A power supply monitoring method for atmosphere light bar production
By adaptively adjusting the truncation order of the SVD filter and combining multiple data to determine the target truncation order, the problem of poor filtering effect caused by the fixed truncation order in the existing technology is solved, and accurate monitoring and anomaly identification of power supply signals are realized.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- DONGGUAN WELLMEI MOLD MFG CO LTD
- Filing Date
- 2026-01-21
- Publication Date
- 2026-06-09
AI Technical Summary
In the existing technology, the signal denoising method based on singular value decomposition (SVD) cannot adapt to different noise intensities and signal dynamic characteristics due to the fixed truncation order, resulting in poor filtering effect and affecting the accuracy of power supply anomaly identification.
By acquiring the current and historical noise sequences of the power supply signal, the adaptive energy threshold and cumulative energy ratio are calculated, the truncation order of the SVD filter is dynamically adjusted, and the target truncation order is determined by combining multiple data sources, thereby enabling signal reconstruction and anomaly detection.
Adaptive SVD filtering was implemented, avoiding filtered waves and under-filtering, improving the noise reduction effect of the power supply signal, and ensuring accurate monitoring of power supply anomalies.
Smart Images
Figure CN122173791A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology. More specifically, this invention relates to a power supply monitoring method for the production of ambient light strips. Background Technology
[0002] With the popularization of smart lighting technology, ambient light strips, as multi-scene adaptable lighting devices, have their optical performance and lifespan directly affected by their production quality. In industrial production, a stable power supply signal is crucial to ensuring the consistency of LED light emission and delaying light decay; therefore, accurate monitoring of the power supply signal is essential. However, electromagnetic interference, high-frequency equipment noise, and line coupling noise commonly found in the production environment can easily lead to power supply signal distortion. Therefore, eliminating noise interference is a prerequisite for ensuring accurate power supply signal detection.
[0003] In related technologies, signal denoising methods based on Singular Value Decomposition (SVD) can suppress noise in power supply signals. The basic principle is as follows: First, the power supply signal sequence is constructed as a Hankel matrix, and SVD decomposition is used to obtain the singular value matrix representing the signal energy distribution. Then, based on the characteristic that noise energy is concentrated in smaller singular values, minor singular values other than the first k principal singular values are truncated, thereby suppressing the noise-dominated high-rank subspace and preserving the low-rank subspace where the principal components of the signal reside. This method effectively filters out various types of mixed noise, such as power frequency harmonic distortion, random impulse noise, and high-frequency electromagnetic interference, while retaining core features such as the fundamental frequency and harmonic components, by separating the statistical differences between the signal and noise. This provides high signal-to-noise ratio preprocessed data for power supply signal anomaly detection. Therefore, this filtering method can be used to filter power supply signals to eliminate the influence of noise.
[0004] However, the cutoff order set in traditional SVD filtering methods is usually a fixed value, which makes it difficult to adapt to different noise intensities and signal dynamic characteristics. This can easily lead to either filtered waves (too low an order, resulting in loss of signal features) or under-filtered waves (too high an order, resulting in residual noise), resulting in poor filtering performance and thus affecting the accuracy of abnormal data identification. Summary of the Invention
[0005] To address the problem that when using the SVD filtering method to filter power supply signals, the fixed cutoff order cannot adapt to different noise intensities and signal dynamic characteristics, resulting in poor filtering performance and thus failing to accurately identify power supply anomalies, this invention provides a power supply monitoring method for ambient light strip production. The method includes: Acquire the ambient light strip production power supply signal within the current time window and preprocess it to obtain the current estimated noise sequence; The power supply signal is converted into a Hankel matrix and singular value decomposition is performed. Based on the singular value matrix in the decomposition result, the cumulative energy ratio of each cutoff order is obtained. The cumulative energy ratio represents the ratio of the sum of squares of the singular values retained at the corresponding cutoff order to the sum of squares of all singular values in the singular value matrix. The ratio of the noise energy of the current estimated noise sequence to that of multiple previous historical estimated noise sequences is used as the current noise level. An adaptive energy threshold is calculated, which is negatively correlated with the current noise level and the degree of clustering of data in the current estimated noise sequence. The minimum truncation order at which the corresponding cumulative energy ratio is greater than the adaptive energy threshold is used as the target truncation order. Based on the target truncation order, each matrix in the decomposition result is truncated and reconstructed to generate the reconstructed Hankel matrix. The target power supply signal is then obtained through a restoration operation, and anomalies are detected in the target power supply signal to identify power supply anomalies.
