A circuit board noise source localization method based on sparse matrix decomposition

Abstract

The present invention discloses a circuit board noise source positioning method based on sparse matrix decomposition, comprising the following steps: real-time collection of multi-channel noise signals from different functional areas of a circuit board through a multi-layer noise perception network; elimination of noise interference between different channels by a cross-domain noise decoupling algorithm, and compression of noise signals to a low-dimensional feature space by a dynamic dimensionality reduction technique; decomposition of noise signals based on non-negative sparse matrix decomposition technology, extraction of normal operation signals and abnormal noise features of the circuit board, and construction of a noise source fingerprint library; optimization of the positioning of noise sources by combining the temporal and spatial changes of noise signals with a multi-dimensional dynamic tracking positioning algorithm; iterative optimization of the noise source positioning model by an adaptive feedback mechanism, and prediction of possible future noise sources in combination with historical noise data, so as to deploy noise suppression measures in advance. The present invention improves the accuracy and efficiency of noise source positioning, and is suitable for noise management of complex electronic equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic equipment noise control, and in particular to a circuit board noise source positioning method based on sparse matrix decomposition. Background Art

[0002] With the widespread use of electronic devices, circuit board noise issues are increasingly becoming a key factor affecting device performance and reliability. In a complex circuit board working environment, noise signals may not only cause device dysfunction, but may also interfere with other systems. Therefore, accurate positioning and suppression of noise sources are crucial to the stable operation of electronic equipment. Although existing noise source positioning technologies can alleviate noise problems to a certain extent, they still have many deficiencies when dealing with complex noise environments, especially in terms of multi-channel noise signal processing, noise source feature extraction, and the accuracy and efficiency of noise source positioning. Existing technologies have obvious defects.

[0003] Existing noise collection methods usually rely on a single sensor or a simple multi-channel sensor network. These methods are difficult to effectively capture multi-channel noise signals from different areas when faced with a complex circuit board working environment, especially when transient and continuous noise exist at the same time. The noise signal collection efficiency is low. Traditional sampling technologies mostly use a fixed sampling frequency and cannot be adaptively adjusted according to changes in noise signals. The collected data often contains a large amount of redundant information, which not only increases the difficulty of processing, but also affects the accuracy of noise source positioning. When faced with complex noise signals in circuit boards, traditional sampling methods cannot effectively distinguish different types of noise characteristics, resulting in low efficiency in noise source positioning.

[0004] The mutual interference problem of multi-channel noise signals is a common problem in the existing technology. When noise sources exist in multiple functional areas of a circuit board at the same time, the noise signals are likely to interfere with each other, resulting in noise mutual interference. Traditional noise processing technologies mostly rely on simple noise elimination algorithms, which are difficult to effectively eliminate noise interference between channels. This technical means is not only unable to accurately distinguish the sources of different noise signals, but may even lead to erroneous signal interpretation, seriously affecting the positioning results of the noise source. In the coordinated processing of multi-channel noise signals, the existing technology has not yet proposed an effective solution, resulting in large errors in the positioning of noise sources.

[0005] Existing dimensionality reduction and feature extraction technologies also have limitations. When processing high-dimensional noise signals, general dimensionality reduction methods such as principal component analysis or singular value decomposition are usually used. These technologies are not ideal when facing complex noise environments. Traditional dimensionality reduction methods cannot effectively optimize the sparse characteristics of noise signals, resulting in the loss of key information. Technologies such as PCA and SVD are mainly used for signal noise reduction or data compression, but are not specifically designed for noise source positioning. It is difficult to retain sufficient noise features for subsequent analysis while achieving efficient dimensionality reduction. Especially when processing transient noise and abnormal noise, the existing technology is seriously insufficient in its ability to extract noise features, which in turn affects the precise positioning of noise sources.

[0006] Matrix decomposition technology is widely used in the field of signal processing, but its use in circuit board noise source positioning still has some limitations. Although traditional matrix decomposition methods can be used for signal compression and dimensionality reduction, they often show poor adaptability in noise source positioning. Existing matrix decomposition methods are difficult to effectively deal with the sparsity of noise signals, resulting in the inability to accurately extract the characteristics of abnormal noise signals. Traditional decomposition methods are mostly used for noise reduction operations in signal processing, but they fail to fully optimize in combination with the needs of noise source positioning, especially in the complex noise environment of circuit boards. Noise signals are diverse and dynamic. Traditional matrix decomposition methods show great limitations when dealing with these complex situations.

[0007] In terms of noise source positioning, existing technologies mainly rely on spectrum analysis and positioning algorithms based on noise intensity. Although spectrum analysis can capture the frequency characteristics of noise signals, it is prone to distortion when processing transient noise and high-frequency changing signals, resulting in large positioning errors. The noise intensity positioning algorithm performs preliminary positioning based on the signal strength of the noise source, but this method cannot accurately calculate the propagation path of the noise source and its dynamic changes in space. In complex noise environments, the positioning accuracy of existing technologies is difficult to meet the needs of actual applications.

