Noise reduction processing method, device, equipment and medium
By combining the ICEEMDAN algorithm and the SANC algorithm, the noise reduction processing is performed on the traction fan soundprint monitoring data, which solves the problem of poor noise reduction effect in the existing technology, and achieves more effective noise filtering and signal retention.
Patent Information
- Application Number
- CN202410291896.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-14
- Publication Date
- 2025-02-18
AI Technical Summary
When the prior art reduces noise-reducing data of the traction fan soundprint monitoring data, some useful soundprint signals are often filtered out, resulting in poor noise reduction.
The improved adaptive noise complete set empirical modal decomposition ICEEMDAN algorithm is used to decompose the voiceprint data into easy-to-process single-frequency signals, and then the decomposed signal is adaptively denoised by the SANC algorithm.
Effectively filter out noise in voiceprint data, improve noise reduction effect, reduce the generation of pseudo-modal components, and solve the residual noise problem in inherent modal components.
Smart Images
Figure CN119479672A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of voiceprint signal processing, and in particular to a noise reduction processing method, device, equipment and medium. Background Art
[0002] In order to ensure the normal operation of electromechanical equipment, it is necessary to perform status detection and fault diagnosis on the electromechanical equipment, such as the traction fan in the locomotive machinery room. At present, voiceprint analysis technology can be used to monitor the status and diagnose faults of traction fans. However, since the locomotive machinery room where the traction fan is located also includes a large number of mechanical and electrical components, the collected traction fan voiceprint monitoring data will be affected by noise data, and the collected voiceprint monitoring data needs to be filtered and denoised.
[0003] However, the existing filtering technologies such as digital filters, wavelet threshold filters, EMD threshold filters, etc., when performing noise reduction on voiceprint monitoring data, will more or less filter out some useful voiceprint signals, resulting in poor noise reduction effect. Summary of the invention
[0004] The present application provides a noise reduction processing method, device, equipment and medium. By adopting an improved adaptive noise complete set empirical mode decomposition (ICEEMDAN) algorithm to decompose voiceprint data into easy-to-process single-frequency signals, and then using the SANC algorithm to perform adaptive noise reduction processing on the decomposed signals, the noise in the voiceprint data can be effectively filtered out and the noise reduction effect can be improved.
[0005] In a first aspect, the present application provides a noise reduction processing method, the method comprising:
[0006] Obtaining an original acoustic signal and a corresponding reference acoustic signal;
[0007] The original acoustic signal and the reference acoustic signal are respectively decomposed by using an improved adaptive noise complete set empirical mode decomposition (ICEEMDAN) algorithm to obtain an original intrinsic mode function set and a reference intrinsic mode function set, wherein the number of intrinsic mode functions in the original intrinsic mode function set and the reference intrinsic mode function set is consistent;
[0008] Using a self-reference adaptive noise cancellation SANC algorithm to perform adaptive noise reduction processing on the original intrinsic mode function set to obtain multiple noise-reduced intrinsic mode functions, wherein the reference signal of the self-reference adaptive noise cancellation SANC algorithm is the reference intrinsic mode function set;
[0009] A plurality of noise-reduced intrinsic mode functions are reconstructed to obtain a noise-reduced acoustic signal.
[0010] Optionally, obtaining an original acoustic signal and a corresponding reference acoustic signal includes:
[0011] Collecting original acoustic signals;
[0012] The original acoustic signal is subjected to time delay processing to obtain a reference acoustic signal.
[0013] Optionally, the original acoustic signal and the reference acoustic signal are decomposed respectively using an improved adaptive noise complete set empirical mode decomposition ICEEMDAN algorithm to obtain an original intrinsic mode function set and a reference intrinsic mode function set, including:
[0014] The original acoustic signal and the reference acoustic signal are decomposed by using the improved adaptive noise complete ensemble empirical mode decomposition (ICEEMDAN) algorithm to obtain multiple original intrinsic mode functions and multiple reference intrinsic mode functions.
[0015] According to the number of the plurality of original intrinsic modal functions and the number of the plurality of reference intrinsic modal functions, an original intrinsic modal function set and a reference intrinsic modal function set are obtained.
[0016] Optionally, obtaining an original intrinsic modal function set and a reference intrinsic modal function set according to the number of the plurality of original intrinsic modal functions and the number of the plurality of reference intrinsic modal functions comprises:
[0017] Comparing the number of the plurality of original intrinsic mode functions with the number of the plurality of reference intrinsic mode functions, determining a minimum value of the number, and taking the minimum value of the number as a retained value;
[0018] Determining the number of the multiple original intrinsic mode functions to be retained according to the retention value, and using the multiple original intrinsic mode functions after retention as the original intrinsic mode function set;
[0019] According to the reserved value, the number of the multiple reference intrinsic modal functions to be retained is determined, and the multiple reference intrinsic modal functions after being retained are used as a reference intrinsic modal function set.
