A method and equipment for monitoring the operating status of computer room equipment based on computer room noise.

Abstract

This application relates to a method and device for monitoring the operational status of data center equipment based on data center noise. The method includes: collecting audio data generated during the operation of the data center equipment; processing the data to extract frequency domain and time domain features; and finally inputting the time domain and frequency domain features into a pre-trained recognition model to obtain the operational status analysis results of the data center equipment, thereby providing a reference for the stability of data center operation. Because this technical solution analyzes the operational status of the data center equipment based on the audio data generated during operation, the monitoring capability is not limited when the system suddenly loses power or restarts.

Description

Technical Field

[0001] This application relates to the field of data center monitoring technology, and in particular to a method and device for monitoring the operating status of data center equipment based on data center noise. Background Technology

[0002] With the rapid development of information technology, data center monitoring systems have been widely used in data centers, server rooms, and other similar settings. These systems are typically used to monitor various indicators and equipment status in the data center in real time to ensure the normal operation of the equipment and the security of the data.

[0003] Existing data center monitoring systems rely on deploying numerous software services for system monitoring. However, when these software services malfunction, or when the system experiences a sudden power outage or restart, the monitoring capabilities of the software monitoring system become limited. Summary of the Invention

[0004] To at least partially overcome the problem that monitoring capabilities are limited in data center monitoring systems in related technologies when abnormalities occur, or when there is a power outage or restart, this application provides a method and device for monitoring the operating status of data center equipment based on data center noise.

[0005] The proposed solution is as follows:

[0006] According to a first aspect of the embodiments of this application, a method for monitoring the operating status of data center equipment based on data center noise is provided, including:

[0007] S1. Collect audio data generated during the operation of equipment in the computer room;

[0008] S2. Adjust the gain of the audio data;

[0009] S3. Normalize the audio data after gain adjustment.

[0010] S4. Filter the normalized audio data based on the preset initial low-frequency cutoff frequency and initial high-frequency cutoff frequency.

[0011] S5. Determine the normal noise power and peak noise power of the audio data after gain adjustment, and the normal signal power and peak signal power of the audio data after filtering.

[0012] S6. Calculate the normal signal-to-noise ratio based on the normal noise power and the normal signal power; calculate the peak signal-to-noise ratio based on the peak noise power and the peak signal power; calculate the signal-to-noise mean square error based on the gain-adjusted audio data and the filtered audio data;

[0013] S7. Determine whether the normal signal-to-noise ratio, the peak signal-to-noise ratio, and the signal-to-noise mean square error meet the preset requirements; if they meet the requirements, output the processed audio data; if they do not meet the requirements, adjust the initial low-frequency cutoff frequency and the initial high-frequency cutoff frequency and then execute steps S4-S6 again until the preset requirements are met.

[0014] S8. After normalizing the audio data output in step S7, divide it into multiple time windows based on the preset frame rate.

[0015] S9. Extract frequency domain features based on the audio data output in step S7; extract time domain features based on the audio data processed in step S8;

[0016] S10. Input the time-domain features and the frequency-domain features into the pre-trained recognition model to obtain the current operating status analysis results of the equipment in the computer room.

[0017] Preferably, the method further includes:

[0018] Acquire sample data; the sample data includes audio data of the computer room equipment under various operating conditions;

[0019] The time-domain and frequency-domain features of the sample data are extracted through steps S2-S9;

[0020] The recognition model is trained using the time-domain and frequency-domain features of the sample data, as well as the operating status of the corresponding computer room equipment, as training data.

[0021] Preferably, the method further includes:

[0022] Create a multi-class CNN model; the CNN model includes convolutional layers, pooling layers, and fully connected layers;

[0023] The CNN model was compiled using the Adam optimizer, the multi-class cross-entropy loss function, and accuracy as evaluation metrics.

[0024] The recognition model is trained using the time-domain and frequency-domain features of the sample data, as well as the operating status of the corresponding computer room equipment, as training data.

[0025] Preferably, the method further includes:

[0026] The EarlyStopping callback is applied during the training of the recognition model to prevent overfitting.

[0027] Preferably, the method further includes:

[0028] Audio data generated during the operation of equipment in the computer room is collected based on a preset period to obtain the analysis results of the operating status of the equipment in the current period.

[0029] Preferably, the preset initial low-frequency cutoff frequency is 1000.

[0030] The preset initial high-frequency cutoff frequency is 5000.