[0008] The feedback optimization mechanism in the existing technology is usually relatively lagging, and fails to effectively combine the real-time collected data and historical data to dynamically adjust the noise source. Most existing solutions can only passively respond to the current noise source and lack the ability to actively predict future noise sources. This lagging feedback mechanism is difficult to adapt to the complex and changeable working environment of the circuit board, resulting in the difficulty of real-time optimization of the noise source positioning model, and the noise suppression effect is also affected. Most of the existing technologies fail to combine historical noise data to predict and intervene in future noise sources in advance, which makes noise management lack foresight when responding to dynamic changes.

[0009] Therefore, how to provide a circuit board noise source localization method based on sparse matrix decomposition is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0010] One object of the present invention is to propose a circuit board noise source positioning method based on sparse matrix decomposition. The present invention makes full use of sparse matrix decomposition technology, multi-channel noise decoupling algorithm and adaptive feedback optimization mechanism, and describes in detail the process of accurately positioning and dynamically optimizing the circuit board noise source through multi-layer noise perception network acquisition, dimensionality reduction processing, noise feature extraction and multi-dimensional dynamic tracking and positioning algorithm in a complex circuit board working environment. The method has the advantages of high noise source positioning accuracy, strong real-time performance and high processing efficiency.

[0011] A circuit board noise source location method based on sparse matrix decomposition according to an embodiment of the present invention comprises the following steps:

[0012] S1. Define a multi-layer noise sensing network on the circuit board. Through micro sensor nodes distributed in different functional areas, multi-channel noise signals are collected synchronously in real time. In combination with an adaptive dynamic frequency domain sampling algorithm, the sampling frequency and acquisition mode are automatically adjusted to adapt to the dynamic changes of transient and continuous noise signals.

[0013] S2. Apply the cross-domain noise decoupling algorithm to the noise signals of each acquisition channel, analyze the feature correlation and difference of multi-channel noise signals, separate the noise interference between channels, optimize the coordinated processing of signals, and use the dynamic dimensionality reduction mechanism to convert the high-dimensional noise signal into compressed noise feature data in the low-dimensional feature space;

[0014] S3. Based on the improved non-negative sparse matrix decomposition algorithm, the compressed noise feature data is decomposed twice. First, the low-rank signal matrix is ​​decomposed to represent the normal operation signal of the circuit board. Secondly, the abnormal noise features are extracted by decomposing the sparse matrix, and the noise source fingerprint library is constructed by combining the dynamic characteristics under different working modes.

[0015] S4. Based on the noise source fingerprint library, a multi-dimensional dynamic tracking and positioning algorithm is applied, combined with the spatiotemporal distribution information of the functional area of ​​the circuit board, the location of the noise source is dynamically calculated, and the noise source positioning path is optimized by analyzing the spatiotemporal changes of the noise intensity, so as to preferentially identify and locate the most disturbing noise source;

[0016] S5. Through the adaptive feedback optimization mechanism, according to the real-time collected noise signals and historical noise source data, the noise source fingerprint library is dynamically adjusted, and the noise source location model is iteratively optimized to adapt to the noise source location under complex working conditions of the circuit board;

[0017] S6. Based on the matching results of historical noise data and the current noise source fingerprint library, predict the noise sources that may be generated in the future and take noise suppression measures in advance;

[0018] The S3 specifically includes:

[0019] S31, based on the improved non-negative sparse matrix decomposition algorithm, the compressed global noise feature matrix Decompose and extract the low-rank signal matrix and the sparse noise matrix :

[0020] ;

[0021] in, for norm, represents the nuclear norm of the matrix, for Regularization term, and is the regularization coefficient;

[0022] S32, sparse noise matrix Represents abnormal noise characteristics, and sparsity is optimized through an adaptive control mechanism:

[0023] ;

[0024] in, for norm, Control the sparsity of sparse matrices, The term is used to enhance the adaptability of sparse matrices;

[0025] S33, combining the dynamic characteristics of the circuit board in different working modes, and constructing a noise source fingerprint library by decomposing the abnormal noise characteristics, wherein the noise source fingerprint library is generated by clustering and anomaly detection of sparse matrices;

[0026] S34. Dynamically update the noise source fingerprint library through an iterative feedback mechanism:

[0027] ;

[0028] in, is the updated sparse matrix, is the previous noise source fingerprint library matrix, is the learning rate, is the gradient adjustment item in the noise source fingerprint library.

[0029] Optionally, the S1 specifically includes:

[0030] S11, deploying micro sensor nodes in multiple functional areas of the circuit board, wherein the sensor nodes form a multi-layer noise sensing network to synchronously collect multi-channel noise signals in the area in real time;

[0031] S12. Through the adaptive dynamic frequency domain sampling algorithm, the sensor node automatically adjusts the sampling frequency according to the transient and continuous characteristics of the noise signal. And the acquisition mode:

[0032] ;

[0033] in, is the minimum sampling frequency, is the maximum sampling frequency, is a coefficient that is dynamically adjusted according to the change of the noise signal;

[0034] S13, the sensor node performs feature analysis based on the frequency, amplitude and duration of the collected noise signal, automatically distinguishes between transient noise signals and continuous noise signals, and for transient noise signals, the sensor node captures its changing characteristics through the set sampling frequency, and for continuous noise signals, avoids data redundancy by adjusting the sampling frequency, and optimizes the use of system resources;

[0035] S14. During the signal acquisition process, the sensor node uses the spectrum characteristics of the noise signal to Dynamically update sampling parameters, the spectrum characteristics reflect the frequency changes of noise signals over time, through real-time analysis According to the changing trend of the sensor node, when the high-frequency noise signal appears, the sensor node sets an appropriate sampling frequency, and when the noise signal tends to be stable, the sampling frequency is adjusted in time to avoid unnecessary data redundancy;

[0036] S15. Each sensor node dynamically adjusts the collection mode according to the noise feature classification, and combines with the multi-layer perception network of the circuit board to achieve comprehensive collection and processing of noise signals and adapt to the complex noise environment in different areas of the circuit board.