[0020] Optionally, the process of obtaining a plurality of original intrinsic mode functions and a plurality of reference intrinsic mode functions includes:
[0021] Performing modal decomposition on Gaussian white noise by using an empirical mode decomposition (EMD) algorithm to obtain N stages of white noise intrinsic modal components, where N is an integer greater than or equal to 0;
[0022] The first-stage white noise intrinsic modal components are added to the original acoustic signal and the reference acoustic signal respectively to construct the original first sequence and the reference first sequence, and the original first sequence and the reference first sequence are modally decomposed by the empirical mode decomposition (EMD) algorithm to obtain the original first-order residual and the reference first-order residual;
[0023] Obtaining an original first-order intrinsic mode function according to a difference between an original acoustic signal and an original first-order residual remainder; and obtaining a reference first-order intrinsic mode function according to a difference between a reference acoustic signal and a reference first-order residual remainder;
[0024] The original first-order residual and the reference first-order residual are added with the second-stage white noise inherent modal component respectively, so as to construct the original second sequence and the reference second sequence, and the original second sequence and the reference second sequence are modally decomposed by the empirical mode decomposition (EMD) algorithm respectively to obtain the original second-order residual and the reference second-order residual;
[0025] According to the difference between the original first-order residual residual and the original second-order residual residual, an original second-order intrinsic mode function is obtained; and according to the difference between the reference first-order residual residual and the reference second-order residual residual, a reference second-order intrinsic mode function is obtained;
[0026] The loop is executed iteratively until the j-th order residual margin meets a preset stop condition, where j is an integer greater than or equal to 2 and less than or equal to N.
[0027] Optionally, a self-reference adaptive noise cancellation SANC algorithm is used to perform adaptive noise reduction processing on the original intrinsic mode function set to obtain multiple noise-reduced intrinsic mode functions, including:
[0028] The original intrinsic mode function set and the reference intrinsic mode function set are input into the self-reference adaptive noise cancellation SANC algorithm in a one-to-one correspondence according to the intrinsic mode function order, and the original intrinsic mode set is denoised to obtain multiple denoised intrinsic mode functions.
[0029] In a second aspect, the present application provides a noise reduction processing device, the device comprising:
[0030] An acquisition unit, used for acquiring an original acoustic signal and a corresponding reference acoustic signal;
[0031] A decomposition unit is used to decompose the original acoustic signal and the reference acoustic signal respectively by using an improved adaptive noise complete set empirical mode decomposition (ICEEMDAN) algorithm to obtain an original intrinsic mode function set and a reference intrinsic mode function set, wherein the number of intrinsic mode functions in the original intrinsic mode function set and the reference intrinsic mode function set is consistent;
[0032] A denoising unit, configured to perform adaptive denoising processing on an original intrinsic mode function set by using a self-reference adaptive noise cancellation SANC algorithm to obtain a plurality of denoised intrinsic mode functions, wherein a reference signal of the self-reference adaptive noise cancellation SANC algorithm is a reference intrinsic mode function set;
[0033] The reconstruction unit is used to reconstruct a plurality of noise reduction intrinsic mode functions to obtain a noise reduction acoustic signal.
[0034] Optionally, the obtaining unit is specifically used for:
[0035] Collecting original acoustic signals;
[0036] The original acoustic signal is subjected to time delay processing to obtain a reference acoustic signal.
[0037] Optionally, the decomposition unit comprises:
[0038] A decomposition subunit is used to decompose the original acoustic signal and the reference acoustic signal respectively by using an improved adaptive noise complete set empirical mode decomposition ICEEMDAN algorithm to obtain a plurality of original intrinsic mode functions and a plurality of reference intrinsic mode functions;
[0039] The obtaining subunit is used to obtain an original intrinsic modal function set and a reference intrinsic modal function set according to the number of the plurality of original intrinsic modal functions and the number of the plurality of reference intrinsic modal functions.
[0040] Optionally, the obtaining subunit is specifically used for:
[0041] Comparing the number of the plurality of original intrinsic mode functions with the number of the plurality of reference intrinsic mode functions, determining a minimum value of the number, and taking the minimum value of the number as a retained value;
[0042] Determining the number of the multiple original intrinsic mode functions to be retained according to the retention value, and using the multiple original intrinsic mode functions after retention as the original intrinsic mode function set;
[0043] According to the reserved value, the number of the multiple reference intrinsic modal functions to be retained is determined, and the multiple reference intrinsic modal functions after being retained are used as a reference intrinsic modal function set.
[0044] Optionally, the decomposition subunit is specifically used for:
[0045] Performing modal decomposition on Gaussian white noise by using an empirical mode decomposition (EMD) algorithm to obtain N stages of white noise intrinsic modal components, where N is an integer greater than or equal to 0;
[0046] The first-stage white noise intrinsic modal components are added to the original acoustic signal and the reference acoustic signal respectively to construct the original first sequence and the reference first sequence, and the original first sequence and the reference first sequence are modally decomposed by the empirical mode decomposition (EMD) algorithm to obtain the original first-order residual and the reference first-order residual;
[0047] Obtaining an original first-order intrinsic mode function according to a difference between an original acoustic signal and an original first-order residual remainder; and obtaining a reference first-order intrinsic mode function according to a difference between a reference acoustic signal and a reference first-order residual remainder;
[0048] The original first-order residual and the reference first-order residual are added with the second-stage white noise inherent modal component respectively, so as to construct the original second sequence and the reference second sequence, and the original second sequence and the reference second sequence are modally decomposed by the empirical mode decomposition (EMD) algorithm respectively to obtain the original second-order residual and the reference second-order residual;
[0049] According to the difference between the original first-order residual residual and the original second-order residual residual, an original second-order intrinsic mode function is obtained; and according to the difference between the reference first-order residual residual and the reference second-order residual residual, a reference second-order intrinsic mode function is obtained;
[0050] The loop is executed iteratively until the j-th order residual margin meets a preset stop condition, where j is an integer greater than or equal to 2 and less than or equal to N.
