and needs to be determined in advance through experiments, and M is the adaptive filter order.

[0023] Preferably, in the second step, the sampling frequency F s ≥4KHz.

[0024] Preferably, in the step 3, the "state determination" subroutine ST includes the following steps ST1 to ST4:

[0025] Step ST1: Calculate the mean amplitude of the main and auxiliary channels for the past L sampling points every L sampling points: If k>L and kmod L=0, where mod is the remainder operation, calculate the mean amplitude of the main channel respectively Auxiliary channel amplitude average and the ratio of the two means

[0026]

[0027] If or Then execute step ST2, otherwise execute step ST3. Among them, the thresholds D1, A1 and A2 vary with different stethoscopes and need to be determined in advance through experiments.

[0028] Step ST2: Calculate the amplitude root mean square S(h) of the main channel signal every window length T.

[0029] If k>L and k mod L=0, where mod is the remainder operation, then the main channel signal d(i) is subjected to an operation of order M. HP The cut-off frequency is F HPHigh-pass filtering processing is performed to obtain where \(i = k - L,\cdots,k - 1\). And calculate the root mean square \(S\) of the amplitude of the main channel signal:

[0030]

[0031] Judge if \(S(h)>T\) h1 and the average value of the main channel amplitude then set \(S\) L \(= 0\) and \(S\) F \(= 1\); otherwise, if \(S(h)<T\) h2 and the ratio of the average amplitudes of the main and auxiliary channels then set \(S\) L \(= 1\) and \(S\) F \(= 0\). Wherein, the thresholds \(T\) h1 and \(T\) h2 vary depending on the stethoscope and need to be determined in advance through experiments.

[0032] Step ST3: Judge if then set \(S\) C \(= 1\), otherwise set \(S\) C \(= 0\). Wherein, the threshold \(D2<D1\), and \(D2\) varies depending on the stethoscope and needs to be determined in advance through experiments.

[0033] Step ST4: Frame number \(f = h + 1\). If \(S\) L 、\(S\) F 、\(S\) C are all \(0\), then set \(S\) R \(= 1\).

[0034] Preferably, in the ST2 step, the window length \(T\in[4,50]\text{ms}\).

[0035] Preferably, in the ST2 step, the orders of the high-pass filter and the low-pass filter are not less than \(1\), that is, \(M\) HP \(\geq1\), \(M\) LP \(\geq1\).

[0036] Preferably, in the ST2 step, the cut-off frequencies of the high-pass filter and the low-pass filter are \(F\) HP \(\in[50,100]\text{Hz}\), \(F\) LP \(\in[0.2,3]\text{Hz}\).

[0037] Preferably, in the 4th step, the memory factor The preferred value of the coefficient \(\theta\) is \(1\).

[0038] Preferably, in the 7th step, the commonly used robust normalized least mean square (NLMS) adaptive filtering algorithm is used to update the weight \(w(k + 1,j)\), and the update formula is:

[0039]

[0040] Preferably, the value range of the adaptive filter order M is: M∈[20,300].

[0041] The technical principle of the algorithm of the present invention is:

[0042] Assuming that the environmental noise is additive, the main channel signal is the noisy body sound d(k)=s(k)+n1(k), where s(k) is the body sound signal at the kth moment, and n1(k) is the environmental noise signal entering the main channel at the kth moment. The environmental noise entering the secondary channel at the kth moment is recorded as n(k), which is processed by the fifth step of the algorithm of the present invention to obtain the main channel noise estimation value y(k) and is used to offset n1(k) to achieve noise reduction.

[0043] Under normal auscultation conditions, as shown in formula (1), the algorithm of the present invention uses the commonly used robust NLMS adaptive filtering algorithm to update the weight w in step 7. k+1,j .