[0037] Optionally, the S2 specifically includes:

[0038] S21, preliminarily process the noise signal of each acquisition channel to generate a matrix related to the characteristics of the noise signal ,in For in time Down, channel The acquired noise signal characteristic data matrix;

[0039] S22. By analyzing the correlation of noise signals between different channels, a cross-channel collaborative processing model is constructed to define the channels. With channel The correlation function between for:

[0040] ;

[0041] in, and Channel and Channel The mean value of the noise signal, and is its standard deviation;

[0042] S23. Based on the results of the inter-channel correlation analysis, a cross-domain noise decoupling algorithm is applied to eliminate the noise interference effect between different channels. By constructing a dimensionality reduction mapping of the noise signal, the high-dimensional signal Project to low-dimensional feature space to generate low-dimensional noise feature matrix :

[0043] ;

[0044] in, is the dimension reduction mapping matrix;

[0045] S24. After cross-channel collaborative processing, the low-dimensional noise feature matrices of all channels are merged to generate a global noise feature matrix .

[0046] Optionally, the S4 specifically includes:

[0047] S41, based on noise source fingerprint library and global noise feature matrix , the multi-dimensional dynamic tracking and positioning algorithm is applied, combined with the spatiotemporal distribution information of different functional areas of the circuit board, the location of the noise source is dynamically calculated, and the initial location of the noise source is determined by analyzing the spatiotemporal characteristics of the noise signal in different areas of the circuit board. ;

[0048] S42. Optimize the location path of the noise source according to the temporal and spatial changes of the noise source signal strength:

[0049] ;

[0050] in, For the optimized position, is the gradient of the noise signal intensity, is the adaptive step size;

[0051] S43, by analyzing the strength and interference weight of the noise source signal , prioritize the noise sources:

[0052] ;

[0053] in, is the signal strength of the noise source, is the noise interference weight, is the noise propagation distance;

[0054] S44. Combined with the temporal and spatial distribution characteristics of the functional areas of the circuit board, multiple noise sources are synchronously located and optimized, so that the circuit board can effectively manage the location and impact of multiple noise sources in a complex noise environment.

[0055] Optionally, the S5 specifically includes:

[0056] S51, through the adaptive feedback optimization mechanism, combined with the real-time acquisition of noise signals and historical noise source data , construct an adaptive weighted noise source model, and assign weights according to the frequency, amplitude and interference of the noise signal through a dynamic weighting algorithm , dynamically adjust the noise source characteristics:

[0057] ;

[0058] in, is the weighting coefficient, is the adjustment coefficient, is the signal change gradient;

[0059] S52, iteratively optimize the noise source positioning model, combined with the adaptive path optimization mechanism, by introducing the time-varying characteristics of the noise signal, the historical and real-time noise signals are integrated into the path optimization model:

[0060] ;

[0061] in, is the time-varying weighting coefficient, is the second-order time derivative of the noise intensity;

[0062] S53. Aiming at the working conditions of complex circuit boards, the noise source fingerprint library is dynamically expanded through an adaptive acceleration feedback mechanism to process noise signals in different frequency ranges and form a multi-layer nested weighted feature model.

[0063] S54. Introduce a multi-level noise source feedback mechanism, build a spatiotemporal mapping model of the noise signal based on the real-time collected noise signal spectrum information, and use the model to update the data of the noise source fingerprint library, so that the noise source positioning process can adapt to real-time changes in a complex working environment.

[0064] Optionally, the S6 specifically includes:

[0065] S61, through historical noise data and the current noise source fingerprint library Matching can be performed to predict possible noise sources in the future:

[0066] ;

[0067] in, represents the noise source signal that may be generated in the future, and is the adjustment coefficient, is the gradient change of historical noise data, is the acceleration term of the current noise source;

[0068] S62, based on the matching results of the noise source fingerprint library, deploy noise suppression measures in advance, according to the noise signal strength and the location of the noise source Calculate the strength of the suppression measure:

[0069] ;

[0070] in, Indicates the intensity of the suppression measures, and is the adjustment coefficient, is the propagation distance of the noise signal, and are the future noise source position and the current noise source position respectively;

[0071] S63. By predicting the noise sources that may be generated in the future, the system deploys noise suppression strategies in advance, adjusts the suppression priority according to the interference and intensity of different noise sources, and dynamically manages the noise sources;

[0072] S64. After the noise suppression strategy is implemented, the noise source fingerprint library is updated based on the feedback mechanism, and the noise sources that may be generated in the future are continuously tracked and adjusted through the adaptive optimization mechanism.