[0031] Determining whether the normal signal-to-noise ratio, the peak signal-to-noise ratio, and the signal-to-noise mean square error meet preset requirements includes:

[0032] Determine whether the normal signal-to-noise ratio, the peak signal-to-noise ratio, and the signal-to-noise mean square error are lower than a preset threshold;

[0033] The threshold values ​​are: normal signal-to-noise ratio (SNR) of 20; peak SNR of 0.9; and SNR mean square error of 0.001.

[0034] Preferably, determining the normal noise power and peak noise power of the gain-adjusted audio data, and the normal signal power and peak signal power of the filtered audio data, includes:

[0035] The gain-adjusted audio data is divided into multiple time windows based on a preset frame rate;

[0036] The signal within each time window is squared, the average value is calculated, and the short-time energy is obtained as the normal noise power; the maximum value is calculated, and the peak value is obtained as the peak noise power.

[0037] The filtered audio data is divided into multiple time windows based on a preset frame rate;

[0038] The signal within each time window is squared, and the average value is calculated to obtain the short-time energy as the normal signal power; the maximum value is calculated to obtain the peak value as the peak signal power.

[0039] Preferably, extracting time-domain features includes:

[0040] The signal within each time window is squared, and the average value is calculated to obtain the short-time energy.

[0041] The zero-crossing rate is obtained by calculating the number of times the signal crosses zero within each time window based on the short-time energy.

[0042] The autocorrelation function of the signal within each time window is calculated based on the zero-crossing rate to obtain short-time autocorrelation features as time-domain features.

[0043] Preferably, extracting frequency domain features includes:

[0044] Perform a Fast Fourier Transform on the audio data output in step S7 to obtain the spectrum.

[0045] The spectrum is processed by a Mel frequency filter bank to obtain a set of filtered spectrums.

[0046] Calculate the power spectrum of each filtered spectrogram;

[0047] The logarithm of the power spectrum and the discrete cosine transform value are calculated to obtain the Mel frequency cepstral coefficients as frequency domain features.

[0048] According to a second aspect of the embodiments of this application, a monitoring device for the operating status of data center equipment based on data center noise is provided, comprising:

[0049] Processor and memory;

[0050] The processor and memory are connected via a communication bus:

[0051] The processor is used to call and execute the program stored in the memory;

[0052] The memory is used to store a program, which is at least used to execute a method for monitoring the operating status of computer room equipment based on computer room noise as described in any of the above.

[0053] The technical solution provided in this application may include the following beneficial effects:

[0054] The sounds produced by data center equipment during normal operation differ from those produced during abnormal states such as restarts, power outages, and overloads. This technical solution leverages this characteristic by collecting audio data generated during the operation of the data center equipment, processing it, and extracting frequency and time domain features. These features are then input into a pre-trained recognition model to obtain an analysis of the current operating status of the data center equipment, thus providing a reference for the stability of the data center operation. Because this technical solution analyzes the operating status of the data center equipment based on the audio data generated during operation, the monitoring capability is not limited during sudden power outages or restarts.

[0055] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0056] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0057] Figure 1 This is a flowchart illustrating a method for monitoring the operating status of computer room equipment based on computer room noise, provided in one embodiment of this application.

[0058] Figure 2This is a schematic diagram of the structure of a data center equipment operation status monitoring device based on data center noise, provided in one embodiment of this application.

[0059] Reference numerals: Processor-21; Memory-22. Detailed Implementation

[0060] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0061] Example 1

[0062] A method for monitoring the operational status of computer room equipment based on computer room noise, referring to... Figure 1 ,include:

[0063] S1. Collect audio data generated during the operation of equipment in the computer room;

[0064] It should be noted that, in order to monitor the operating status of the equipment in the data center in real time, this embodiment collects audio data generated during the operation of the equipment in the data center based on a preset period to obtain the analysis results of the operating status of the equipment in the data center within the current period.

[0065] In practice, the preset period can be, but is not limited to, 3 minutes.

[0066] Audio data generated by the equipment in the computer room during operation can be collected through microphones, thus comprehensively capturing the operating noise of the equipment in the computer room.

[0067] Preferably, audio data with a duration of at least 30 seconds is collected.

[0068] S2. Adjust the gain of the audio data;

[0069] The acquired audio data undergoes gain adjustment: the audio data is saved as an audio file, a gain factor is set (e.g., increasing the gain by 10dB), and decibels are converted to linear gain. The audio data is then loaded, the gain is applied to the audio data, generating the audio data with the 10dB increase, and the audio data is exported as an audio file.