[0051] Optionally, the noise reduction unit is specifically used for:
[0052] The original intrinsic mode function set and the reference intrinsic mode function set are input into the self-reference adaptive noise cancellation SANC algorithm in a one-to-one correspondence according to the intrinsic mode function order, and the original intrinsic mode set is denoised to obtain multiple denoised intrinsic mode functions.
[0053] In a third aspect, the present application provides an electronic device, the device comprising a memory and a processor:
[0054] The memory is used to store computer programs;
[0055] The processor is used to execute the method provided by the first aspect according to the computer program.
[0056] In a fourth aspect, the present application further provides a computer-readable storage medium, which is used to store a computer program, and the computer program is used to execute the method provided in the first aspect above.
[0057] It can be seen that this application has the following beneficial effects:
[0058] The present application provides a noise reduction processing method. First, after obtaining an original acoustic signal and a corresponding reference acoustic signal, the original acoustic signal and the reference acoustic signal are decomposed by the ICEEMDAN algorithm to obtain an original intrinsic modal function set and a reference intrinsic modal function set, wherein the number of intrinsic modal functions in the original intrinsic modal function set and the reference intrinsic modal function set is consistent; then the original intrinsic modal function set is adaptively denoised by the SANC algorithm to obtain multiple denoised intrinsic modal functions, wherein the reference signal of the SANC algorithm is the reference intrinsic modal function set; finally, the multiple denoised intrinsic modal functions are reconstructed to obtain a noise-reduced acoustic signal. In this process, the original acoustic signal is decomposed into an intrinsic modal function set by the ICEEMDAN algorithm, and then when performing adaptive noise reduction processing, the SANC algorithm is used to directly use the time delay of the original acoustic signal as the reference acoustic signal without collecting additional reference signal data, thereby reducing the complexity and cost of data collection required to achieve adaptive noise elimination, and can effectively filter out various types of noise and improve the noise reduction effect. BRIEF DESCRIPTION OF THE DRAWINGS
[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0060] Figure 1 A schematic diagram of a noise reduction process in an embodiment of the present application;
[0061] Figure 2 A schematic diagram of a process flow of a noise reduction method according to an embodiment of the present application;
[0062] Figure 3 A schematic diagram of a process flow of a noise reduction method according to an embodiment of the present application;
[0063] Figure 4 This is a schematic diagram of the structure of a noise reduction processing device in an embodiment of the present application;
[0064] Figure 5 This is a schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION
[0065] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0066] The "first" in the names such as "first sequence" mentioned in the embodiments of the present application is only used as a name identifier and does not mean the first in order. The same rule applies to "second", "third", etc.
[0067] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.
[0068] First, the technical terms involved in the embodiments of the present application are explained as follows:
[0069] (1) ICEEMDAN algorithm: Improved adaptive noise complete set empirical mode decomposition algorithm is a signal processing method mainly used for signal decomposition. It combines the advantages of empirical mode decomposition (EMD) algorithm and wavelet denoising technology to remove noise and extract effective features of the signal. The main steps include: Signal decomposition: ICEEMDAN algorithm first uses empirical mode decomposition (EMD) to decompose the original signal to obtain multiple intrinsic mode functions (IMF); Noise removal: The noise in each IMF is removed by an adaptive noise algorithm; Complete integration: The denoised IMFs are fully integrated to obtain the decomposed signal.
[0070] (2) SANC algorithm: Self-reference adaptive noise cancellation algorithm, which is a process that passes a noise-contaminated signal through a filter to suppress the noise and allow the signal to pass through without distortion. Moreover, this adaptive processing process does not require prior knowledge of the characteristics of the signal and noise.
[0071] (3) EMD algorithm: The key to this algorithm is empirical mode decomposition, which can decompose complex signals into a finite number of intrinsic mode functions (IMFs). The decomposed IMF components contain local characteristic signals of the original signal at different time scales.
[0072] At present, the state monitoring and fault diagnosis of the traction fan in the locomotive machinery room are carried out through voiceprint analysis technology. However, since the locomotive machinery room where the traction fan is located also contains a large number of mechanical and electrical components, the collected voiceprint monitoring data will be affected by the noise data, so it is necessary to reduce the noise of the voiceprint monitoring data. However, the commonly used noise reduction technologies such as digital filters, wavelet threshold filters, EMD threshold filters, etc., will more or less filter out useful voiceprint signals, resulting in poor noise reduction effects.
[0073] In the embodiment of the present application, the voiceprint data is decomposed into a single-frequency signal that is easy to process by using the improved adaptive noise complete set empirical mode decomposition ICEEMDAN algorithm, and then the decomposed signal is subjected to adaptive noise reduction processing by the SANC algorithm, which can effectively filter out the noise in the voiceprint data and improve the noise reduction effect. In specific implementation, the method may include, for example: firstly, obtaining an original acoustic signal and a corresponding reference acoustic signal, and then using the ICEEMDAN algorithm to decompose the original acoustic signal and the reference acoustic signal respectively, and after obtaining the original intrinsic mode function set and the reference intrinsic mode function set, using the SANC algorithm to perform adaptive noise reduction processing on the original intrinsic mode function set to obtain multiple noise reduction intrinsic mode functions, and the reference signal of the self-reference adaptive noise elimination SANC algorithm is the reference intrinsic mode function set; finally, the multiple noise reduction intrinsic mode functions are reconstructed to obtain a noise reduction acoustic signal.