[0006] This invention determines the adaptive energy threshold by integrating data from multiple sources, ensuring the accuracy of the adaptive energy threshold. Furthermore, it uses the minimum truncation order corresponding to a cumulative energy ratio greater than the adaptive energy threshold as the target truncation order. This allows for dynamic adjustment of the truncation order while ensuring optimal denoising performance when performing SVD filtering based on the target truncation order. This not only avoids filtering waves and under-filtering but also guarantees the filtering effect, thereby reducing the impact of noise on the power supply anomaly identification results and achieving accurate monitoring of power supply anomalies.
[0007] Preferably, the current method for obtaining the estimated noise sequence includes: The power supply signal is low-pass filtered to obtain a denoised power supply signal, and the sequence formed by the difference between the corresponding data in the power supply signal and the denoised power supply signal is used as the current estimated noise sequence.
[0008] The current estimated noise sequence calculated by this invention can provide a data basis for subsequently determining the adaptive energy threshold.
[0009] Preferably, the historical noise sequence is generated based on the power supply signals collected in historical neighboring windows prior to the current time window; the current noise level satisfies the following relationship: ; In the formula, The current noise level; For the currently estimated noise sequence, the th The possible values of each data point; This represents the number of sampling points within the sampling time window. For the first The nth historical estimated noise sequence The possible values of each data point; Estimate the number of noisy sequences in history; This is the normalization function.
[0010] This invention constructs a relative quantitative index of noise intensity, namely the difference between the noise energy levels of the current estimated noise sequence and the historical estimated noise sequence, to calculate the current noise level. This can eliminate the interference of absolute noise energy fluctuations caused by environmental differences on the evaluation results, thereby achieving unbiased dynamic calibration of the noise level and ensuring the accuracy of the current noise level.
[0011] Preferably, the degree of clustering of data in the current estimated noise sequence is used as the energy focus of the current estimated noise sequence. The method for obtaining the energy focus includes: Determine the power spectrum of the current noise sequence, and select the first few peaks from the power spectrum as the main peaks in descending order of power value; By combining the energy proportion of the main peaks, the inter-peak difference, and the peak isolation, the energy focus of the current noise sequence is calculated using a weighted multiplication formula. Among them, energy percentage represents the ratio of the sum of the power values of the main peaks to the total power of the power spectrum; interpeak difference represents the ratio of the standard deviation to the range of the power values of the main peaks; peak isolation represents the average difference between the power value of the main peak and the power values of the two adjacent peaks.
[0012] This invention combines data from different angles to calculate the energy focus of the current estimated noise sequence, ensuring the accuracy of the calculation results.
[0013] Preferably, the energy focus satisfies the following relationship: ; In the formula, The energy focus of the current noise sequence; The power spectrum of the current noise sequence is the first... The power values of the main peaks; The number of main peaks selected; The first power in this power spectrum Power values at each frequency; This represents the total number of frequencies in the power spectrum. The variance of the power values for all major peaks; The range of power values for all major peaks; The first power in this power spectrum The average power value of the two adjacent peaks to the left and right of the main peak.
[0014] Preferably, the adaptive energy threshold satisfies the following relationship: ; In the formula, An adaptive energy threshold; The energy focus of the current noise sequence; This represents the current noise level.