[0070] S3. Normalize the audio data after gain adjustment.

[0071] The gain-adjusted audio data is normalized to the range [-1, 1] to facilitate subsequent processing.

[0072] S4. Filter the normalized audio data based on the preset initial low-frequency cutoff frequency and initial high-frequency cutoff frequency.

[0073] Preferably, the preset initial low-frequency cutoff frequency is 1000.

[0074] The preset initial high-frequency cutoff frequency is 5000.

[0075] Define a bandpass filter with a low-frequency cutoff of 1000 Hz and a high-frequency cutoff of 5000 Hz. Apply this filter to the normalized audio data to remove or reduce background noise, interference, or other frequency components in the sound signal.

[0076] S5. Determine the normal noise power and peak noise power of the audio data after gain adjustment, and the normal signal power and peak signal power of the audio data after filtering.

[0077] Determine the normal noise power and peak noise power of the gain-adjusted audio data, and the normal signal power and peak signal power of the filtered audio data, including:

[0078] The gain-adjusted audio data is divided into multiple time windows based on a preset frame rate;

[0079] The signal within each time window is squared, the average value is calculated, and the short-time energy is obtained as the normal noise power; the maximum value is calculated, and the peak value is obtained as the peak noise power.

[0080] The filtered audio data is divided into multiple time windows based on a preset frame rate;

[0081] The signal within each time window is squared, and the average value is calculated to obtain the short-time energy as the normal signal power; the maximum value is calculated to obtain the peak value as the peak signal power.

[0082] The preset frame rate can be, but is not limited to, 250 frames per second.

[0083] S6. Calculate the normal signal-to-noise ratio based on the normal noise power and normal signal power; calculate the peak signal-to-noise ratio based on the peak noise power and peak signal power; calculate the signal-to-noise mean square error based on the audio data after gain adjustment and the audio data after filtering.

[0084] This step calculates the quantization metrics before and after filtering:

[0085] Calculate the normal signal-to-noise ratio (SNR): SNR = 10 * log10 (normal signal power / normal noise power);

[0086] Calculate Peak Signal-to-Noise Ratio (PNSR): PNSR = Peak Signal Power / Peak Noise Power;

[0087] Calculate the mean square error (MSE): MSE = 1 / N*Σ[(noise signal (i) corresponding to the audio data after gain adjustment - filtered signal (i) corresponding to the audio data after filtering)^2].

[0088] S7. Determine whether the normal signal-to-noise ratio, peak signal-to-noise ratio, and signal-to-noise mean square error meet the preset requirements; if they do, output the processed audio data; if they do not, adjust the initial low-frequency cutoff frequency and the initial high-frequency cutoff frequency and then execute steps S4-S6 again until the preset requirements are met.

[0089] Determine whether the normal signal-to-noise ratio, peak signal-to-noise ratio, and mean square error of signal-to-noise ratio meet the preset requirements, including:

[0090] Determine whether the normal signal-to-noise ratio, peak signal-to-noise ratio, and signal-to-noise mean square error are lower than preset thresholds;

[0091] The threshold values ​​are: normal signal-to-noise ratio (SNR) of 20; peak SNR of 0.9; and SNR mean square error of 0.001.

[0092] In this technical solution, the threshold values ​​for each parameter of the quantitative indicators of the data center environment are as follows: SNR threshold is 20, PNSR threshold is 0.9, and MSE threshold is 0.001.

[0093] S8. After normalizing the audio data output in step S7, divide it into multiple time windows based on the preset frame rate.

[0094] It should be noted that the preset frame rate here is also 250 frames per second.

[0095] S9. Extract frequency domain features based on the audio data output in step S7; extract time domain features based on the audio data processed in step S8;

[0096] Extracting temporal features, including:

[0097] The signal within each time window is squared, and the average value is calculated to obtain the short-time energy.

[0098] The zero-crossing rate is obtained by calculating the number of times the signal crosses zero within each time window based on the short-time energy.

[0099] The autocorrelation function of the signal within each time window is calculated based on the zero-crossing rate to obtain short-time autocorrelation features as time-domain features.

[0100] Extracting frequency domain features includes:

[0101] Perform a Fast Fourier Transform on the audio data output in step S7 to obtain the spectrum.