[0073] The beneficial effects of the present invention are:

[0074] The present invention introduces a multi-layer noise perception network to collect multi-channel noise signals in different functional areas of the circuit board in real time, and combines an adaptive dynamic frequency domain sampling algorithm to dynamically adjust the sampling frequency and mode according to the transient and continuous characteristics of the noise signal. It adopts a cross-channel noise decoupling algorithm to effectively solve the mutual interference problem between multi-channel noise signals. By analyzing the correlation between different channels, the cross-interference of noise signals can be eliminated on the basis of collaborative processing.

[0075] Through the improved non-negative sparse matrix decomposition algorithm, the present invention can decompose the low-rank signal matrix and the sparse noise matrix from the compressed noise feature data, effectively extract the normal operating signal and abnormal noise characteristics of the circuit board, make full use of the sparsity of the noise signal, combine with the adaptive control mechanism, optimize the decomposition process of the sparse matrix, and make the extraction of abnormal noise more accurate, especially in the processing of transient noise and abnormal noise under complex working modes. This method can effectively construct a noise source fingerprint library and provide accurate feature data for subsequent noise source positioning.

[0076] The present invention uses a multi-dimensional dynamic tracking and positioning algorithm, combined with the spatiotemporal distribution information of the functional area of ​​the circuit board, to dynamically calculate the location of the noise source. By analyzing the spatiotemporal changes and propagation paths of the noise source signal strength, the positioning path of the noise source can be optimized, and the most interfering noise source can be quickly identified and located.

[0077] The present invention can also dynamically adjust the noise source fingerprint library and iteratively optimize the noise source positioning model through an adaptive feedback optimization mechanism, combined with real-time collected noise signals and historical noise source data. This mechanism gives the system a strong adaptive ability, enabling it to adapt to the dynamic changes of the complex working environment of the circuit board and maintain efficient noise source positioning and management. The present invention can predict the noise sources that may be generated in the future based on historical noise data and current noise source characteristics, and deploy corresponding noise suppression measures in advance. The noise management capability greatly improves the system's preventiveness, reduces the probability of noise problems, and enhances the initiative of noise suppression. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0079] Figure 1 A flowchart of a circuit board noise source location method based on sparse matrix decomposition proposed by the present invention;

[0080] Figure 2 A schematic diagram of extracting noise signal features and building a noise source fingerprint library based on a non-negative sparse matrix decomposition algorithm;

[0081] Figure 3 Flowchart of the multi-dimensional dynamic tracking and positioning algorithm for noise source positioning. DETAILED DESCRIPTION

[0082] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0083] refer to Figure 1-3 , a circuit board noise source location method based on sparse matrix decomposition, comprising the following steps:

[0084] S1. Define a multi-layer noise sensing network on the circuit board. Through micro sensor nodes distributed in different functional areas, multi-channel noise signals are collected synchronously in real time. In combination with an adaptive dynamic frequency domain sampling algorithm, the sampling frequency and acquisition mode are automatically adjusted to adapt to the dynamic changes of transient and continuous noise signals.

[0085] S2. Apply the cross-domain noise decoupling algorithm to the noise signals of each acquisition channel, analyze the feature correlation and difference of multi-channel noise signals, separate the noise interference between channels, optimize the coordinated processing of signals, and use the dynamic dimensionality reduction mechanism to convert the high-dimensional noise signal into compressed noise feature data in the low-dimensional feature space;

[0086] S3. Based on the improved non-negative sparse matrix decomposition algorithm, the compressed noise feature data is decomposed twice. First, the low-rank signal matrix is ​​decomposed to represent the normal operation signal of the circuit board. Secondly, the abnormal noise features are extracted by decomposing the sparse matrix, and the noise source fingerprint library is constructed by combining the dynamic characteristics under different working modes.

[0087] S4. Based on the noise source fingerprint library, a multi-dimensional dynamic tracking and positioning algorithm is applied, combined with the spatiotemporal distribution information of the functional area of ​​the circuit board, the location of the noise source is dynamically calculated, and the noise source positioning path is optimized by analyzing the spatiotemporal changes of the noise intensity, so as to preferentially identify and locate the most disturbing noise source;

[0088] S5. Through the adaptive feedback optimization mechanism, according to the real-time collected noise signals and historical noise source data, the noise source fingerprint library is dynamically adjusted, and the noise source location model is iteratively optimized to adapt to the noise source location under complex working conditions of the circuit board;

[0089] S6. Based on the matching results of historical noise data and the current noise source fingerprint library, predict the noise sources that may be generated in the future and take noise suppression measures in advance;

[0090] S3 specifically includes:

[0091] S31, based on the improved non-negative sparse matrix decomposition algorithm, the compressed global noise feature matrix Decompose and extract the low-rank signal matrix and the sparse noise matrix :

[0092] ;

[0093] in, for norm, represents the nuclear norm of the matrix, for Regularization term, and is the regularization coefficient;

[0094] S32, sparse noise matrix Represents abnormal noise characteristics, and sparsity is optimized through an adaptive control mechanism:

[0095] ;

[0096] in, for norm, Control the sparsity of sparse matrices, The term is used to enhance the adaptability of sparse matrices;

[0097] S33, combining the dynamic characteristics of the circuit board in different working modes, and constructing a noise source fingerprint library by decomposing the abnormal noise characteristics, wherein the noise source fingerprint library is generated by clustering and anomaly detection of sparse matrices;

[0098] S34. Dynamically update the noise source fingerprint library through an iterative feedback mechanism:

[0099] ;

[0100] in, is the updated sparse matrix, is the previous noise source fingerprint library matrix, is the learning rate, is the gradient adjustment item in the noise source fingerprint library.