[0074] It can be seen that the method provided by the implementation of this application combines the ICEEMDAN algorithm and the SANC algorithm, and proposes a new joint noise reduction algorithm ICEEMDAN-SANC algorithm. By performing noise reduction processing on the simulation signal and the actual voiceprint data, various types of noise in the locomotive machinery room in the original acoustic signal, such as stationary noise and transient noise, can be effectively filtered out, thereby improving the noise reduction effect.
[0075] To facilitate understanding of the specific implementation of the noise reduction processing method provided in the embodiment of the present application, it will be described below with reference to the accompanying drawings.
[0076] It should be noted that the subject implementing the noise reduction processing method may be a noise reduction processing device provided in the embodiment of the present application, and the noise reduction processing device may be carried in an electronic device or a functional module of an electronic device. The electronic device in the embodiment of the present application may be any device capable of implementing the noise reduction processing method in the embodiment of the present application, for example, an Internet of Things (IoT) device.
[0077] Figure 1 A flowchart of a noise reduction processing method provided in an embodiment of the present application is shown in FIG. The method can be applied to a noise reduction processing device, which can be, for example, Figure 4The noise reduction processing device 400 shown in FIG. 1 may also be integrated in Figure 5 Functional modules in the electronic device 500 are shown.
[0078] like Figure 1 As shown, the method includes the following S101 to S104:
[0079] S101: Obtain an original acoustic signal and a corresponding reference acoustic signal.
[0080] In order to filter and reduce the noise of the collected original acoustic signal, it is first necessary to obtain the original acoustic signal and the corresponding reference acoustic signal, and then use the ICEEMDAN algorithm to decompose the original acoustic signal and the reference acoustic signal respectively. After obtaining the original intrinsic mode function set and the reference intrinsic mode function set, the SANC algorithm is used to perform adaptive noise reduction on the original intrinsic mode function set to obtain multiple noise-reduced intrinsic mode functions; finally, the multiple noise-reduced intrinsic mode functions are reconstructed to obtain the noise-reduced acoustic signal.
[0081] Therefore, the embodiment of the present application obtains the original acoustic signal and the corresponding reference acoustic signal through S101, which makes preliminary preparations for subsequently obtaining the original intrinsic modal function set and the reference intrinsic modal function set.
[0082] As an example, S101 may include: collecting original acoustic signals, such as collecting soundprint monitoring data of traction fans, and then delaying the original acoustic signals, that is, using the delayed original acoustic signals as reference acoustic signals, thereby obtaining reference acoustic signals.
[0083] S102: Decomposing the original acoustic signal and the reference acoustic signal respectively by using the ICEEMDAN algorithm to obtain an original intrinsic modal function set and a reference intrinsic modal function set, wherein the number of intrinsic modal functions in the original intrinsic modal function set and the number of intrinsic modal functions in the reference intrinsic modal function set are consistent.
[0084] As an example, S102 may include: S1021, using the ICEEMDAN algorithm to decompose the original acoustic signal and the reference acoustic signal respectively to obtain multiple original intrinsic modal functions and multiple reference intrinsic modal functions; that is, multi-order original intrinsic modal functions and multi-order reference intrinsic modal functions; S1022, according to the number of multiple original intrinsic modal functions and the number of multiple reference intrinsic modal functions, obtain an original intrinsic modal function set and a reference intrinsic modal function set.
[0085] Specifically, the process of S1021 may include: (1) performing modal decomposition on Gaussian white noise by using an EMD algorithm to obtain N stages of white noise intrinsic modal components, where N is an integer greater than or equal to 0; (2) adding the first stage of white noise intrinsic modal components to the original acoustic signal and the reference acoustic signal, respectively, to construct an original first sequence and a reference first sequence, and performing modal decomposition on the original first sequence and the reference first sequence by using an EMD algorithm, respectively, to obtain an original first-order residual residual and a reference first-order residual residual; (3) obtaining an original first-order intrinsic modal function according to the difference between the original acoustic signal and the original first-order residual residual; and obtaining a reference first-order intrinsic modal function according to the difference between the reference acoustic signal and the reference first-order residual residual. state function; (4) adding the second-stage white noise intrinsic modal component to the original first-order residual and the reference first-order residual respectively, constructing the original second sequence and the reference second sequence, and performing modal decomposition on the original second sequence and the reference second sequence respectively by the EMD algorithm, obtaining the original second-order residual and the reference second-order residual; (5) obtaining the original second-order intrinsic modal function according to the difference between the original first-order residual and the original second-order residual; and obtaining the reference second-order intrinsic modal function according to the difference between the reference first-order residual and the reference second-order residual; (6) looping and iterating until the j-th order residual satisfies the preset stop condition, where j is an integer greater than or equal to 2 and less than or equal to N. Wherein, when the residual is a monotonic function and cannot be decomposed any further, it means that the decomposition needs to be stopped, and the condition is set as the preset stop condition.
[0086] Since the number of intrinsic modal functions in the original intrinsic modal function set and the reference intrinsic modal function set is consistent, S1022 specifically includes: first, comparing the number of multiple original intrinsic modal functions with the number of multiple reference intrinsic modal functions, determining the minimum number, and taking the minimum number as the retention value; according to the retention value, determining the retention number of multiple original intrinsic modal functions, and taking the retained multiple original intrinsic modal functions as the original intrinsic modal function set; according to the retention value, determining the retention number of multiple reference intrinsic modal functions, and taking the retained multiple reference intrinsic modal functions as the reference intrinsic modal function set.