[0015] Preferably, obtaining the cumulative energy ratio for each cutoff order includes: Extract the diagonal elements of the singular value matrix sequentially from the top left to the bottom right to form a singular value array; Starting from the first element of the singular value array, select the same number of elements as the preset truncation order, calculate the ratio of the sum of squares of all selected elements to the sum of squares of all elements in the singular value array, and obtain the cumulative energy ratio of the truncation order.
[0016] This invention selects a target cutoff order based on the cumulative energy ratio, ensuring that the selected target cutoff order has the best filtering effect, thereby improving the filtering effect.
[0017] Preferably, the method for obtaining the reconstructed Hankel matrix includes: By retaining the first few column vectors of the left singular matrix, singular value matrix, and right singular matrix in the decomposition results, we obtain the reconstructed left singular matrix, reconstructed singular value matrix, and reconstructed right singular matrix; wherein, the number of column vectors retained is the same as the target truncation order; The reconstructed left singular matrix, the reconstructed singular value matrix, and the reconstructed right singular matrix are multiplied sequentially to obtain the reconstructed Hankel matrix.
[0018] Preferably, obtaining the target power supply signal through a restoration operation includes: Obtain the elements of each anti-diagonal line in the reconstructed Hankel matrix, and take the average value of each anti-diagonal line element as a signal value. Obtain the sequence of signal values corresponding to all anti-diagonal lines to obtain the target power supply signal.
[0019] Preferably, power supply anomalies are identified by detecting anomalies in the target power supply signal, including: The box plot method is used to detect anomalies in the target power supply signal, and power supply anomalies are identified based on the anomaly detection results.
[0020] The present invention has the following effects: On the one hand, this invention dynamically adjusts the truncation order when performing SVD filtering on the power supply signal through an adaptive energy threshold, which can avoid overfitting and underfitting, thereby improving the filtering effect. On the other hand, this invention determines the adaptive energy threshold by integrating multiple data sources, ensuring the accuracy of the determination result. Furthermore, based on the relationship between the cumulative energy ratio and the adaptive energy threshold, a target truncation order is selected, ensuring that the filtering effect at the target truncation order is the optimal filtering effect, further guaranteeing the accuracy of the filtering effect. This effectively eliminates the influence of noise on the identification of power supply anomalies, enabling precise monitoring of power supply anomalies. Attached Figure Description
[0021] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent upon reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the invention are illustrated by way of example and not limitation, and like or corresponding reference numerals denote like or corresponding parts, wherein: Figure 1 This is a flowchart illustrating the steps of a power supply monitoring method for the production of ambient light strips according to an embodiment of the present invention. Detailed Implementation
[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0023] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
[0024] Reference Figure 1 A power supply monitoring method for the production of ambient light strips, comprising steps S1-S4, as detailed below: S1: Obtain the ambient light strip production power supply signal within the current time window and perform preprocessing to obtain the current estimated noise sequence.
[0025] It should be noted that when monitoring the power supply during the production of ambient light strips, the main task is to ensure the stability of the power supply voltage. Therefore, this invention selects the power supply voltage as the power supply signal to be collected. Of course, the power supply current or other signals can also be collected as the power supply signal.
[0026] The current time window refers to the sampling time window with the current moment as the end point; the current estimated noise sequence refers to the data sequence used to evaluate the noise energy level and distribution in the voltage signal, so as to provide a data basis for subsequent analysis and processing operations.
[0027] Optionally, the duration of the sampling time window can be set to... and at sampling frequency The power supply voltage of the ambient light strips during production is continuously collected within each sampling time window, thus providing a data foundation for subsequent analysis. In this embodiment... =1s, =1000Hz.
[0028] In an exemplary embodiment of the present invention, the determination of the current estimated noise sequence can be achieved through the following steps: The power supply signal is low-pass filtered to obtain a denoised power supply signal, and the sequence formed by the difference between the corresponding data in the power supply signal and the denoised power supply signal is used as the current estimated noise sequence.
[0029] It should be noted that low-pass filtering can quickly remove high-frequency switching noise and electromagnetic interference. Therefore, low-pass filtering can be used to filter power supply signals. The process of low-pass filtering the signal is existing technology and will not be described in detail in this embodiment.