[0044]

[0045] Where, e(k) = d(k) - y(k), w(k+1,j) and w(k,j) are the weights of the adaptive filter at the k+1th and kth moments respectively, and δ(k) is the variable step threshold. The expression for iterative update of the variable step threshold δ(k) is shown in (2):

[0046]

[0047] Formula (1) can be regarded as a step size of NLMS adaptive filtering algorithm. For this algorithm, when the small step size assumption is met, on the one hand, the larger the step size μ(k), the higher the adaptive filtering convergence rate; on the other hand, its steady-state mean square weight deviation (i.e., steady-state misalignment) and steady-state mean square error are both proportional to μ. 2 (k)| k→∞ , that is, the steady-state step size μ(k)| k→∞ The smaller it is, the smaller the steady-state misalignment is. Therefore, after the environmental noise channel characteristics suddenly change, resetting the variable step size threshold δ(k) to a larger initial value δ0 can increase the step size μ(k) of the weight update and improve the convergence speed. As shown in formula (2), as the number of iterations k increases, δ(k) decreases monotonically, and the step size μ(k) also decreases monotonically until the steady-state step size μ(k)| k→∞ →0, which plays a role in reducing steady-state imbalance.

[0048] In this way, the key to the algorithm's rapid convergence and low steady-state misalignment is to accurately determine when the environmental noise channel characteristics have undergone a sudden change. As described in the background technology introduction, common channel characteristic mutations usually occur when ① the stethoscope moves on the body surface and generates high-amplitude friction noise; ② the stethoscope leaves the body surface in a short period of time. For the first case, the friction sound generated by the movement of the stethoscope has a high amplitude and the energy is distributed in a very wide frequency band, resulting in a large root mean square of the main channel signal for a period of time. For the second case, the stethoscope leaves the human body, and the signals picked up by the main and auxiliary channels are all environmental noise; at this time, because the low-pass filtering effect of the main channel is more obvious than the low-pass filtering effect of the auxiliary channel, the amplitude of the main channel signal is small, while the amplitude of the auxiliary channel signal is relatively high, so it can be determined whether the stethoscope is off the body based on whether the root mean square of the main channel signal and the amplitude ratio of the main and auxiliary channel signals meet the threshold condition. The present invention designs an algorithm based on the above characteristics to determine whether the stethoscope moves on the body surface or leaves the body surface.

[0049] The movement of the stethoscope on the body surface indicates that the same subject is still being auscultated; in this case, when the movement is completed, there is no need to significantly change the weights of the adaptive filter, and the weights of the adaptive filter obtained before the movement of the stethoscope are used.

[0050] If the stethoscope is pressed on the body surface again for the next auscultation after leaving the body surface, the variable step size threshold δ(k) needs to be reset to a larger initial value δ0 so as to update the adaptive filter weights with a larger step size.

[0051] Accordingly, the present invention divides the mutations in the auscultation process into the following three types: ① the stethoscope rubs against the body surface; ② the stethoscope leaves the human body; ③ the stethoscope contacts the human body again after leaving the body, and sets a normal auscultation flag S R , in vitro marker S L , friction mark S F , Start auscultation sign S C The basis for the value change of each status flag variable is as follows:

[0052] ① Friction mark: calculate the mean amplitude of the main channel signal And the main channel signal amplitude root mean square S(h):

[0053]

[0054] When the stethoscope is rubbed against the body surface, the average amplitude and root mean square of the main channel signal are higher. And S(h)>T h1 When the stethoscope is judged to be in a friction state, set S F =1, S R =0.

[0055] 2. In vitro marker: Calculate the ratio of the mean amplitude of the main and auxiliary channel signals

[0056]

[0057] When the stethoscope leaves the human body, the signals picked up by the main and auxiliary channels are both environmental noise, the former has a smaller amplitude and the latter has a larger amplitude. And S(h) <T h2 When the stethoscope is judged to be in the off-body state, set S L =1, S R =0.

[0058] ③Start auscultation mark: calculate the mean amplitude of the main channel

[0059]

[0060] When auscultation begins, the stethoscope contacts the body surface and generates high-amplitude vibrations in a short period of time, and the amplitude of the main channel signal increases. When the stethoscope is judged to be in contact state, S C =1.

[0061] The present invention determines whether to update the adaptive filter weight and variable step size threshold according to the value of the status flag:

[0062] If the out-of-body mark S L =1 or friction mark S F =1, then S R =0, do not update the adaptive filter weights and variable step size threshold.