[0087] The retention value can be obtained by the formula k=min(n,m), where the k value represents the retention value, n represents the number of multiple original intrinsic mode functions, m represents the number of multiple reference intrinsic mode functions, and min represents the minimum value.
[0088] It should be noted that, according to the retained values, the order of obtaining the original intrinsic modal function set and the reference intrinsic modal function set is not limited. The original intrinsic modal function set can be obtained first, or the reference intrinsic modal function set can be obtained first, or the original intrinsic modal function set and the reference intrinsic modal function set can be obtained at the same time.
[0089] In order to ensure that the orders of the original intrinsic mode function and the reference intrinsic mode function in the subsequent denoising process are one-to-one corresponding, it is necessary to align the number of original intrinsic mode functions and reference intrinsic mode functions. Therefore, the minimum value of the two is required to be retained as the retained value, so as to obtain the original intrinsic mode function set and the reference intrinsic mode function set according to the retained value.
[0090] In this process, the ICEEMDAN algorithm can reduce the generation of pseudo-modal components and solve the residual noise problem in the intrinsic modal components, thereby improving the signal-to-noise ratio, and the acoustic signal can be decomposed into a single-frequency signal that is easy to process by the SANC algorithm.
[0091] S103: using the SANC algorithm to perform adaptive noise reduction processing on the original intrinsic mode function set to obtain a plurality of noise-reduced intrinsic mode functions, wherein the reference signal of the SANC algorithm is the reference intrinsic mode function set.
[0092] As an example, S103 may include: inputting the original intrinsic modal function set and the reference intrinsic modal function set into the SANC algorithm in a one-to-one correspondence according to the intrinsic modal function order, performing noise reduction on the original intrinsic modal set, and obtaining multiple noise-reduced intrinsic modal functions. In this process, a noise addition method and a decomposition strategy are introduced to improve the stability and accuracy of the decomposition, and the SANC algorithm is used without collecting additional reference acoustic signals, but directly using the time delay of the original signal as the reference signal, thereby reducing the complexity and cost of data collection required to achieve adaptive noise elimination.
[0093] S104: reconstructing a plurality of noise reduction intrinsic mode functions to obtain a noise reduction acoustic signal.
[0094] As an example, S104 may include: adding and combining multiple noise reduction intrinsic mode functions according to the hierarchical phase, that is, obtaining a noise reduction acoustic signal. In this process, it is ensured that the decomposed acoustic signal retains all the information of the original acoustic signal, thereby improving the accuracy of the decomposition.
[0095] It can be seen that the embodiment of the present application first obtains the original acoustic signal and the corresponding reference acoustic signal, and then uses the ICEEMDAN algorithm to decompose the original acoustic signal and the reference acoustic signal respectively, and obtains the original intrinsic modal function set and the reference intrinsic modal function set, wherein the number of intrinsic modal functions in the original intrinsic modal function set and the reference intrinsic modal function set is consistent; then the SANC algorithm is used to perform adaptive noise reduction processing on the original intrinsic modal function set to obtain multiple noise reduction intrinsic modal functions, and the reference signal of the self-reference adaptive noise elimination SANC algorithm is the reference intrinsic modal function set; finally, the multiple noise reduction intrinsic modal functions are reconstructed to obtain the noise reduction acoustic signal. In this process, the ICEEMDAN algorithm can reduce the generation of pseudo-modal components and solve the residual noise problem in the intrinsic modal component, thereby improving the effect of filtering noise reduction, and when performing adaptive noise reduction processing, the SANC algorithm can be used without collecting additional reference signal data, but directly using the delay of the original acoustic signal as the reference acoustic signal, thereby reducing the complexity and cost of data acquisition required to achieve adaptive noise elimination.
[0096] In order to make the method provided in the embodiment of the present application clearer and easier to understand, Figure 2 and Figure 3 A specific example of this method is described.
[0097] like Figure 2 As shown, the implementation may include:
[0098] S201: Collecting original acoustic signals.
[0099] As an example, S201 may include: collecting soundprint monitoring data of the traction fan as an original acoustic signal.
[0100] S202: Perform time delay processing on the original acoustic signal to obtain a reference acoustic signal.
[0101] like Figure 3 As shown, after the original acoustic signal is subjected to time delay processing, the delayed voiceprint monitoring data is obtained, and the delayed voiceprint monitoring data is used as the reference acoustic signal.
[0102] S203: Decomposing the original acoustic signal and the reference acoustic signal respectively by using the ICEEMDAN algorithm to obtain a plurality of original intrinsic mode functions and a plurality of reference intrinsic mode functions.
[0103] like Figure 3 As shown, the original acoustic signal and the reference acoustic signal are respectively input into the ICEEMDAN algorithm for decomposition, and multi-order original intrinsic mode functions and multi-order reference intrinsic mode functions are obtained.
