Intelligent vibration monitoring and adaptive optimization device based on neural network

Abstract

The present invention relates to the field of vibration monitoring, and in particular to an intelligent vibration monitoring and adaptive optimization device based on a neural network. The present invention provides a vibration monitoring unit, a data storage unit, an analysis and determination unit, a vibration verification unit, and an optimization unit. The vibration monitoring unit is used to determine whether to send an interrupt signal to wake up a microcontroller, the data storage unit is used to record the timestamp of each wake-up moment and store the timestamp in a data list, the analysis and determination unit is used to determine the submission status of the data, and perform feature extraction on the original vibration data to calculate the vibration trigger characterization coefficient and identify the vibration trigger source. The vibration verification unit calculates the vibration comparison similarity to determine the accuracy of the analysis and determination unit's identification, and the optimization unit is used to continuously collect new vibration samples and integrate the vibration samples into the existing model to improve the accuracy of the model and the sensitivity of vibration monitoring.

Description

Technical Field

[0001] The present invention relates to the field of vibration monitoring, and in particular to an intelligent vibration monitoring and self-adaptive optimization device based on a neural network. Background Art

[0002] In recent years, the development of neural networks has provided a powerful tool for data processing. Neural networks have been widely used in many fields such as image recognition, speech recognition, and natural language processing. In the field of vibration monitoring, in order to ensure the safety of underground high-voltage cables, communication optical fibers, pipeline systems, and important buildings near the ground, neural networks are combined with 4G networks to identify and classify different vibration patterns, enabling remote data transmission. Monitoring data can be uploaded to cloud servers in real time to facilitate remote monitoring and data analysis.

[0003] Chinese Patent Publication No. CN106534439A discloses a vibration monitoring method, device, and terminal. The method includes: collecting data on vibration generated by a terminal through a pressure sensor, wherein the vibration is generated by a motor after the terminal receives a test instruction from a PC, and the test instruction includes a vibration pattern and time interval; and sending the data to the PC, wherein the data is used by the PC to monitor whether the vibration generated by the terminal meets pre-set standards. The present invention also discloses a vibration monitoring device and terminal, which solves the problem in related technologies of inaccurate measurements caused by the inability to fundamentally measure vibration, achieving accurate measurement of motor vibration quality and improving user experience.

[0004] However, the prior art still has the following problems:

[0005] When monitoring vibration sources, most existing technologies are suitable for detecting large-scale vibration events, such as earthquakes and abnormal operation of large machinery. However, they have poor sensitivity when monitoring small regular vibration sources near the ground in real life, such as the vibration of large vehicle engines and the vibration of pile drivers. They are unable to accurately and timely determine the vibration source, resulting in low accuracy and high false alarm rate. Summary of the Invention

[0006] To this end, the present invention provides an intelligent vibration monitoring and adaptive optimization device based on a neural network to solve the problem of poor sensitivity when monitoring small regular vibration sources near the ground in real life, such as the vibration of large vehicle engines and the vibration during pile driver operations. The vibration source cannot be accurately and timely determined, resulting in low accuracy and high false alarm rate.

[0007] To achieve the above objectives, the present invention provides an intelligent vibration monitoring and adaptive optimization device based on a neural network, which comprises:

[0008] a vibration monitoring unit, configured to obtain a vibration signal and perform threshold monitoring on the vibration signal to determine whether to issue an interrupt signal to wake up the microcontroller;

[0009] a data storage unit connected to the vibration monitoring unit, configured to respond to the awakening state of the microcontroller, record a timestamp of each awakening moment, and store the timestamp in a data list;

[0010] The analyzing and determining unit is connected to the data storage unit and is used to determine the submission status of the data based on the amount of data or the time interval in the data list, wherein:

[0011] If the data volume is greater than or equal to the capacity of the vibration list and / or the time interval between two vibration events is less than a preset value and the time is discontinuous, feature extraction is performed on the raw vibration data and the feature data is submitted to the neural network model;

[0012] wherein the neural network model determines a vibration trigger characterization coefficient based on the characteristic data to identify the vibration trigger source;

[0013] a vibration verification unit connected to the vibration monitoring unit and the analysis and determination unit, and calculating the vibration comparison similarity to determine the accuracy of the analysis and determination unit's identification;

[0014] The optimization unit is connected to the vibration monitoring unit, the analysis and determination unit, and the vibration verification unit, and is used to continuously collect new vibration samples and integrate the vibration samples into the existing model to improve the accuracy of the model.