[0101] In this implementation, S1 specifically includes:

[0102] S11, deploying micro sensor nodes in multiple functional areas of the circuit board, the sensor nodes forming a multi-layer noise sensing network, and synchronously collecting multi-channel noise signals in the area in real time;

[0103] S12. Through the adaptive dynamic frequency domain sampling algorithm, the sensor node automatically adjusts the sampling frequency according to the transient and continuous characteristics of the noise signal. And the acquisition mode:

[0104] ;

[0105] in, is the minimum sampling frequency, is the maximum sampling frequency, is a coefficient that is dynamically adjusted according to the change of the noise signal;

[0106] S13, the sensor node performs feature analysis based on the frequency, amplitude and duration of the collected noise signal, automatically distinguishes between transient noise signals and continuous noise signals, and for transient noise signals, the sensor node captures its changing characteristics through the set sampling frequency, and for continuous noise signals, avoids data redundancy by adjusting the sampling frequency, and optimizes the use of system resources;

[0107] S14. During the signal acquisition process, the sensor node uses the spectrum characteristics of the noise signal to Dynamically update sampling parameters, the spectrum characteristics reflect the frequency changes of noise signals over time, through real-time analysis According to the changing trend of the sensor node, when the high-frequency noise signal appears, the sensor node sets an appropriate sampling frequency, and when the noise signal tends to be stable, the sampling frequency is adjusted in time to avoid unnecessary data redundancy;

[0108] S15. Each sensor node dynamically adjusts the collection mode according to the noise feature classification, and combines with the multi-layer perception network of the circuit board to achieve comprehensive collection and processing of noise signals and adapt to the complex noise environment in different areas of the circuit board.

[0109] In this implementation, S2 specifically includes:

[0110] S21, preliminarily process the noise signal of each acquisition channel to generate a matrix related to the characteristics of the noise signal ,in For in time Down, channel The acquired noise signal characteristic data matrix;

[0111] S22. By analyzing the correlation of noise signals between different channels, a cross-channel collaborative processing model is constructed to define the channels. With channel The correlation function between for:

[0112] ;

[0113] in, and Channel and Channel The mean value of the noise signal, and is its standard deviation;

[0114] S23. Based on the results of the inter-channel correlation analysis, a cross-domain noise decoupling algorithm is applied to eliminate the noise interference effect between different channels. By constructing a dimensionality reduction mapping of the noise signal, the high-dimensional signal Project to low-dimensional feature space to generate low-dimensional noise feature matrix :

[0115] ;

[0116] in, is the dimension reduction mapping matrix;

[0117] S24. After cross-channel collaborative processing, the low-dimensional noise feature matrices of all channels are merged to generate a global noise feature matrix .

[0118] In this implementation, S4 specifically includes:

[0119] S41, based on noise source fingerprint library and global noise feature matrix , the multi-dimensional dynamic tracking and positioning algorithm is applied, combined with the spatiotemporal distribution information of different functional areas of the circuit board, the location of the noise source is dynamically calculated, and the initial location of the noise source is determined by analyzing the spatiotemporal characteristics of the noise signal in different areas of the circuit board. ;

[0120] S42. Optimize the location path of the noise source according to the temporal and spatial changes of the noise source signal strength:

[0121] ;

[0122] in, For the optimized position, is the gradient of the noise signal intensity, is the adaptive step size;

[0123] S43, by analyzing the strength and interference weight of the noise source signal , prioritize the noise sources:

[0124] ;

[0125] in, is the signal strength of the noise source, is the noise interference weight, is the noise propagation distance;

[0126] S44. Combined with the temporal and spatial distribution characteristics of the functional areas of the circuit board, multiple noise sources are synchronously located and optimized, so that the circuit board can effectively manage the location and impact of multiple noise sources in a complex noise environment.

[0127] In this implementation, S5 specifically includes:

[0128] S51, through the adaptive feedback optimization mechanism, combined with the real-time acquisition of noise signals and historical noise source data , construct an adaptive weighted noise source model, and assign weights according to the frequency, amplitude and interference of the noise signal through a dynamic weighting algorithm , dynamically adjust the noise source characteristics:

[0129] ;

[0130] in, is the weighting coefficient, is the adjustment coefficient, is the signal change gradient;

[0131] S52, iteratively optimize the noise source positioning model, combined with the adaptive path optimization mechanism, by introducing the time-varying characteristics of the noise signal, the historical and real-time noise signals are integrated into the path optimization model:

[0132] ;

[0133] in, is the time-varying weighting coefficient, is the second-order time derivative of the noise intensity;

[0134] S53. Aiming at the working conditions of complex circuit boards, the noise source fingerprint library is dynamically expanded through an adaptive acceleration feedback mechanism to process noise signals in different frequency ranges and form a multi-layer nested weighted feature model.