[0104] The process of obtaining multiple original intrinsic modal functions and multiple reference intrinsic modal functions includes: performing modal decomposition on Gaussian white noise through an EMD algorithm to obtain N stages of white noise intrinsic modal components, wherein N is an integer greater than or equal to 0; adding the first stage of white noise intrinsic modal components to the original acoustic signal and the reference intrinsic modal function respectively to construct an original first sequence and a reference first sequence, and performing modal decomposition on the original first sequence and the reference first sequence respectively through an EMD algorithm to obtain an original first-order residual residue and a reference first-order residual residue; obtaining an original first-order intrinsic modal function according to the difference between the original acoustic signal and the original first-order residual residue; and according to the difference between the reference acoustic signal and the reference first-order residual residue, Obtain a reference first-order intrinsic modal function; add the second-stage white noise intrinsic modal component to the original first-order residual residue and the reference first-order residual residue, respectively, construct the original second sequence and the reference second sequence, and perform modal decomposition on the original second sequence and the reference second sequence respectively through the EMD algorithm to obtain the original second-order residual residue and the reference second-order residual residue; obtain the original second-order intrinsic modal function according to the difference between the original first-order residual residue and the original second-order residual residue; and obtain the reference second-order intrinsic modal function according to the difference between the reference first-order residual residue and the reference second-order residual residue; loop execution and iteration until the j-th order residual residue meets the preset stop condition, where j is an integer greater than or equal to 2 and less than or equal to N. Wherein, when the residual residue is a monotonic function and can no longer be decomposed, it means that the decomposition needs to be stopped, and the condition is set as the preset stop condition.
[0105] S204: Compare the number of the plurality of original intrinsic mode functions with the number of the plurality of reference intrinsic mode functions, determine a minimum number, and use the minimum number as a retained value.
[0106] like Figure 3 As described, it is necessary to use the formula k=min(n,m) to align the number of multiple original intrinsic modal functions and the number of multiple reference intrinsic modal functions. For example, the number n of original intrinsic modal functions is 5, and the number m of reference intrinsic modal functions is 3, that is, the minimum value of the number is determined to be 3, and the minimum value of the number is used as the retained value k, that is, the obtained retained value k is 3.
[0107] S205: Determine the number of the multiple original intrinsic modal functions to be retained according to the retention value, and use the multiple original intrinsic modal functions after retention as the original intrinsic modal function set.
[0108] As shown above, for example, if the retention value k is 3, it means that only the original intrinsic modal functions of order 1-3 among the multiple original intrinsic modal functions need to be retained, then the original intrinsic modal functions of order 4 and 5 among the multiple original intrinsic modal functions are deleted to obtain the original intrinsic modal function set.
[0109] S206: Determine the number of reference intrinsic modal functions to be retained according to the retention value, and use the retained reference intrinsic modal functions as a reference intrinsic modal function set.
[0110] As shown above, for example, if the retention value k is 3, it means that only the reference intrinsic modal functions of order 1-3 among the multiple reference intrinsic modal functions need to be retained. If the multiple reference intrinsic modal functions only contain reference intrinsic modal functions of order 1-3, there is no need to delete them, and the reference intrinsic modal function set is directly obtained.
[0111] S207: inputting the original intrinsic mode function set and the reference intrinsic mode function set into the SANC algorithm in a one-to-one correspondence according to the intrinsic mode function orders, performing denoising on the original intrinsic mode set, and obtaining a plurality of denoised intrinsic mode functions.
[0112] like Figure 3 As shown, the original intrinsic modal function of each order in the original intrinsic modal function set and the reference intrinsic modal function of each order in the reference intrinsic modal function set are input into the SANC algorithm in a one-to-one correspondence according to the order, so that the original intrinsic modal function of each order in the original intrinsic modal function set is denoised to obtain multiple denoised intrinsic modal functions, namely, multi-order denoised intrinsic modal functions.
[0113] S208: reconstructing a plurality of noise reduction intrinsic mode functions to obtain a noise reduction acoustic signal.
[0114] like Figure 3 As shown, finally, multiple noise reduction intrinsic mode functions are reconstructed to obtain noise reduction acoustic signals as the original acoustic signals after noise reduction, thereby achieving the noise reduction effect.
[0115] This embodiment provides a noise reduction processing method, which combines the ICEEMDAN algorithm and the SANC algorithm, and proposes a new joint noise reduction algorithm ICEEMDAN-SANC algorithm. By performing noise reduction processing on the simulation signal and the actual voiceprint data, various types of noise in the locomotive machinery room in the original acoustic signal, such as stationary noise and transient noise, can be effectively filtered out, thereby improving the noise reduction effect.
[0116] See also Figure 4 , the embodiment of the present application provides a noise reduction processing device 400, the device comprising:
[0117] An obtaining unit 401 is used to obtain an original acoustic signal and a corresponding reference acoustic signal;
[0118] A decomposition unit 402 is used to decompose the original acoustic signal and the reference acoustic signal respectively by using an improved adaptive noise complete set empirical mode decomposition (ICEEMDAN) algorithm to obtain an original intrinsic mode function set and a reference intrinsic mode function set, wherein the number of intrinsic mode functions in the original intrinsic mode function set and the reference intrinsic mode function set is consistent;
[0119] A noise reduction unit 403 is used to perform adaptive noise reduction processing on the original intrinsic mode function set by using a self-reference adaptive noise cancellation SANC algorithm to obtain a plurality of noise-reduced intrinsic mode functions, wherein the reference signal of the self-reference adaptive noise cancellation SANC algorithm is a reference intrinsic mode function set;
[0120] The reconstruction unit 404 is used to reconstruct the multiple noise reduction intrinsic mode functions to obtain the noise reduction acoustic signal.