[0015] Furthermore, the vibration monitoring unit is used to obtain vibration signals and uses a multi-range amplification factor to monitor the signal of the same vibration sensor to ensure that vibration signals from weak to strong can be accurately captured.

[0016] Furthermore, the vibration monitoring unit determines whether to send an interrupt signal to wake up the microcontroller, wherein:

[0017] If the detected vibration signal exceeds the preset vibration signal threshold, an interrupt signal is triggered to wake up the microcontroller, and the timestamp of the wake-up time is stored in the data table;

[0018] Alternatively, the monitoring can be continued until the microcontroller is woken up periodically by the real-time clock.

[0019] Furthermore, the analysis and determination unit determines the submission status of the data based on the amount of data or time interval in the data list, wherein:

[0020] If the data volume is greater than or equal to the vibration list capacity and / or the time interval between two vibration events is less than a preset value and the time interval is discontinuous, determining that the submission status is submittable;

[0021] Alternatively, the submission status is unsubmittable and is continuously monitored.

[0022] Furthermore, the process of extracting features from the original vibration data by the analysis and determination unit includes:

[0023] Preprocessing of raw vibration data, including filtering, noise reduction and detrending;

[0024] Extract the variance characteristics, peak characteristics and power spectrum density characteristics of the preprocessed vibration data.

[0025] Furthermore, the process of calculating the vibration trigger characterization coefficient by the analysis and determination unit includes:

[0026] Determine the ratio of the variance characteristic to the benchmark variance characteristic as the first vibration influence coefficient;

[0027] Determine the ratio of the peak characteristic to the reference peak characteristic as the second vibration influence coefficient;

[0028] Determine the ratio of the power spectrum density characteristic to the reference power spectrum density characteristic as the third vibration influence coefficient;

[0029] A weighted sum of the first vibration influence coefficient, the second vibration influence coefficient, and the third vibration influence coefficient is determined to be a vibration trigger characterization coefficient.

[0030] Furthermore, the analysis and determination unit is pre-set with vibration trigger characterization coefficient intervals corresponding to different vibration sources.

[0031] Furthermore, the process of the analysis and determination unit identifying the vibration trigger source includes:

[0032] Determining a vibration trigger characterization coefficient interval in which the vibration trigger characterization coefficient lies;

[0033] If the vibration trigger characterization coefficient is greater than or equal to the vibration trigger characterization coefficient threshold, the vibration source corresponding to the vibration trigger characterization coefficient interval is determined to be a vibration trigger source.

[0034] Furthermore, the vibration trigger characterization coefficient threshold is a variable value, wherein,

[0035] The neural network stores the vibration trigger characterization coefficient in real time to update the vibration trigger characterization coefficient threshold.

[0036] Furthermore, the process of the vibration verification unit determining the accuracy of the identification by the analysis and determination unit includes:

[0037] Determine the ratio of the vibration trigger characterization coefficient to the original vibration trigger characterization coefficient as the vibration comparison similarity;

[0038] If the vibration comparison similarity is greater than or equal to the vibration comparison similarity value threshold, it is determined that the analysis and determination unit has made accurate identification.

[0039] Compared with the prior art, the present invention is provided with a vibration monitoring unit, a data storage unit, an analysis and determination unit, a vibration verification unit and an optimization unit. The vibration monitoring unit is used to determine whether to send an interrupt signal to wake up the microcontroller, the data storage unit is used to record the timestamp of each wake-up moment and store the timestamp in a data list, the analysis and determination unit is used to determine the submission status of the data, and perform feature extraction on the original vibration data to calculate the vibration trigger characterization coefficient and identify the vibration trigger source. The vibration verification unit calculates the vibration comparison similarity to determine the accuracy of the analysis and determination unit's identification, and the optimization unit is used to continuously collect new vibration samples and integrate the vibration samples into the existing model to improve the accuracy of the model and the sensitivity of vibration monitoring.