[0030] Optionally, a low-pass filter with a cutoff frequency of 100Hz can be used to filter the power supply signal to obtain a denoised power supply signal. The current estimated noise sequence, composed of all the differences, can then be obtained by calculating the differences between corresponding data in the power supply signals before and after denoising. This embodiment does not impose any particular limitation on the cutoff frequency of the selected low-pass filter.
[0031] In another embodiment, a high-pass filter, a band-stop filter, or the like can also be used to filter the power supply signal. This embodiment does not impose any particular limitation on the selected filtering method.
[0032] S2: Convert the power supply signal into a Hankel matrix and perform singular value decomposition. Based on the singular value matrix in the decomposition result, obtain the cumulative energy ratio of each truncation order. The cumulative energy ratio represents the ratio of the sum of squares of the singular values retained at the corresponding truncation order to the sum of squares of all singular values in the singular value matrix.
[0033] It should be noted that the Hankel matrix is a special type of matrix where all elements on each subdiagonal are identical, and it can be used to represent the autocorrelation and cross-correlation of a signal. Singular Value Decomposition (SVD) decomposes a signal matrix into three submatrices: a left singular matrix, a singular value matrix, and a right singular matrix, to reconstruct the signal by projecting the original signal onto an eigenspace spanned by singular vectors. The elements on the main diagonal of the singular value matrix are called singular values.
[0034] Next, taking the power supply signal as a voltage signal as an example, the construction process of the Hankel matrix and the singular value decomposition process of the constructed Hankel matrix will be explained in detail: First, the voltage signal can be denoted as... , In the formula, It is a voltage signal. , These are the voltage values at the first and second sampling points of the voltage signal, respectively. This represents the number of sampling points within the sampling time window.
[0035] Then, based on the signal length of the voltage signal, the number of rows and columns of the Hankel matrix to be constructed can be determined. In this embodiment, the number of rows is set as follows: Satisfying the relation: , number of columns Satisfying the relation: In the formula, A function that returns the minimum value; This represents the number of sampling points within the sampling time window. The divisor symbol is used. represent The value after dividing by 3 and rounding down. It should be noted that, to avoid excessive computational complexity due to an excessively large number of rows in the Hankel matrix, this invention limits the number of rows and columns of the constructed Hankel matrix based on the signal length of the voltage signal.
[0036] Next, the number of rows in the Hankel matrix to be constructed can be used as a basis. The number of columns in the Hankel matrix to be constructed and voltage signal Construct the Hankel matrix of the voltage signal. The specific structure of this matrix is as follows: ; In the formula, The constructed Hankel matrix; The first voltage signal Voltage values at each sampling point.
[0037] Finally, the constructed Hankel matrix can be... Perform SVD decomposition to obtain the decomposition results, which satisfy the following relation: In the formula, It is a left singular matrix; It is a singular value matrix, which is a diagonal matrix and represents the energy distribution of the voltage signal; It is a right singular matrix. It should be noted that the process of converting the voltage signal into a Hankel matrix and the method for determining each matrix in the decomposition result are existing technologies, and will not be described in detail in this embodiment.
[0038] Furthermore, once the decomposition results are obtained, the cumulative energy ratio of each truncation order can be calculated based on the singular value matrix in the decomposition results. It should be noted that since the singular value matrix represents the energy distribution of the signal, the cumulative energy ratio of each truncation order can be calculated based on the singular value matrix.
[0039] In an exemplary embodiment of the present invention, the cumulative energy ratio of each cutoff order can be determined through the following steps: Step 1: Extract the diagonal elements of the singular value matrix sequentially from the top left to the bottom right to form a singular value array; Optional, the total number of data in the singular value array. Satisfying the relation: In the formula, A function that returns the minimum value; The number of rows in the Hankel matrix to be constructed; The number of columns is the number of columns in the Hankel matrix to be constructed.