[0102] The spectrum is processed by a Mel frequency filter bank to obtain a set of filtered spectrums.

[0103] Calculate the power spectrum of each filtered spectrogram;

[0104] The logarithm of the power spectrum and the discrete cosine transform value are calculated to obtain the Mel frequency cepstral coefficients as frequency domain features.

[0105] S10. Input the time-domain features and frequency-domain features into the pre-trained recognition model to obtain the current operating status analysis results of the equipment in the computer room.

[0106] It should be noted that the method also includes:

[0107] Obtain sample data; the sample data includes audio data of the computer room equipment under various operating conditions;

[0108] Extract the time-domain and frequency-domain features of the sample data through steps S2-S9;

[0109] The time-domain and frequency-domain features of the sample data, as well as the operating status of the corresponding computer room equipment, are used as training data to train the recognition model.

[0110] Sample data can be obtained by repeating steps S1-S9 n times (n>=10) and recording the actual operating status of the data center equipment in n times. The time domain and frequency domain feature values ​​of each time are labeled with status (the labeled dataset needs to meet all operating states of the data center equipment, such as normal operation, restart, power failure, overload, etc.).

[0111] Then create a multi-class CNN model; the CNN model includes convolutional layers, pooling layers, and fully connected layers;

[0112] The CNN model was compiled using the Adam optimizer, the multi-class cross-entropy loss function, and accuracy as evaluation metrics.

[0113] The time-domain and frequency-domain features of the sample data, as well as the operating status of the corresponding computer room equipment, are used as training data to train the recognition model.

[0114] Preferably, the time-domain and frequency-domain features extracted from the sample data are standardized, and the features are then merged.

[0115] During the training process, the sample data is divided into a training set and a test set.

[0116] The training set is used to train the recognition model, while the test set is used to test the trained recognition model. Only after passing the test can the model be put into use.

[0117] During the training of the recognition model, the EarlyStopping callback is used to prevent overfitting. The accuracy on the output test set is then used to adjust the model's hyperparameters, such as learning rate, batch size, number of layers, and number of hidden units, to find the optimal configuration.

[0118] In practice, the recognition model is deployed on a server, and audio data of at least 30 seconds generated by the computer room equipment during operation is collected every 3 minutes. After processing the audio data as described in steps S2-S9, the time domain features and frequency domain features are input into the pre-trained recognition model to obtain the current operating status analysis results of the computer room equipment.

[0119] The current operational status analysis results of the equipment in the data center can be broadcast to the maintenance personnel via voice, thereby enabling data center monitoring.

[0120] It is understandable that the sounds generated by data center equipment during normal operation differ from those generated during abnormal states such as restarts, power outages, and overloads. Based on this characteristic, this technical solution collects audio data generated by the data center equipment during operation, processes it to extract frequency and time domain features, and finally inputs these features into a pre-trained recognition model to obtain the operational status analysis results of the data center equipment. This provides a reference for the stability of data center operations. Because this technical solution analyzes the operational status of the data center equipment based on the audio data generated during operation, monitoring capabilities are not limited during sudden power outages or restarts.

[0121] Example 2

[0122] A monitoring device for the operating status of computer room equipment based on computer room noise, referring to Figure 2 ,include:

[0123] Processor 21 and memory 22;

[0124] Processor 21 and memory 22 are connected via a communication bus:

[0125] The processor 21 is used to call and execute the program stored in the memory 22;

[0126] The memory 22 is used to store a program, which is used to execute at least one of the above embodiments of a method for monitoring the operating status of computer room equipment based on computer room noise.

[0127] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0128] It should be noted that in the description of this application, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this application, unless otherwise stated, "a plurality of" means at least two.

[0129] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the function involved, as will be understood by those skilled in the art to which embodiments of this application pertain.

[0130] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0131] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0132] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0133] The storage media mentioned above can be read-only memory, disk, or optical disk, etc.