[0040] In particular, the present invention performs real-time threshold monitoring on the monitoring area to determine whether an interrupt signal needs to be sent to wake up the microcontroller. Under the existing scheme, when monitoring vibration signals, single-channel single-range measurement is mostly used. This scheme results in a very narrow range of threshold setting and poor dynamic range. This monitoring scheme is more suitable for detecting large-scale vibration events, such as earthquakes, abnormal operation of large machinery, etc. If the existing scheme is still used when monitoring small regular vibration sources near the ground in real life, the monitoring sensitivity will be reduced and the vibration source cannot be identified in time. Based on this, the present invention considers using multiple ranges of amplification to monitor the signal of the same vibration sensor to ensure that it can accurately capture vibration signals from weak to strong. Once a vibration signal is detected, an interrupt signal is immediately sent to wake up the microcontroller. At the same time, the present invention sets a daily monitoring time. If there is no vibration signal within the daily monitoring time, the real-time clock will immediately wake up the microcontroller for data storage and storage after the daily monitoring time ends, to ensure all-round collection of vibration events, thereby improving monitoring sensitivity and comprehensiveness.

[0041] In particular, the present invention performs feature extraction on the original vibration data to calculate the vibration trigger characterization coefficient. In actual situations, the judgment of the vibration trigger source is mostly based on the original vibration data, but some of the data in the original vibration data is invalid data. Calculation according to this method will lead to a waste of computing power. Based on this, the present invention performs feature extraction on the original vibration data and determines the vibration trigger characterization coefficient based on the feature data, which reduces the calculation frequency and improves the calculation speed.

[0042] In particular, the present invention adopts a neural network model to determine the vibration trigger characterization coefficient based on characteristic data to identify the source of the vibration trigger. In real life, the judgment of the source of the vibration trigger is mostly based on a fixed threshold, but for small vibrations, their vibration frequency and amplitude are relatively small. If the fixed threshold judgment method is still used, it may lead to an increase in the false alarm rate or even missed alarms. Based on this, the present invention introduces an intelligent recognition technology based on a neural network, records each vibration data and combines previous data to calculate the vibration trigger characterization coefficient threshold in real time, reducing the number of false alarms, accurately judging the source of the vibration trigger, and improving the recognition accuracy of the vibration trigger source. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic structural diagram of an intelligent vibration monitoring and adaptive optimization device based on a neural network according to an embodiment of the invention;

[0044] Figure 2 A logic block diagram of determining whether to send an interrupt signal to wake up a microcontroller according to an embodiment of the invention;

[0045] Figure 3 A logic block diagram of a determination unit in an embodiment of the present invention determining a submission status of data based on the amount of data or time interval in a data list;

[0046] Figure 4 This is a schematic diagram of an intelligent vibration monitoring and adaptive optimization device based on a neural network according to an embodiment of the invention. DETAILED DESCRIPTION

[0047] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0048] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0049] It should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the term "connection" should be understood in a broad sense. For example, it can mean a fixed connection, a detachable connection, or an integral connection; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0050] See also Figures 1-4 As shown, Figure 1This is a schematic diagram of the structure of an intelligent vibration monitoring and adaptive optimization device based on a neural network according to an embodiment of the invention. Figure 2 This is a logic block diagram of an embodiment of the invention in response to the recognition result of the pre-recognition unit. Figure 3 This is a logic block diagram for determining whether inertial data should be clustered according to an embodiment of the present invention. Figure 4 This is a logic block diagram of determining whether sub-inertial data is abnormal inertial data according to an embodiment of the invention. The intelligent vibration monitoring and adaptive optimization device based on a neural network of the present invention includes:

[0051] a vibration monitoring unit, configured to obtain a vibration signal and perform threshold monitoring on the vibration signal to determine whether to issue an interrupt signal to wake up the microcontroller;

[0052] a data storage unit connected to the vibration monitoring unit, configured to respond to the awakening state of the microcontroller, record a timestamp of each awakening moment, and store the timestamp in a data list;

[0053] The analyzing and determining unit is connected to the data storage unit and is used to determine the submission status of the data based on the amount of data or the time interval in the data list, wherein:

[0054] If the data volume is greater than or equal to the capacity of the vibration list and / or the time interval between two vibration events is less than a preset value and the time is discontinuous, feature extraction is performed on the raw vibration data and the feature data is submitted to the neural network model;

[0055] wherein the neural network model determines a vibration trigger characterization coefficient based on the characteristic data to identify the vibration trigger source;

[0056] a vibration verification unit connected to the vibration monitoring unit and the analysis and determination unit, and calculating the vibration comparison similarity to determine the accuracy of the analysis and determination unit's identification;

[0057] The optimization unit is connected to the vibration monitoring unit, the analysis and determination unit, and the vibration verification unit, and is used to continuously collect new vibration samples and integrate the vibration samples into the existing model to improve the accuracy of the model.