[0040] It should be noted that the singular value array is constructed in this invention to facilitate the subsequent calculation of the cumulative energy ratio.
[0041] Step 2: Starting from the first element of the singular value array, select the same number of elements as the preset truncation order, calculate the ratio of the sum of squares of all selected elements to the sum of squares of all elements in the singular value array, and obtain the cumulative energy ratio of the truncation order.
[0042] Specifically, the cumulative energy ratio for any cutoff order satisfies the following relationship: ; In the formula, The truncation order is The cumulative energy ratio at that time; The singular value array is the first The possible values of each data point; To truncate the order, it should be noted that... The value of is related to the value of the adaptive energy threshold determined in subsequent steps, and since the calculation of the cumulative energy ratio of each cutoff order is for determining the target cutoff order, this invention hereby... The value of is not explained in detail here; it will be described in detail later when the target truncation order is determined. This represents the total number of data in the singular value array; This is the summation symbol.
[0043] It should be noted that since the principle of SVD filtering is to reconstruct the signal by retaining the principal singular values in the singular value matrix of the Hankel matrix to remove noise, and the number of principal singular values retained is the same as the truncation order, this invention can provide a reference for subsequent selection of the target truncation order by measuring the cumulative energy ratio under different truncation orders.
[0044] S3: The ratio of the noise energy of the current estimated noise sequence to that of multiple previous historical estimated noise sequences is taken as the current noise level. An adaptive energy threshold is calculated. The adaptive energy threshold is negatively correlated with the current noise level and the degree of clustering of data in the current estimated noise sequence. The minimum truncation order with a cumulative energy ratio greater than the adaptive energy threshold is taken as the target truncation order.
[0045] It should be noted that, since the fundamental and harmonic frequencies of the power supply signal exhibit a low-rank structure in the Hankel matrix, their energy is mainly concentrated on a few major singular values, and these major singular values account for a relatively high proportion of the total energy. In contrast, the energy of noise is dispersed across a large number of small singular values, which account for a relatively low proportion of the total energy. Therefore, this invention, based on this characteristic, uses the estimated energy distribution of the noise sequence to determine an adaptive energy threshold, which can provide a reference for determining the subsequent target truncation order.
[0046] The current noise level refers to data that measures the degree to which the acquired power supply signal is affected by noise within the current time window.
[0047] In one exemplary embodiment of the present invention, the historical estimated noise sequence is generated based on the power supply signals collected in historical neighboring windows prior to the current time window. It should be noted that the method for determining the historical estimated noise sequence is the same as the method for determining the current estimated noise sequence, and will not be repeated here.
[0048] Specifically, the current noise level satisfies the following relationship: ; In the formula, The current noise level; For the currently estimated noise sequence, the th The possible values of each data point; This represents the number of sampling points within the sampling time window. For the first The nth historical estimated noise sequence The possible values of each data point; In this embodiment, to estimate the number of historical noise sequences, =10; This is the normalization function.
[0049] in, It reflects the average level of noise energy in the referenced historical estimated noise sequence; It reflects the overall level of noise energy in the currently estimated noise sequence; This reflects the normalized value that represents the ratio of the noise energy of the current estimated noise sequence to that of several previous historical estimated noise sequences; when When the value of is large, it indicates that the noise energy of the current estimated noise sequence is larger than that of the historical estimated noise sequence, and the current noise level is relatively large. This can further indicate that the power supply signal obtained within the current time window is more affected by noise.
[0050] It should be noted that the current noise level only reflects the degree to which the power supply signal is affected by noise within the current time window, but it cannot provide information on the distribution of noise. Therefore, this invention also calculates the degree of clustering of data in the currently estimated noise sequence to measure the distribution information of noise, thereby providing data support for determining the adaptive energy threshold.