[0134] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0135] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for monitoring the operating status of computer room equipment based on computer room noise, characterized in that, include: S1. Collect audio data generated during the operation of equipment in the computer room; S2. Adjust the gain of the audio data; S3. Normalize the audio data after gain adjustment. S4. Filter the normalized audio data based on the preset initial low-frequency cutoff frequency and initial high-frequency cutoff frequency. S5. Determine the normal noise power and peak noise power of the audio data after gain adjustment, and the normal signal power and peak signal power of the audio data after filtering. S6. Calculate the normal signal-to-noise ratio based on the normal noise power and the normal signal power; calculate the peak signal-to-noise ratio based on the peak noise power and the peak signal power; calculate the signal-to-noise mean square error based on the gain-adjusted audio data and the filtered audio data; S7. Determine whether the normal signal-to-noise ratio, the peak signal-to-noise ratio, and the signal-to-noise mean square error meet the preset requirements; if they do, output the processed audio data. If the requirements are not met, adjust the initial low-frequency cutoff frequency and the initial high-frequency cutoff frequency and then execute steps S4-S6 again until the preset requirements are met. S8. After normalizing the audio data output in step S7, divide it into multiple time windows based on the preset frame rate. S9. Extract frequency domain features based on the audio data output in step S7; Temporal features are extracted from the audio data processed in step S8. S10. Input the time-domain features and the frequency-domain features into the pre-trained recognition model to obtain the current operating status analysis results of the equipment in the computer room; The preset initial low-frequency cutoff frequency is 1000. The preset initial high-frequency cutoff frequency is 5000. Determining whether the normal signal-to-noise ratio, the peak signal-to-noise ratio, and the signal-to-noise mean square error meet preset requirements includes: Determine whether the normal signal-to-noise ratio, the peak signal-to-noise ratio, and the signal-to-noise mean square error are lower than a preset threshold; The threshold values ​​for normal signal-to-noise ratio (SNR) are 20, peak SNR are 0.9, and mean square error of signal-to-noise ratio (MSNR) are 0.

001.

2. The method according to claim 1, characterized in that, The method further includes: Acquire sample data; the sample data includes audio data of the computer room equipment under various operating conditions; The time-domain and frequency-domain features of the sample data are extracted through steps S2-S9; The recognition model is trained using the time-domain and frequency-domain features of the sample data, as well as the operating status of the corresponding computer room equipment.

3. The method according to claim 2, characterized in that, The method further includes: Create a multi-class CNN model; the CNN model includes convolutional layers, pooling layers, and fully connected layers; The CNN model was compiled using the Adam optimizer, the multi-class cross-entropy loss function, and accuracy as evaluation metrics. The recognition model is trained using the time-domain and frequency-domain features of the sample data, as well as the operating status of the corresponding computer room equipment.

4. The method according to claim 3, characterized in that, The method further includes: The EarlyStopping callback is applied during the training of the recognition model to prevent overfitting.

5. The method according to claim 1, characterized in that, The method further includes: The system collects audio data generated by the equipment in the computer room during operation at a preset period to obtain the analysis results of the equipment's operating status within the current period.

6. The method according to claim 1, characterized in that, Determine the normal noise power and peak noise power of the gain-adjusted audio data, and the normal signal power and peak signal power of the filtered audio data, including: The gain-adjusted audio data is divided into multiple time windows based on a preset frame rate; The signal within each time window is squared, the average value is calculated, and the short-time energy is obtained as the normal noise power; the maximum value is calculated, and the peak value is obtained as the peak noise power. The filtered audio data is divided into multiple time windows based on a preset frame rate; The signal within each time window is squared, and the average value is calculated to obtain the short-time energy as the normal signal power; the maximum value is calculated to obtain the peak value as the peak signal power.

7. The method according to claim 6, characterized in that, Extracting temporal features includes: The signal within each time window is squared, and the average value is calculated to obtain the short-time energy. The zero-crossing rate is obtained by calculating the number of times the signal crosses zero within each time window based on the short-time energy. The autocorrelation function of the signal within each time window is calculated based on the zero-crossing rate to obtain short-time autocorrelation features as time-domain features.

8. The method according to claim 6, characterized in that, Extracting frequency domain features includes: Perform a Fast Fourier Transform on the audio data output in step S7 to obtain the spectrum. The spectrum is processed by a Mel frequency filter bank to obtain a set of filtered spectrums. Calculate the power spectrum of each filtered spectrogram; The logarithm of the power spectrum and the discrete cosine transform value are calculated to obtain the Mel frequency cepstral coefficients as frequency domain features.

9. A monitoring device for the operating status of computer room equipment based on computer room noise, characterized in that, include: Processor and memory; The processor and memory are connected via a communication bus: The processor is used to call and execute the program stored in the memory; The memory is used to store a program, which is at least used to execute the method for monitoring the operating status of computer room equipment based on computer room noise as described in any one of claims 1-8.

Images