[0063] If you start auscultation, mark S C =1; then reset the variable step size threshold to δ(k)=δ0.

[0064] If the normal state flag S R =1, the adaptive filter weight and variable step size threshold are iteratively updated according to equation (1) and equation (2) respectively.

[0065] A variable step-size adaptive noise reduction device for auscultation, characterized in that it implements the above-mentioned variable step-size adaptive noise reduction algorithm for auscultation.

[0066] A storage medium, characterized in that the storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the above-mentioned variable step-size adaptive noise reduction algorithm for auscultation.

[0067] A computing device comprises a processor and a memory for storing a program executable by the processor, wherein when the processor executes the program stored in the memory, the variable step-size adaptive noise reduction algorithm for auscultation is implemented.

[0068] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0069] 1. The present invention aims at the influence of the stethoscope moving on or off the body surface on the ambient sound noise reduction effect during auscultation, and determines the rule of updating the filter weight with a variable step size according to the state of the stethoscope, so as to obtain the ambient sound noise reduction effect with fast convergence and low offset;

[0070] 2. The algorithm of the present invention has a small amount of calculation, a fast response speed, and low requirements on the computing power of the hardware equipment. It is particularly suitable for small wearable stethoscopes and small electronic stethoscopes. At the same time, the algorithm of the present invention is also suitable for use in electronic stethoscope auxiliary diagnosis and treatment systems for hospitals and homes. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 Flowchart of the algorithm of the present invention.

[0072] Figure 2 This is the curve showing the change of the main and auxiliary channel signal amplitudes over time when the stethoscope is rubbed on the body surface.

[0073] Figure 3 This is the curve of the change in the amplitude of the main and auxiliary channel signals before and after the stethoscope touches the human body (the sudden change in amplitude in the figure corresponds to the moment when it just touches the human body).

[0074] Figure 4 This is a curve showing the change of the ambient noise signal measured by the main and auxiliary channels when the stethoscope is out of the body over time.

[0075] Figure 5 This is the result obtained by applying the algorithm described in Example 1 (variable step size algorithm) to ambient sound noise reduction. DETAILED DESCRIPTION

[0076] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0077] Embodiment 1

[0078] The present embodiment provides a variable step-size adaptive noise reduction algorithm for auscultation. The algorithm aims to determine the rule for updating the filter weights with a variable step-size according to the state of the stethoscope, in order to achieve a noise reduction effect with fast convergence and low offset, in view of the influence of the stethoscope moving on or off the body surface during auscultation on the noise reduction effect of ambient sound.

[0079] To implement the above variable step size adaptive noise reduction algorithm for auscultation, this embodiment includes the following steps:

[0080] Step 1, call the "initialization" subroutine S.

[0081] Step 2: With sampling frequency F s =6K Hz to collect the main channel noisy body sound signal d(k) and the sub-channel environmental noise signal n(k) at the kth moment respectively.

[0082] Step 3: Call the "status judgment" subroutine ST.

[0083] Step 4: Determine the corresponding operation based on the current status. The rules are as follows:

[0084] If the out-of-body mark S L =1 or friction mark S F =1, then set S R =0, and jump to step 3;

[0085] If you start auscultation, mark S C =1; then take δ(k)=δ0, and set S C =0, S R =1;

[0086] If the normal state flag S R =1, then the step size threshold is changed Where n(k) = [n(k), n(k-1), ..., n(k-149)] T .

[0087] Step 5: Calculate the main channel noise estimate y(k):

[0088]

[0089] Step 6: Calculate the error signal e(k):

[0090] e(k)=d(k)-y(k).

[0091] Step 7: Update the adaptive filter weights w(k+1,j).

[0092] Step 8, take e(k) as the filtered result at the current moment and output e(k); determine whether the adaptive noise reduction termination condition is met: if the adaptive noise reduction termination indicator variable is true, then end the adaptive noise reduction algorithm, otherwise k=k+1, jump to step 2 to calculate the error signal e(k) at the next moment.