[0051] In an exemplary embodiment of the present invention, the degree of clustering of data in the current estimated noise sequence is used as the energy focus of the current estimated noise sequence. The energy focus of the current estimated noise sequence can be determined through the following steps: Step 1: Determine the power spectrum of the current noise sequence, and select the first few peaks from the power spectrum as the main peaks in descending order of power value; The power spectrum, also known as the power spectral density, has frequency on the horizontal axis and power on the vertical axis. It is defined as the signal power per unit frequency band and is used to represent how the signal power changes with frequency. It should be noted that the process of determining the power spectrum is prior art and will not be described in detail in this embodiment.
[0052] Optionally, the top three peaks in the power spectrum can be selected as the main peaks in descending order of power value. This embodiment does not impose a particular limitation on the number of main peaks selected.
[0053] Step 2: Combining the energy proportion of the main peaks, the inter-peak difference, and the peak isolation, calculate the energy focus of the current noise sequence using a weighted multiplication formula.
[0054] Among them, the energy percentage of the main peak represents the ratio of the sum of the power values of the main peaks to the total power of the power spectrum; the interpeak difference of the main peaks represents the ratio of the standard deviation to the range of the power values of the main peaks; and the peak isolation of the main peaks represents the average difference between the power value of the main peak and the power values of the two adjacent peaks.
[0055] Specifically, the energy focus of the current noise sequence satisfies the following relationship: ; In the formula, The energy focus of the current noise sequence; The power spectrum of the current noise sequence is the first... The power values of the main peaks; In this embodiment, the number of main peaks selected is... =3; The first power in this power spectrum Power values at each frequency; This represents the total number of frequencies in the power spectrum. The variance of the power values for all major peaks; The range is the power values of all major peaks, where the range is the difference between the maximum and minimum power values of all major peaks; The first power in this power spectrum The average power value of the two adjacent peaks to the left and right of the main peak.
[0056] in, It reflects the energy proportion of the main peaks. The larger the value, the more the data in the current estimated noise sequence is concentrated on a few frequency components, and the greater the corresponding energy focus.
[0057] It reflects the inter-peak difference of the main peaks. The larger the value, the smaller the range of the main peaks and the greater the fluctuation. This indicates that the data in the current estimated noise sequence is more concentrated on individual frequencies corresponding to the main peaks, and the corresponding energy focus is greater. This reflects the peak isolation of the main peak. The larger the value, the greater the difference between the main peak and its adjacent peaks. In other words, the more isolated the main peak is, the greater the confidence that the larger values in the current estimated noise sequence are concentrated on individual frequencies, and the greater the corresponding energy focus.
[0058] Next, the adaptive energy threshold can be calculated by combining the current noise level and the energy focus of the current estimated noise sequence. Specifically, the adaptive energy threshold satisfies the following relationship: ; In the formula, An adaptive energy threshold; Estimate the energy focus of the current noise sequence; This represents the current noise level.
[0059] in, The larger the value, the greater the current noise level. In this case, if... When the value is also large, it indicates that the power supply signal obtained within the current time window is greatly affected by noise and the noise distribution is relatively concentrated. Setting a smaller adaptive energy threshold can reduce the truncation order, thereby effectively suppressing noise and avoiding under-filtering.
[0060] Conversely, if The smaller the value, the lower the current noise level. In this case, if... When the value is also small, it indicates that the power supply signal obtained within the current time window is less affected by noise and the noise distribution is relatively dispersed. By setting a larger adaptive energy threshold, the truncation order can be increased, thereby effectively preserving useful information in the power supply signal and avoiding filtering waves.
[0061] In another embodiment, the relation can also be used: Calculate the adaptive energy threshold; where, This is the normalization function.
[0062] Next, the determination of the target truncation order will be explained in detail: First, obtain the adaptive energy threshold; then, from the cutoff order... Starting with 1, gradually increase the value (by 1 each time), and calculate each value sequentially according to the formula for calculating the cumulative energy ratio in step S2. The cumulative energy percentage under the current value, up to the present. The target cutoff order is obtained when the percentage of accumulated energy under the current value first exceeds the adaptive energy threshold. This value ensures that the filtering effect at the target truncation order is the best, thereby improving the filtering effect.