[0135] S54. Introduce a multi-level noise source feedback mechanism, build a spatiotemporal mapping model of the noise signal based on the real-time collected noise signal spectrum information, and use the model to update the data of the noise source fingerprint library, so that the noise source positioning process can adapt to real-time changes in a complex working environment.

[0136] In this implementation, S6 specifically includes:

[0137] S61, through historical noise data and the current noise source fingerprint library Matching can be performed to predict possible noise sources in the future:

[0138] ;

[0139] in, represents the noise source signal that may be generated in the future, and is the adjustment coefficient, is the gradient change of historical noise data, is the acceleration term of the current noise source;

[0140] S62, based on the matching results of the noise source fingerprint library, deploy noise suppression measures in advance, according to the noise signal strength and the location of the noise source Calculate the strength of the suppression measure:

[0141] ;

[0142] in, Indicates the intensity of the suppression measures, and is the adjustment coefficient, is the propagation distance of the noise signal, and are the future noise source position and the current noise source position respectively;

[0143] S63. By predicting the noise sources that may be generated in the future, the system deploys noise suppression strategies in advance, adjusts the suppression priority according to the interference and intensity of different noise sources, and dynamically manages the noise sources;

[0144] S64. After the noise suppression strategy is implemented, the noise source fingerprint library is updated based on the feedback mechanism, and the noise sources that may be generated in the future are continuously tracked and adjusted through the adaptive optimization mechanism.

[0145] Embodiment 1:

[0146] In order to verify the feasibility of the present invention in implementation, the present invention is applied to the production workshop of an electronic manufacturing enterprise, mainly for accurately locating and analyzing the noise source in the circuit board of the electronic equipment to improve the stability and reliability of the equipment in the production process. The environment in the production workshop is relatively complex, including various types of electronic equipment working, among which the noise signals generated by some equipment affect the normal operation of other equipment, especially when the equipment is started, the noise interference problem is particularly prominent. The traditional noise detection method is difficult to accurately identify and locate the noise source due to the complex and uneven distribution of the noise signal, which affects the normal operation efficiency of the production line. Therefore, the production workshop urgently needs an efficient noise source locating method so that the noise source can be quickly identified and measures can be taken to suppress noise interference.

[0147] In the production workshop scenario, the method of the present invention first deploys micro sensor nodes on the main electronic equipment of the production line to form a multi-layer noise perception network. Each sensor node synchronously collects the multi-channel noise signals generated during the operation of the equipment in real time to ensure comprehensive noise monitoring of different areas of the equipment. During the collection process, the system combines an adaptive dynamic frequency domain sampling algorithm to dynamically adjust the sampling frequency and collection mode according to the transient and continuous characteristics of the equipment noise signal. When the equipment makes a sudden noise, the system can capture the instantaneous changing noise signal. When the equipment runs for a long time, the system reduces the sampling frequency to avoid the generation of a large amount of redundant data, thereby optimizing data processing efficiency.

[0148] Through the cross-channel noise decoupling algorithm, the system performs correlation analysis on the noise signals collected by different channels and eliminates the noise interference between channels. The noise signal of each channel is mapped to a low-dimensional feature space to generate compressed noise feature data. Based on the improved non-negative sparse matrix decomposition algorithm, the system decomposes these compressed noise feature data and extracts the normal operating signal and abnormal noise signal of the circuit board respectively. Through the sparse matrix obtained by decomposition, the system constructs a noise source fingerprint library of the equipment and records the noise characteristics of each device under different operating conditions. In order to further optimize the positioning of the noise source, the system combines a multi-dimensional dynamic tracking and positioning algorithm, uses the spatiotemporal distribution information of different functional areas of the circuit board, dynamically calculates the position of the noise source, and optimizes the positioning path of the noise source according to the noise signal strength and propagation path, and preferentially identifies and locates the most interfering noise source.

[0149] The application process of this method in the production workshop shows that when the equipment is running, the noise signal in a certain area frequently exceeds the expected value. The system successfully captures the abnormal noise signal in the area through a multi-layer perception network. After analyzing the noise signal in the area, the system finds that the noise signal has a strong transient characteristic. The sampling frequency is adjusted in real time with the drastic changes in the noise intensity, ensuring the accurate capture of transient noise. By comparing the noise source fingerprint library with historical data, the system infers that the noise source comes from a specific component when the equipment is started. During the operation of the production line, the system can adjust the fingerprint library in time and further optimize the noise source positioning path through real-time data analysis and feedback, and quickly locate the specific location of the noise source.

[0150] In order to verify the beneficial effects of the present invention, we compared the effects of using the technology of the present invention and the traditional noise source positioning method in the same scene. In terms of the positioning time of the noise source when the equipment is running, the average positioning time of the technology of the present invention is 15 seconds, while the traditional method takes an average of 60 seconds, which significantly shortens the positioning time. In terms of positioning accuracy, the positioning error of the noise source using the technology of the present invention is less than 5 cm, while the positioning error of the traditional method is greater than 20 cm. In terms of noise source recognition rate, the technology of the present invention can identify more than 95% of the noise sources, while the recognition rate of the traditional method is less than 80%. These data show that the technology of the present invention is significantly superior to the traditional method in terms of positioning time, accuracy and recognition rate.