[0093] The "initialization" subroutine S in step 1 includes the following steps S1 and S2:

[0094] Step S1: Set the normal state flag S R =0, out-of-body mark S L=0, friction mark S F =0, start auscultation mark S C =0;

[0095] In step S2, the current time sequence number k=0, the frame sequence number h=0, the variable step threshold δ(k)=δ0, the adaptive filter weight w(0,j)=0, j∈{0,…,149}, where the adaptive filter order is 150, and δ0 is determined in the following way:

[0096] The stethoscope is pressed against the thigh for a period of time, during which the M sequence signal is played through the speaker to simulate the environmental noise. The volume of the speaker is controlled so that the noise volume measured near the stethoscope is approximately 75-95dB. The main and auxiliary channel signals are collected and recorded as {d δ1 ,d δ2 ,…,d δ2000},{n δ1 ,n δ2 ,…,n δ2000},but

[0097]

[0098] Among them, in the said step 3, the "state judgment" subroutine ST includes the following steps ST1 to ST4:

[0099] Step ST1: Calculate the mean amplitude of the main and auxiliary channels for the past 50 samples every L = 50 sampling points: If k>50 and k mod 50 = 0, where mod is the remainder operation, calculate the mean amplitude of the main channel respectively Auxiliary channel amplitude average and the ratio of the two means

[0100]

[0101] If or Then execute step ST2, otherwise execute step ST3. Wherein, D1, A1 and A2 are determined as follows:

[0102] Press the stethoscope on the body surface and rub it, collect the main channel signal and record it as {d DF1 ,d DF2 ,…,d DF200},but

[0103] Place the stethoscope in an environment with an ambient noise level greater than 40 dB, with the stethoscope surface that contacts the human body suspended in the air. Collect the main and auxiliary channel signals and record them as {d A1 ,d A2 ,…,d A200},{n A1 ,n A2 ,…,n A200},but

[0104] Step ST2: Perform a multiplication of the main channel signal d(i) into N HP =1 and the cutoff frequency is F HP =50Hz high-pass filtering, we get Where i = k-50,…,k-1. Calculate the RMS value of the main channel signal S(h):

[0105]

[0106] If S(h)>T h1 And the main channel amplitude mean Then set S L = 0 and S F =1; otherwise, if S(h) <T h2 And the ratio of the main and auxiliary channel amplitude means Then set S L =1 and S F =0.

[0107] Among them, the threshold T h1 It can be determined as follows:

[0108] Rub the stethoscope on the body surface, collect the main channel signal and perform an M-order HP =1 and the cutoff frequency is F HP =50Hz high-pass filtering, we get Calculate the RMS value of the main channel signal with a frame length of 50 and a frame shift of 10 m∈{1,…,200}. Then

[0109] Among them, the threshold T h2 It can be determined as follows:

[0110] Place the stethoscope in an environment with an ambient noise greater than 40dB, with the stethoscope surface that contacts the human body suspended in the air. Collect the main channel signal and perform an M-order HP =1 and the cutoff frequency is F HP =50Hz high-pass filtering, we get Calculate the RMS value of the main channel signal with a frame length of 50 and a frame shift of 10 m∈{1,…,200}. Then

[0111] Step ST3: Determine if Then set S C=1, otherwise set S C =0. The threshold D2 can be determined in the following manner:

[0112] Place the stethoscope in an environment with an ambient noise level less than 40 dB, with the stethoscope surface that contacts the human body suspended in the air. Collect the main channel signal and record it as {d D1 ,d D2 ,…,d D200},but

[0113] Step ST4: Frame number h = h + 1. L , S F , S C If both are 0, then set S R =1.

[0114] A variable step-size adaptive noise reduction device for auscultation implements the variable step-size adaptive noise reduction algorithm for auscultation.

[0115] A storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the above-mentioned variable step-size adaptive noise reduction algorithm for auscultation.

[0116] A computing device comprises a processor and a memory for storing a program executable by the processor, wherein when the processor executes the program stored in the memory, the variable step-size adaptive noise reduction algorithm for auscultation is implemented.