[0063] S4: Based on the target truncation order, each matrix in the decomposition result is truncated and reconstructed to generate the reconstructed Hankel matrix. The target power supply signal is obtained through the restoration operation, so as to identify power supply anomalies by detecting anomalies in the target power supply signal.
[0064] In an exemplary embodiment of the present invention, the reconstructed Hankel matrix can be determined through the following steps: By retaining the first few column vectors of the left singular matrix, singular value matrix, and right singular matrix in the decomposition results, we obtain the reconstructed left singular matrix, reconstructed singular value matrix, and reconstructed right singular matrix; wherein, the number of column vectors retained is the same as the target truncation order; The reconstructed left singular matrix, the reconstructed singular value matrix, and the reconstructed right singular matrix are multiplied sequentially to obtain the reconstructed Hankel matrix.
[0065] Specifically, the reconstructed Hankel matrix can be expressed as follows: ; In the formula, This is the reconstructed Hankel matrix; The order of truncation is set to the target. This is the reconstructed left singular matrix; The reconstructed singular value matrix; This is the reconstructed right singular matrix.
[0066] It should be noted that the process of reconstructing the Hankel matrix based on the truncation order is an existing technique.
[0067] In an exemplary embodiment of the present invention, the target power supply signal can be determined through the following steps: Obtain the elements of each anti-diagonal line in the reconstructed Hankel matrix, and take the average value of each anti-diagonal line element as a signal value. Obtain the sequence of signal values corresponding to all anti-diagonal lines to obtain the target power supply signal.
[0068] The target power supply signal refers to the denoised power supply signal. It should be noted that the process of restoring the reconstructed Hankel matrix to obtain the corresponding signal is existing technology and will not be described in detail here.
[0069] In one exemplary embodiment of the present invention, power supply anomaly identification can be achieved through the following steps: The box plot method is used to detect anomalies in the target power supply signal, and power supply anomalies are identified based on the anomaly detection results.
[0070] Optionally, anomaly detection algorithms such as LOF (Local Outlier Factor) and COF (Connectivity-Based Outlier Factor) can also be used to detect anomalies in the target power supply signal. This embodiment does not impose any particular limitation on the type of anomaly detection algorithm selected.
[0071] It should be noted that when using the box plot method to detect anomalies in the target power supply signal, the outliers obtained are the abnormal data in the power supply signal. Therefore, when a power supply anomaly is detected, it can be fed back to the staff in real time through a projection display system, and an alarm can be issued to adjust the power supply situation of the ambient light strip production in a timely manner to ensure production safety. The process of using the box plot method to identify abnormal data is existing technology and will not be described in detail in this embodiment.
[0072] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise explicitly specified.
[0073] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.
Claims
1. A power supply monitoring method for the production of ambient light strips, characterized in that, include: Acquire the ambient light strip production power supply signal within the current time window and preprocess it to obtain the current estimated noise sequence; The power supply signal is converted into a Hankel matrix and singular value decomposition is performed. Based on the singular value matrix in the decomposition result, the cumulative energy ratio of each truncation order is obtained. The cumulative energy ratio represents the ratio of the sum of squares of the singular values retained at the corresponding truncation order to the sum of squares of all singular values in the singular value matrix. The ratio of the noise energy of the current estimated noise sequence to that of multiple previous historical estimated noise sequences is used as the current noise level. An adaptive energy threshold is calculated, which is negatively correlated with the current noise level and the degree of clustering of data in the current estimated noise sequence. The minimum truncation order at which the corresponding cumulative energy ratio is greater than the adaptive energy threshold is used as the target truncation order. Based on the target truncation order, each matrix in the decomposition result is truncated and reconstructed to generate the reconstructed Hankel matrix. The target power supply signal is then obtained through a restoration operation, and anomalies are detected in the target power supply signal to identify power supply anomalies.