[0121] Optionally, the obtaining unit 401 is specifically used for:
[0122] Collecting original acoustic signals;
[0123] The original acoustic signal is subjected to time delay processing to obtain a reference acoustic signal.
[0124] Optionally, the decomposition unit 402 includes:
[0125] A decomposition subunit is used to decompose the original acoustic signal and the reference acoustic signal respectively by using an improved adaptive noise complete set empirical mode decomposition ICEEMDAN algorithm to obtain a plurality of original intrinsic mode functions and a plurality of reference intrinsic mode functions;
[0126] The obtaining subunit is used to obtain an original intrinsic modal function set and a reference intrinsic modal function set according to the number of the plurality of original intrinsic modal functions and the number of the plurality of reference intrinsic modal functions.
[0127] Optionally, the obtaining subunit is specifically used for:
[0128] Comparing the number of the plurality of original intrinsic mode functions with the number of the plurality of reference intrinsic mode functions, determining a minimum value of the number, and taking the minimum value of the number as a retained value;
[0129] Determining the number of the multiple original intrinsic mode functions to be retained according to the retention value, and using the multiple original intrinsic mode functions after retention as the original intrinsic mode function set;
[0130] According to the reserved value, the number of the multiple reference intrinsic modal functions to be retained is determined, and the multiple reference intrinsic modal functions after being retained are used as a reference intrinsic modal function set.
[0131] Optionally, the decomposition subunit is specifically used for:
[0132] Performing modal decomposition on Gaussian white noise by using an empirical mode decomposition (EMD) algorithm to obtain N stages of white noise intrinsic modal components (IMF), where N is an integer greater than or equal to 0;
[0133] The first-stage white noise intrinsic modal components are added to the original acoustic signal and the reference acoustic signal respectively to construct the original first sequence and the reference first sequence, and the original first sequence and the reference first sequence are modally decomposed by the empirical mode decomposition (EMD) algorithm to obtain the original first-order residual and the reference first-order residual;
[0134] Obtaining an original first-order intrinsic mode function according to a difference between an original acoustic signal and an original first-order residual remainder; and obtaining a reference first-order intrinsic mode function according to a difference between a reference acoustic signal and a reference first-order residual remainder;
[0135] The original first-order residual and the reference first-order residual are added with the second-stage white noise inherent modal component respectively, so as to construct the original second sequence and the reference second sequence, and the original second sequence and the reference second sequence are modally decomposed by the empirical mode decomposition (EMD) algorithm respectively to obtain the original second-order residual and the reference second-order residual;
[0136] According to the difference between the original first-order residual residual and the original second-order residual residual, an original second-order intrinsic mode function is obtained; and according to the difference between the reference first-order residual residual and the reference second-order residual residual, a reference second-order intrinsic mode function is obtained;
[0137] The loop is executed iteratively until the j-th order residual margin meets a preset stop condition, where j is an integer greater than or equal to 2 and less than or equal to N.
[0138] Optionally, the noise reduction unit 403 is specifically configured to:
[0139] The original intrinsic mode function set and the reference intrinsic mode function set are input into the self-reference adaptive noise cancellation SANC algorithm in a one-to-one correspondence according to the intrinsic mode function order, and the original intrinsic mode set is denoised to obtain multiple denoised intrinsic mode functions.
[0140] It should be noted that the specific implementation method and effect of the noise reduction processing device 400 can be found in the above Figure 1 or Figure 2 The relevant descriptions in the provided methods will not be repeated here.
[0141] The present application embodiment also provides an electronic device 500, such as Figure 5 As shown, the device 500 includes a memory 501 and a processor 502:
[0142] The memory 501 is used to store computer programs;
[0143] The processor 502 is used to execute the above Figure 1 or Figure 2 The method provided.
[0144] In addition, the present application also provides a computer-readable storage medium, the computer-readable storage medium is used to store a computer program, and the computer program is used to execute Figure 1 or Figure 2 The method provided.
[0145] Through the description of the above implementation methods, it can be known that those skilled in the art can clearly understand that all or part of the steps in the above-mentioned embodiment method can be implemented by means of software plus a general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a storage medium, such as a read-only memory (ROM) / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network communication device such as a router) to execute the methods described in each embodiment of the present application or some parts of the embodiments.
[0146] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is only schematic, in which the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the goals of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.
[0147] The above description is merely an exemplary embodiment of the present application and is not intended to limit the protection scope of the present application.
Claims
1. A noise reduction method, characterized in that: The method comprises: Obtaining an original acoustic signal and a corresponding reference acoustic signal; Decomposing the original acoustic signal and the reference acoustic signal respectively by using an improved adaptive noise complete set empirical mode decomposition (ICEEMDAN) algorithm to obtain an original intrinsic mode function set and a reference intrinsic mode function set, wherein the number of intrinsic mode functions in the original intrinsic mode function set and the reference intrinsic mode function set is consistent; Using a self-reference adaptive noise cancellation SANC algorithm to perform adaptive noise reduction processing on the original intrinsic mode function set to obtain a plurality of noise-reduced intrinsic mode functions, wherein a reference signal of the self-reference adaptive noise cancellation SANC algorithm is the reference intrinsic mode function set; The multiple noise reduction intrinsic mode functions are reconstructed to obtain a noise reduction acoustic signal.
2. The method according to claim 1, characterized in that The obtaining of the original acoustic signal and the corresponding reference acoustic signal comprises: collecting the original acoustic signal; The original acoustic signal is subjected to time delay processing to obtain the reference acoustic signal.