[0151] After a period of operation, the production workshop reported that the overall operating status of the equipment has been significantly improved. Since the noise source can be quickly located and effectively suppressed, the equipment failure rate has been reduced by 20%, production efficiency has been increased by 15%, and the operating noise of the equipment has been reduced by an average of 10 decibels, greatly improving the working environment of the workers.

[0152] Table 1 Comparison of noise source location technologies

[0153]

[0154] By comparing the data tables, it can be seen that the noise source positioning technology of the present invention is significantly superior to the traditional technology in all aspects. The technology of the present invention can significantly shorten the positioning time of the noise source, improve the accuracy to within 5 cm, and increase the recognition rate of the noise source to more than 95%. The equipment failure rate is reduced by 20%, the production efficiency is improved by 15%, and the equipment noise level is reduced by 10 decibels. In comparison, the traditional technology has no obvious improvement in these aspects, the positioning error is large, the recognition rate is low, and there is almost no effect on the optimization of equipment failure rate and production efficiency.

[0155] In summary, the present invention successfully solves the problems of difficult noise source positioning, low processing efficiency, large signal error, etc. in complex noise environments through the combination of innovative noise signal acquisition, dimensionality reduction processing, sparse matrix decomposition and dynamic positioning algorithm. Both experimental data and actual production applications have proved the effectiveness and superiority of this method. Through the adaptive feedback mechanism, the present invention also has the ability of real-time adjustment and dynamic optimization, ensuring that the noise source positioning is still highly accurate and efficient in complex and changeable working environments.

[0156] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A circuit board noise source location method based on sparse matrix decomposition, characterized in that: The steps include: S1. Define a multi-layer noise sensing network on the circuit board. Through micro sensor nodes distributed in different functional areas, multi-channel noise signals are collected synchronously in real time. In combination with an adaptive dynamic frequency domain sampling algorithm, the sampling frequency and acquisition mode are automatically adjusted to adapt to the dynamic changes of transient and continuous noise signals. S2. Apply the cross-domain noise decoupling algorithm to the noise signals of each acquisition channel, analyze the feature correlation and difference of multi-channel noise signals, separate the noise interference between channels, optimize the coordinated processing of signals, and use the dynamic dimensionality reduction mechanism to convert the high-dimensional noise signal into compressed noise feature data in the low-dimensional feature space; S3. Based on the improved non-negative sparse matrix decomposition algorithm, the compressed noise feature data is decomposed twice. First, the low-rank signal matrix is ​​decomposed to represent the normal operation signal of the circuit board. Secondly, the abnormal noise features are extracted by decomposing the sparse matrix, and the noise source fingerprint library is constructed by combining the dynamic characteristics under different working modes. S4. Based on the noise source fingerprint library, a multi-dimensional dynamic tracking and positioning algorithm is applied, combined with the spatiotemporal distribution information of the functional area of ​​the circuit board, the location of the noise source is dynamically calculated, and the noise source positioning path is optimized by analyzing the spatiotemporal changes of the noise intensity, so as to preferentially identify and locate the most disturbing noise source; S5. Through the adaptive feedback optimization mechanism, according to the real-time collected noise signals and historical noise source data, the noise source fingerprint library is dynamically adjusted, and the noise source location model is iteratively optimized to adapt to the noise source location under complex working conditions of the circuit board; S6. Based on the matching results of historical noise data and the current noise source fingerprint library, predict the noise sources that may be generated in the future and take noise suppression measures in advance; The S3 specifically includes: S31, based on the improved non-negative sparse matrix decomposition algorithm, the compressed global noise feature matrix Decompose and extract the low-rank signal matrix and the sparse noise matrix : ； in, for norm, represents the nuclear norm of the matrix, for Regularization term, and is the regularization coefficient; S32, sparse noise matrix Represents abnormal noise characteristics, and sparsity is optimized through an adaptive control mechanism: ； in, for norm, Control the sparsity of sparse matrices, The term is used to enhance the adaptability of sparse matrices; S33, combining the dynamic characteristics of the circuit board in different working modes, and constructing a noise source fingerprint library by decomposing the abnormal noise characteristics, wherein the noise source fingerprint library is generated by clustering and anomaly detection of sparse matrices; S34. Dynamically update the noise source fingerprint library through an iterative feedback mechanism: ； in, is the updated sparse matrix, is the previous noise source fingerprint library matrix, is the learning rate, is the gradient adjustment item in the noise source fingerprint library.

2. A circuit board noise source location method based on sparse matrix decomposition according to claim 1, characterized in that: The S1 specifically includes: S11, deploying micro sensor nodes in multiple functional areas of the circuit board, wherein the sensor nodes form a multi-layer noise sensing network to synchronously collect multi-channel noise signals in the area in real time; S12. Through the adaptive dynamic frequency domain sampling algorithm, the sensor node automatically adjusts the sampling frequency according to the transient and continuous characteristics of the noise signal. And the acquisition mode: ； in, is the minimum sampling frequency, is the maximum sampling frequency, is a coefficient that is dynamically adjusted according to the change of the noise signal; S13, the sensor node performs feature analysis based on the frequency, amplitude and duration of the collected noise signal, automatically distinguishes between transient noise signals and continuous noise signals, and for transient noise signals, the sensor node captures its changing characteristics through the set sampling frequency, and for continuous noise signals, avoids data redundancy by adjusting the sampling frequency, and optimizes the use of system resources; S14. During the signal acquisition process, the sensor node uses the spectrum characteristics of the noise signal to Dynamically update sampling parameters, the spectrum characteristics reflect the frequency changes of noise signals over time, through real-time analysis According to the changing trend of the sensor node, when the high-frequency noise signal appears, the sensor node sets an appropriate sampling frequency, and when the noise signal tends to be stable, the sampling frequency is adjusted in time to avoid unnecessary data redundancy; S15. Each sensor node dynamically adjusts the collection mode according to the noise feature classification, and combines with the multi-layer perception network of the circuit board to achieve comprehensive collection and processing of noise signals and adapt to the complex noise environment in different areas of the circuit board.