[0058] Specifically, there is no limitation on the method of obtaining the vibration signal. In this embodiment, a vibration sensor is used. Those skilled in the art can select the range of the vibration sensor according to the actual vibration conditions to be detected, which will not be elaborated here.

[0059] Specifically, the original output signal of the vibration sensor will first be preprocessed by a filtering circuit to remove noise interference. The preprocessing operation includes removing trend items, filtering, signal noise reduction, etc. This is an existing technology and will not be repeated here.

[0060] It is understandable that in the low power mode, the microcontroller is usually in a dormant state until it is awakened by a real-time clock or an external interrupt to perform further processing.

[0061] Specifically, the vibration monitoring unit obtains the vibration signal and uses a multi-range amplification factor to monitor the signal of the same vibration sensor to ensure that it can accurately capture vibration signals from weak to strong.

[0062] Specifically, there is no limitation on the amplification factor of the vibration sensor signal, as long as it can ensure the detection of small vibrations. Those skilled in the art can determine it according to actual conditions, and details will not be given here.

[0063] Specifically, the vibration monitoring unit determines whether to issue an interrupt signal to wake up the microcontroller, wherein:

[0064] If the detected vibration signal exceeds the preset vibration signal threshold, an interrupt signal is triggered to wake up the microcontroller, and the timestamp of the wake-up time is stored in the data table;

[0065] Alternatively, the monitoring can be continued until the microcontroller is woken up periodically by the real-time clock.

[0066] Specifically, the preset vibration signal threshold is pre-set, and a number of vibration signals of the same type are obtained for analysis, and an average value of the number of vibration signals of the same type is obtained. The preset vibration signal threshold is determined within a range of 0.45 times to 0.6 times the average value of the vibration signal.

[0067] Specifically, the predetermined duration of the timing clock is determined according to actual conditions, wherein:

[0068] Set the scheduled time from 9:00 to 6:00 on weekdays to 3 hours;

[0069] Set the scheduled duration from 6:00 am on weekdays to 9:00 am the next day and on weekends to 6 hours.

[0070] It is understandable that the data list has a maximum retention period, and only consecutive vibration events occurring within a short time window will be saved in the same list, wherein the maximum retention period is 24 hours.

[0071] The present invention performs real-time threshold monitoring on the monitoring area to determine whether an interrupt signal needs to be sent to wake up the microcontroller. Under the existing scheme, when monitoring vibration signals, single-channel single-range measurement is mostly used. This scheme results in a very narrow range of threshold setting and poor dynamic range. This monitoring scheme is more suitable for detecting large-scale vibration events, such as earthquakes and abnormal operation of large machinery. If the existing scheme is still used when monitoring small regular vibration sources near the ground in real life, the monitoring sensitivity will be reduced and the vibration source cannot be identified in time. Based on this, the present invention considers using multiple ranges of amplification to monitor the signal of the same vibration sensor to ensure that it can accurately capture vibration signals from weak to strong. Once a vibration signal is detected, an interrupt signal is immediately sent to wake up the microcontroller. At the same time, the present invention sets a daily monitoring time. If there is no vibration signal within the daily monitoring time, the real-time clock will immediately wake up the microcontroller for data storage and storage after the daily monitoring time ends, to ensure all-round collection of vibration events, thereby improving monitoring sensitivity and comprehensiveness.

[0072] Specifically, the analysis and determination unit determines the submission status of the data based on the amount of data or the time interval in the data list, wherein:

[0073] If the data volume is greater than or equal to the vibration list capacity and / or the time interval between two vibration events is less than a preset value and the time interval is discontinuous, determining that the submission status is submittable;

[0074] Alternatively, the submission status is unsubmittable and is continuously monitored.

[0075] Specifically, the preset value is pre-set, and the durations of several vibration signals are obtained in advance and the average value of the durations is calculated. The preset value is set within 0.1 times to 0.2 times the average value of the duration.

[0076] Specifically, the process of feature extraction of the original vibration data by the analysis and determination unit includes:

[0077] Preprocessing of raw vibration data, including filtering, noise reduction and detrending;

[0078] Extract the variance characteristics, peak characteristics and power spectrum density characteristics of the preprocessed vibration data.

[0079] Specifically, the present invention performs feature extraction on the original vibration data to calculate the vibration trigger characterization coefficient. In actual situations, the judgment of the vibration trigger source is mostly based on the original vibration data, but some of the data in the original vibration data is invalid data. Calculation according to this method will lead to a waste of computing power. Based on this, the present invention performs feature extraction on the original vibration data and determines the vibration trigger characterization coefficient based on the feature data, which reduces the calculation frequency and improves the calculation speed.