[0117] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A variable step size adaptive noise reduction algorithm for auscultation, characterized in that The steps include: Step 1, call the "initialization" subroutine S; Step 2: With sampling frequency F s The main channel noisy body sound signal d(k) and the secondary channel environmental noise signal n(k) are collected at the kth moment respectively; Step 3, call the "status judgment" subroutine ST; Step 4: Determine the corresponding operation based on the current status. The rules are as follows: If the out-of-body mark S L =1 or friction mark S F =1, then set S R =0, and jump to step 3; If you start auscultation, mark S C =1; then take δ(k)=δ0, and set S C =0, S R =1; If the normal state flag S R =1, then the step size threshold is changed Where the memory factor α∈(0,1) and the side channel noise vector is n(k)=[n(k),n(k-1),…,n(k-M+1)] T ; Step 5: Calculate the main channel noise estimate y(k): Step 6: Calculate the error signal e k : e(k)=d(k)-y(k); Step 7, update the adaptive filter weight w(k+1,j); Step 8, take e(k) as the filtered result at the current moment and output e(k); determine whether the adaptive noise reduction termination condition is met: if the adaptive noise reduction termination indicator variable is true, then end the adaptive noise reduction algorithm, otherwise k=k+1, jump to step 2 to calculate the error signal e(k) at the next moment.

2. The variable step size adaptive noise reduction algorithm for auscultation according to claim 1, characterized in that: In the step 3, the "state determination" subroutine ST includes the following steps ST1 to ST4: Step ST1: If or Then execute step ST2, otherwise execute step ST3; Step ST2: If S(h)>T h1 And the main channel amplitude mean Then set S L = 0 and S F =1; otherwise, if S(h) <T h2 And the ratio of the main channel amplitude to the secondary channel amplitude Then set S L =1 and S F =0; Step 3: If Then set S C =1, otherwise set S C =0; Step ST4: Frame number h = h + 1. L , S F , S C If both are 0, then set S R =1.

3. The variable step size adaptive noise reduction algorithm for auscultation according to claim 1, characterized in that: In step 4, the corresponding operation is determined according to the current state, and the rules are as follows: If the out-of-body mark S L =1 or friction mark S F =1, then set S R =0, and jump to step 3; If you start auscultation, mark S C =1; then take δ(k)=δ0, and set S C =0, S R =1; If the normal state flag S R =1, then the step size threshold is changed Where the memory factor α∈(0,1) and the side channel noise vector is n(k)=[n(k),n(k-1),…,n(k-M+1)] T .

4. The variable step size adaptive noise reduction algorithm for auscultation according to claim 2, characterized in that: In the "state judgment" subroutine ST, in step ST1, and The calculation method is: Calculate the mean amplitude of the main channel every L sampling points Auxiliary channel amplitude average and the ratio of the two means 5. The variable step size adaptive noise reduction algorithm for auscultation according to claim 2, characterized in that: In the "state judgment" subroutine ST, in step ST2, S(h) is calculated as follows: If k>L and k mod L=0, where mod is the remainder operation, then the main channel signal d(i) is subjected to an operation of order M. HP The cut-off frequency is F HP After high-pass filtering, we get Where i = kL,…,k-1; and calculate the amplitude root mean square S of the main channel signal:

6. The variable step size adaptive noise reduction algorithm for auscultation according to claim 1, characterized in that: In step 5, the main channel noise estimation value y(k) is calculated as follows:

7. The variable step size adaptive noise reduction algorithm for auscultation according to claim 1, characterized in that: In step 6, the error signal e(k) is calculated as follows: e(k)=d(k)-y(k).

8. The variable step size adaptive noise reduction algorithm for auscultation according to claim 1, characterized in that: In the step 7, the adaptive filter weight w(k+1,j) is updated in the following manner:

9. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, enables the processor to execute the variable step-size adaptive noise reduction algorithm for auscultation according to any one of claims 1 to 8.

10. A computing device comprising a processor and a memory for storing a program executable by the processor, characterized in that: When the processor executes the program stored in the memory, the variable step-size adaptive noise reduction algorithm for auscultation described in any one of claims 1 to 8 is implemented.

Images