3. The method according to claim 1, characterized in that The improved adaptive noise complete set empirical mode decomposition (ICEEMDAN) algorithm is used to decompose the original acoustic signal and the reference acoustic signal respectively to obtain an original intrinsic mode function set and a reference intrinsic mode function set, including: Decomposing the original acoustic signal and the reference acoustic signal respectively by using the improved adaptive noise complete set empirical mode decomposition (ICEEMDAN) algorithm to obtain a plurality of original intrinsic mode functions and a plurality of reference intrinsic mode functions; The original intrinsic modal function set and the reference intrinsic modal function set are obtained according to the number of the plurality of original intrinsic modal functions and the number of the plurality of reference intrinsic modal functions.
4. The method according to claim 3, characterized in that The obtaining, according to the number of the plurality of original intrinsic modal functions and the number of the plurality of reference intrinsic modal functions, the set of original intrinsic modal functions and the set of reference intrinsic modal functions comprises: Comparing the number of the plurality of original intrinsic mode functions with the number of the plurality of reference intrinsic mode functions, determining a minimum number, and using the minimum number as a retained value; Determining the number of the plurality of original intrinsic modal functions to be retained according to the retention value, and taking the plurality of original intrinsic modal functions after retention as the original intrinsic modal function set; According to the reserved value, the reserved number of the plurality of reference intrinsic modal functions is determined, and the reserved plurality of reference intrinsic modal functions is used as the reference intrinsic modal function set.
5. The method according to claim 3, characterized in that: The process of obtaining the plurality of original intrinsic mode functions and the plurality of reference intrinsic mode functions comprises: Performing modal decomposition on Gaussian white noise by using an empirical mode decomposition (EMD) algorithm to obtain N stages of white noise intrinsic modal components, where N is an integer greater than or equal to 0; Adding the first-stage white noise intrinsic modal component to the original acoustic signal and the reference acoustic signal respectively, constructing an original first sequence and a reference first sequence, and performing modal decomposition on the original first sequence and the reference first sequence respectively by using the empirical mode decomposition (EMD) algorithm to obtain an original first-order residual remainder and a reference first-order residual remainder; Obtaining an original first-order intrinsic mode function according to a difference between the original acoustic signal and the original first-order residual remainder; and obtaining a reference first-order intrinsic mode function according to a difference between the reference acoustic signal and the reference first-order residual remainder; Adding the second-stage white noise inherent modal components to the original first-order residual and the reference first-order residual to construct an original second sequence and a reference second sequence, and performing modal decomposition on the original second sequence and the reference second sequence by using the empirical mode decomposition (EMD) algorithm to obtain an original second-order residual and a reference second-order residual; According to the difference between the original first-order residual residual and the original second-order residual residual, an original second-order intrinsic mode function is obtained; and according to the difference between the reference first-order residual residual and the reference second-order residual residual, a reference second-order intrinsic mode function is obtained; The loop is executed iteratively until the j-th order residual margin meets a preset stop condition, where j is an integer greater than or equal to 2 and less than or equal to N.
6. The method according to claim 1, characterized in that The method of using the self-reference adaptive noise cancellation SANC algorithm to perform adaptive noise reduction processing on the original intrinsic mode function set to obtain multiple noise-reduced intrinsic mode functions includes: The original intrinsic modal function set and the reference intrinsic modal function set are input into the self-reference adaptive noise cancellation SANC algorithm in a one-to-one correspondence according to the intrinsic modal function orders, and the original intrinsic modal set is denoised to obtain the multiple denoised intrinsic modal functions.
7. A noise reduction processing device, characterized in that: The device comprises: An acquisition unit, used for acquiring an original acoustic signal and a corresponding reference acoustic signal; A decomposition unit, used for decomposing the original acoustic signal and the reference acoustic signal respectively by using an improved adaptive noise complete set empirical mode decomposition (ICEEMDAN) algorithm to obtain an original intrinsic mode function set and a reference intrinsic mode function set, wherein the number of intrinsic mode functions in the original intrinsic mode function set and the number of intrinsic mode functions in the reference intrinsic mode function set are consistent; A denoising unit, configured to perform adaptive denoising processing on the original intrinsic modal function set by using a self-reference adaptive noise cancellation SANC algorithm to obtain a plurality of denoised intrinsic modal functions, wherein a reference signal of the self-reference adaptive noise cancellation SANC algorithm is the reference intrinsic modal function set; The reconstruction unit is used to reconstruct the multiple noise reduction intrinsic mode functions to obtain a noise reduction acoustic signal.
8. The device according to claim 7, characterized in that The decomposition unit comprises: A decomposition subunit, used to decompose the original acoustic signal and the reference acoustic signal respectively by using the improved adaptive noise complete set empirical mode decomposition (ICEEMDAN) algorithm to obtain a plurality of original intrinsic mode functions and a plurality of reference intrinsic mode functions; The obtaining subunit is used to obtain the original intrinsic modal function set and the reference intrinsic modal function set according to the number of the plurality of original intrinsic modal functions and the number of the plurality of reference intrinsic modal functions.
9. An electronic device, characterized in that: The device includes a memory and a processor, and the processor is used to execute a program stored in the memory to run the method according to any one of claims 1 to 6.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store a computer program, and the computer program is used to execute the method according to any one of claims 1 to 6.