2. The power supply monitoring method for the production of ambient light strips according to claim 1, characterized in that, The method for obtaining the current estimated noise sequence includes: The power supply signal is low-pass filtered to obtain a denoised power supply signal, and the sequence formed by the difference between the corresponding data in the power supply signal and the denoised power supply signal is used as the current estimated noise sequence.
3. The power supply monitoring method for the production of ambient light strips according to claim 2, characterized in that, The historical estimated noise sequence is generated based on the power supply signals collected in historical neighboring windows prior to the current time window; the current noise level satisfies the following relationship: ； In the formula, The current noise level; For the currently estimated noise sequence, the th The possible values of each data point; This represents the number of sampling points within the sampling time window. For the first The nth historical estimated noise sequence The possible values of each data point; Estimate the number of noisy sequences in history; This is the normalization function.
4. The power supply monitoring method for the production of ambient light strips according to claim 1, characterized in that, The degree of clustering of data in the current estimated noise sequence is used as the energy focus of the current estimated noise sequence. The method for obtaining the energy focus includes: Determine the power spectrum of the current noise sequence, and select the first few peaks from the power spectrum as the main peaks in descending order of power value; By combining the energy proportion of the main peaks, the inter-peak difference, and the peak isolation, the energy focus of the current noise sequence is calculated using a weighted multiplication formula. Wherein, the energy percentage represents the ratio of the sum of the power values of the main peaks to the total power of the power spectrum; the inter-peak difference represents the ratio of the standard deviation to the range of the power values of the main peaks; and the peak isolation represents the average difference between the power value of the main peak and the power values of the two adjacent peaks.
5. The power supply monitoring method for the production of ambient light strips according to claim 4, characterized in that, The energy focusing degree satisfies the following relationship: ； In the formula, The energy focus of the current noise sequence; The power spectrum of the current noise sequence is the first... The power values of the main peaks; The number of main peaks selected; The first power in this power spectrum Power values at each frequency; This represents the total number of frequencies in the power spectrum. The variance of the power values for all major peaks; The range of power values for all major peaks; The first power in this power spectrum The average power value of the two adjacent peaks to the left and right of the main peak.
6. A power supply monitoring method for the production of ambient light strips according to claim 3 or 5, characterized in that, The adaptive energy threshold satisfies the following relationship: ； In the formula, An adaptive energy threshold; The energy focus of the current noise sequence; This represents the current noise level.
7. The power supply monitoring method for the production of ambient light strips according to claim 1, characterized in that, The process of obtaining the cumulative energy ratio for each cutoff order includes: Extract the diagonal elements of the singular value matrix sequentially from the top left to the bottom right to form a singular value array; Starting from the first element of the singular value array, select the same number of elements as the preset truncation order, calculate the ratio of the sum of squares of all selected elements to the sum of squares of all elements in the singular value array, and obtain the cumulative energy ratio of the truncation order.
8. The power supply monitoring method for the production of ambient light strips according to claim 1, characterized in that, The method for obtaining the reconstructed Hankel matrix includes: The first few column vectors of the left singular matrix, singular value matrix, and right singular matrix in the decomposition result are retained to obtain the reconstructed left singular matrix, reconstructed singular value matrix, and reconstructed right singular matrix; wherein the number of column vectors retained is the same as the target truncation order; The reconstructed left singular matrix, the reconstructed singular value matrix, and the reconstructed right singular matrix are multiplied sequentially to obtain the reconstructed Hankel matrix.
9. A power supply monitoring method for the production of ambient light strips according to claim 8, characterized in that, The process of obtaining the target power supply signal through the restoration operation includes: Obtain the elements of each anti-diagonal line in the reconstructed Hankel matrix, and take the average value of each anti-diagonal line element as a signal value. Obtain the sequence of signal values corresponding to all anti-diagonal lines to obtain the target power supply signal.
10. A power supply monitoring method for the production of ambient light strips according to claim 9, characterized in that, The step of identifying power supply anomalies by detecting anomalies in the target power supply signal includes: The target power supply signal is subjected to anomaly detection using the box plot method, and power supply anomalies are identified based on the anomaly detection results.