3. The circuit board noise source localization method based on sparse matrix decomposition according to claim 1, characterized in that: The S2 specifically includes: S21, preliminarily process the noise signal of each acquisition channel to generate a matrix related to the characteristics of the noise signal ,in For in time Down, channel The acquired noise signal characteristic data matrix; S22. By analyzing the correlation of noise signals between different channels, a cross-channel collaborative processing model is constructed to define the channels. With channel The correlation function between for: ； in, and Channel and Channel The mean value of the noise signal, and is its standard deviation; S23. Based on the results of the inter-channel correlation analysis, a cross-domain noise decoupling algorithm is applied to eliminate the noise interference effect between different channels. By constructing a dimensionality reduction mapping of the noise signal, the high-dimensional signal Project to low-dimensional feature space to generate low-dimensional noise feature matrix : ； in, is the dimension reduction mapping matrix; S24. After cross-channel collaborative processing, the low-dimensional noise feature matrices of all channels are merged to generate a global noise feature matrix .

4. The circuit board noise source location method based on sparse matrix decomposition according to claim 1, characterized in that: The S4 specifically includes: S41, based on noise source fingerprint library and global noise feature matrix , the multi-dimensional dynamic tracking and positioning algorithm is applied, combined with the spatiotemporal distribution information of different functional areas of the circuit board, the location of the noise source is dynamically calculated, and the initial location of the noise source is determined by analyzing the spatiotemporal characteristics of the noise signal in different areas of the circuit board. ; S42. Optimize the location path of the noise source according to the temporal and spatial changes of the noise source signal strength: ； in, For the optimized position, is the gradient of the noise signal intensity, is the adaptive step size; S43, by analyzing the strength and interference weight of the noise source signal , prioritize the noise sources: ； in, is the signal strength of the noise source, is the noise interference weight, is the noise propagation distance; S44. Combined with the temporal and spatial distribution characteristics of the functional areas of the circuit board, multiple noise sources are synchronously located and optimized, so that the circuit board can effectively manage the location and impact of multiple noise sources in a complex noise environment.

5. The circuit board noise source location method based on sparse matrix decomposition according to claim 1, characterized in that: The S5 specifically includes: S51, through the adaptive feedback optimization mechanism, combined with the real-time acquisition of noise signals and historical noise source data , construct an adaptive weighted noise source model, and assign weights according to the frequency, amplitude and interference of the noise signal through a dynamic weighting algorithm , dynamically adjust the noise source characteristics: ； in, is the weighting coefficient, is the adjustment coefficient, is the signal change gradient; S52, iteratively optimize the noise source positioning model, combined with the adaptive path optimization mechanism, by introducing the time-varying characteristics of the noise signal, the historical and real-time noise signals are integrated into the path optimization model: ； in, is the time-varying weighting coefficient, is the second-order time derivative of the noise intensity; S53. Aiming at the working conditions of complex circuit boards, the noise source fingerprint library is dynamically expanded through an adaptive acceleration feedback mechanism to process noise signals in different frequency ranges and form a multi-layer nested weighted feature model. S54. Introduce a multi-level noise source feedback mechanism, build a spatiotemporal mapping model of the noise signal based on the real-time collected noise signal spectrum information, and use the model to update the data of the noise source fingerprint library, so that the noise source positioning process can adapt to real-time changes in a complex working environment.

6. The circuit board noise source location method based on sparse matrix decomposition according to claim 1, characterized in that: The S6 specifically includes: S61, through historical noise data and the current noise source fingerprint library Matching can be performed to predict possible noise sources in the future: ； in, represents the noise source signal that may be generated in the future, and is the adjustment coefficient, is the gradient change of historical noise data, is the acceleration term of the current noise source; S62, based on the matching results of the noise source fingerprint library, deploy noise suppression measures in advance, according to the noise signal strength and the location of the noise source Calculate the strength of the suppression measure: ； in, Indicates the intensity of the suppression measures, and is the adjustment coefficient, is the propagation distance of the noise signal, and are the future noise source position and the current noise source position respectively; S63. By predicting the noise sources that may be generated in the future, the system deploys noise suppression strategies in advance, adjusts the suppression priority according to the interference and intensity of different noise sources, and dynamically manages the noise sources; S64. After the noise suppression strategy is implemented, the noise source fingerprint library is updated based on the feedback mechanism, and the noise sources that may be generated in the future are continuously tracked and adjusted through the adaptive optimization mechanism.

Images