[0080] Specifically, the process of analyzing and determining the unit to calculate the vibration trigger characterization coefficient includes:

[0081] Determine the ratio of the variance characteristic to the benchmark variance characteristic as the first vibration influence coefficient;

[0082] Determine the ratio of the peak characteristic to the reference peak characteristic as the second vibration influence coefficient;

[0083] Determine the ratio of the power spectrum density characteristic to the reference power spectrum density characteristic as the third vibration influence coefficient;

[0084] A weighted sum of the first vibration influence coefficient, the second vibration influence coefficient, and the third vibration influence coefficient is determined to be a vibration trigger characterization coefficient.

[0085] Specifically, the benchmark variance feature is pre-set, wherein the variance features of several vibration signals are obtained in advance for analysis, the average value of the variance features of the several vibration signals is obtained, and the benchmark variance feature is set within 1.15 times to 1.3 times the average value of the variance feature.

[0086] Specifically, the reference peak feature is pre-set, wherein the peak features of several vibration signals are obtained in advance for analysis, the average value of the peak features of the several vibration signals is obtained, and the reference peak feature is set within 1.1 times to 1.25 times the average value of the peak feature.

[0087] Specifically, the benchmark power spectrum density feature is pre-set, wherein the power spectrum density features of several vibration signals are obtained in advance for analysis, the average value of the power spectrum density features of the several vibration signals is obtained, and the benchmark power spectrum density feature is set within 1.12 times to 1.27 times the average value of the power spectrum density feature.

[0088] Specifically, the weight coefficient of the first vibration influence coefficient is set to 0.32, the weight coefficient of the second vibration influence coefficient is set to 0.35, and the weight coefficient of the third vibration influence coefficient is set to 0.33.

[0089] Specifically, the analysis and determination unit is pre-set with vibration trigger characterization coefficient intervals corresponding to different vibration sources.

[0090] Specifically, the process of analyzing and determining the vibration trigger source includes:

[0091] Determining a vibration trigger characterization coefficient interval in which the vibration trigger characterization coefficient lies;

[0092] If the vibration trigger characterization coefficient is greater than or equal to the vibration trigger characterization coefficient threshold, the vibration source corresponding to the vibration trigger characterization coefficient interval is determined to be a vibration trigger source.

[0093] Specifically, the threshold of the vibration trigger characterization coefficient is selected within the interval [0.89, 1.21].

[0094] Specifically, the vibration trigger characterization coefficient threshold is a variable value, where

[0095] The neural network stores the vibration trigger characterization coefficient in real time to update the vibration trigger characterization coefficient threshold.

[0096] Specifically, the present invention adopts a neural network model to determine the vibration trigger characterization coefficient based on characteristic data to identify the source of the vibration trigger. In real life, the judgment of the source of the vibration trigger is mostly based on a fixed threshold, but for small vibrations, their vibration frequency and amplitude are relatively small. If the fixed threshold judgment method is still used, it may lead to an increase in the false alarm rate or even missed alarms. Based on this, the present invention introduces an intelligent recognition technology based on a neural network, records each vibration data and combines previous data to calculate the vibration trigger characterization coefficient threshold in real time, reducing the number of false alarms, accurately judging the source of the vibration trigger, and improving the recognition accuracy of the vibration trigger source.

[0097] Specifically, the process of the vibration verification unit determining the accuracy of the identification by the analysis determination unit includes determining a ratio of the vibration trigger characterization coefficient to the original vibration trigger characterization coefficient as the vibration comparison similarity;

[0098] If the vibration comparison similarity is greater than or equal to the vibration comparison similarity value threshold, it is determined that the analysis and determination unit has made accurate identification.

[0099] Specifically, the original vibration trigger characterization coefficient is obtained based on the original data of the vibration signal, which is a prior art and will not be described in detail.

[0100] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. An intelligent vibration monitoring and adaptive optimization device based on a neural network, characterized in that: include: a vibration monitoring unit, configured to obtain a vibration signal and perform threshold monitoring on the vibration signal to determine whether to issue an interrupt signal to wake up the microcontroller; a data storage unit connected to the vibration monitoring unit, configured to respond to the awakening state of the microcontroller, record a timestamp of each awakening moment, and store the timestamp in a data list; The analyzing and determining unit is connected to the data storage unit and is used to determine the submission status of the data based on the amount of data or the time interval in the data list, wherein: If the data volume is greater than or equal to the capacity of the vibration list and / or the time interval between two vibration events is less than a preset value and the time is discontinuous, feature extraction is performed on the raw vibration data and the feature data is submitted to the neural network model; wherein the neural network model determines a vibration trigger characterization coefficient based on the characteristic data to identify the vibration trigger source; a vibration verification unit connected to the vibration monitoring unit and the analysis and determination unit, and calculating the vibration comparison similarity to determine the accuracy of the analysis and determination unit's identification; The optimization unit is connected to the vibration monitoring unit, the analysis and determination unit, and the vibration verification unit, and is used to continuously collect new vibration samples and integrate the vibration samples into the existing model to improve the accuracy of the model.

2. The intelligent vibration monitoring and adaptive optimization device based on neural network according to claim 1 is characterized in that: The vibration monitoring unit is used to obtain vibration signals and uses multiple ranges of amplification to monitor the signals of the same vibration sensor to ensure that vibration signals from weak to strong can be accurately captured.

3. The intelligent vibration monitoring and adaptive optimization device based on neural network according to claim 1, characterized in that: The vibration monitoring unit determines whether to send an interrupt signal to wake up the microcontroller, wherein, If the detected vibration signal exceeds the preset vibration signal threshold, an interrupt signal is triggered to wake up the microcontroller, and the timestamp of the wake-up time is stored in the data table; Or, continue monitoring until the microcontroller is woken up by the real-time clock.

4. The intelligent vibration monitoring and adaptive optimization device based on neural network according to claim 1, characterized in that: The analyzing and determining unit determines the submission status of the data based on the amount of data or the time interval in the data list, wherein: If the data volume is greater than or equal to the vibration list capacity and / or the time interval between two vibration events is less than a preset value and the time interval is discontinuous, determining that the submission status is submittable; Alternatively, the submission status is unsubmittable and is continuously monitored.

5. The intelligent vibration monitoring and adaptive optimization device based on neural network according to claim 1 is characterized in that: The process of extracting features from the original vibration data by the analysis and determination unit includes: Preprocessing of raw vibration data, including filtering, noise reduction and detrending; Extract the variance characteristics, peak characteristics and power spectrum density characteristics of the preprocessed vibration data.

6. The intelligent vibration monitoring and adaptive optimization device based on neural network according to claim 1, characterized in that: The process of calculating the vibration trigger characterization coefficient by the analysis and determination unit includes: Determine the ratio of the variance characteristic to the benchmark variance characteristic as the first vibration influence coefficient; Determine the ratio of the peak characteristic to the reference peak characteristic as the second vibration influence coefficient; Determine the ratio of the power spectrum density characteristic to the reference power spectrum density characteristic as the third vibration influence coefficient; A weighted sum of the first vibration influence coefficient, the second vibration influence coefficient, and the third vibration influence coefficient is determined to be a vibration trigger characterization coefficient.

7. The intelligent vibration monitoring and adaptive optimization device based on neural network according to claim 1, characterized in that: The analysis and determination unit is pre-set with vibration trigger characterization coefficient intervals corresponding to different vibration sources.

8. The intelligent vibration monitoring and adaptive optimization device based on neural network according to claim 1, characterized in that: The process of the analysis and determination unit identifying the vibration trigger source includes: Determining a vibration trigger characterization coefficient interval in which the vibration trigger characterization coefficient lies; If the vibration trigger characterization coefficient is greater than or equal to the vibration trigger characterization coefficient threshold, the vibration source corresponding to the vibration trigger characterization coefficient interval is determined to be a vibration trigger source.

9. The neural network-based intelligent vibration monitoring and adaptive optimization device according to claim 8, characterized in that: The vibration trigger characterization coefficient threshold is a variable value, wherein, The neural network stores the vibration trigger characterization coefficient in real time to update the vibration trigger characterization coefficient threshold.

10. The intelligent vibration monitoring and adaptive optimization device based on neural network according to claim 1, characterized in that: The process of the vibration verification unit determining the accuracy of the identification by the analysis and determination unit includes: Determine the ratio of the vibration trigger characterization coefficient to the original vibration trigger characterization coefficient as the vibration comparison similarity; If the vibration comparison similarity is greater than or equal to the vibration comparison similarity value threshold, it is determined that the analysis and determination unit has made accurate identification.

Images