Condition monitoring system and method for induction motor

Abstract

The invention provides a state monitoring system and method for an induction motor, and the system comprises at least one vibration sensor which is disposed on at least one surface of the induction motor, and is used for collecting a vibration signal; a signal conditioning module supporting the Internet of Things and configured to receive the vibration signal and transmit the signal to a computing device; the control platform is configured to analyze the vibration signal based on a fault mode and automatically determine a fault type in the induction motor; and the storage device is used for storing and managing the vibration signal and the analysis result for subsequent analysis. The method is simple to realize, can accurately identify the abnormal root cause based on multiple groups of data, and can automatically and instantly diagnose the fault mode of the induction motor. The method not only can accelerate the recognition of abnormal conditions, but also can simplify the process of determining the specific mode of faults of the induction motor.

Description

Technical Field

[0001] The invention belongs to the field of electronic technology and relates to a state monitoring system and method for an induction motor. Background Art

[0002] Induction motors are an integral part of many industrial and commercial applications, converting electrical energy into mechanical motion efficiently and reliably. These motors operate on the principle of electromagnetic induction and consist of a stator and a rotor. AC power supplied to the stator windings creates a rotating magnetic field that induces current in the rotor, resulting in its rotation and mechanical output. They are widely used in applications such as fans, pumps, compressors, and conveyors, and induction motors are available in a variety of designs and power ratings and are the workhorses of various industries. They have attracted interest from many industries due to their ruggedness, reliability, and relatively low cost compared to other types of motors. To ensure the efficiency and safe operation of induction motors, monitoring systems are often employed.

[0003] Monitoring systems for induction motors are essential to ensure their optimal performance and longevity. These systems are designed to track key parameters and conditions to promote smooth operation and early identification of potential problems. By adopting predictive maintenance practices, these systems can reduce downtime and avoid catastrophic failures. Monitoring can be accomplished through a variety of means, including specialized sensors, advanced data analysis techniques, and specialized software tools. Parameters such as vibration, temperature, current, voltage, and power factor are commonly observed to detect deviations from expected norms, which can indicate impending problems.

[0004] Detecting anomalies in induction motors is a complex task fraught with many challenges. These challenges stem from the complex operating environments in which these motors are used, including dynamic conditions, load fluctuations, and temperature variations. A major obstacle is the lack of labeled data that accurately represents normal and abnormal motion behavior. This scarcity hinders the development of effective anomaly detection algorithms. Furthermore, since anomalies are rare compared to normal instances, the inherent class imbalance in this data can bias machine learning models towards identifying normal behavior while having difficulty identifying anomalies. Feature engineering, a critical step in the process that requires selecting relevant motion behavior attributes, can be daunting and may not cover all operational nuances.

[0005] Additionally, the non-stationary nature of induction motor behavior due to gradual wear, maintenance, and operational changes challenges the performance of traditional anomaly detection methods. Transient effects during startup, shutdown, or rapid load adjustments introduce complications, requiring the distinction between transient deviations and true anomalies. The effects of noise, sensor errors, and sensor measurement drift further confuse the situation, creating the need to distinguish true anomalies from data inaccuracies. Additionally, the complex interaction of multiple simultaneous anomalies or cascading failures complicates the detection process. Model selection that strikes a balance between complexity and effectiveness is another hurdle, as overly complex models risk overfitting, while overly simple models may miss subtle anomalies.

[0006] There are some existing products in the market, namely handheld portable wireless sensors, which can detect anomalies based on the collected vibration and temperature data using ISO standards. However, it only provides a general health status classification based on international standards. Another monitoring system claims to be versatile, scalable and customizable to meet specific application needs based on vibration data. Such systems provide 24 / 7 visibility to evaluate the health and protection of individual chassis and can monitor online conditions. It provides multiple readings related to motor health, such as RMS speed, speed peak and acceleration. However, these data lack interpretability and may be difficult for users to understand. Both systems cannot provide abnormal diagnosis results, such as which failure modes belong to which components. If anomalies are detected, maintenance measures must be taken. However, it is challenging for users to distinguish the failure mode or component that caused the problem based on unfamiliar readings alone. As a result, the entire fault identification process is inefficient and prone to human errors.

[0007] Another example of a monitoring and diagnostic system is disclosed in a Chinese patent document with document number CN103995229B, in which it monitors and diagnoses the health of a motor based on the Mahalanobis distance. The system includes collecting signals such as vibration, current, and speed of the motor, calculating features, selecting feature vectors, and constructing a Mahalanobis space representing the normal working state to determine the health state of the test motor signal. Similar to the above, Jin et al. proposed a health index called Mahalanobis distance (MD) to evaluate the health of cooling fans and induction motors using vibration signals. Both systems use statistical process control methods, which have their own limitations. For example, it works best when the data is normally distributed and does not contain outliers. However, in real life, data may often be biased or contain outliers, resulting in inaccurate predictions and limited applicability of the method. In addition, the Mahalanobis Taguchi system relies heavily on statistical calculations and complex algorithmic models, which may pose challenges to those who do not have the necessary statistical expertise or resources.

[0008] Patent document US8473228B2 of the United States proposes another method for detecting rotor-side anomalies in an induction motor, wherein the method processes and rectifies low-frequency signals of voltage and current obtained from the motor. The present invention also includes a system for detecting anomalies. The method is said to be able to identify equivalents, substitutes, and modifications. However, by obtaining low-frequency signals, the data rate may be small and the communication may be slow, thus relaying insufficient or incorrect anomaly information. This may ultimately lead to inaccurate detection or false alarms. In addition, it should be noted that the method only covers the rotor side and does not cover all components inside the induction motor.

[0009] On the other hand, the international patent document with document number WO2020180887A discloses an integrated machine learning system for detecting anomalies in motors, in which it applies the Internet of Things and machine learning to detect anomalies in data signals obtained by sensors mounted on machine parts. It is also able to adjust the discrimination threshold developed by SVDD using the Mahalanobis kernel and applied to anomaly detection in real wireless sensor network data sets. Although in some cases, inter-sample noise can be handled by subsequent stages of the algorithm, statistical models and anomaly prediction through artificial intelligence are possible. However, while it is a cutting-edge system, the cost of developing such a system may be high, and it may not be suitable for diagnosing anomalies in induction motors.

[0010] Therefore, there is an obvious and urgent need to develop a comprehensive system or method that can accurately pinpoint the potential triggers behind anomalies using different data sets. This requirement stems from the need for a solution that can autonomously identify and classify data while minimizing its dependence on complex statistical calculations and complex algorithmic models that usually require expertise in the field. The essence of this requirement is that the proposed system can achieve rapid and automatic diagnosis of induction motor failure modes. Summary of the invention

[0011] The object of the present invention is to overcome the above-mentioned shortcomings and provide a condition monitoring system and method for an induction motor, which can solve the problem in the prior art that it is not possible to simply realize rapid automatic diagnosis of induction motor fault modes.

[0012] In order to achieve the above object, the technical solution adopted by the present invention is: a state monitoring system for an induction motor, characterized in that it includes:

[0013] A vibration sensor, at least one of which is mounted on at least one surface of the induction motor to collect vibration signals;

[0014] a signal conditioning module, which supports the Internet of Things and is configured to receive the vibration signal and transmit the signal to a computing device;

[0015] a control platform configured to analyze the vibration signal based on the fault mode and automatically determine the type of fault in the induction motor;

[0016] The storage device is used to store and manage vibration signals and analysis results for subsequent analysis.

[0017] Further, the vibration sensor is an accelerometer sensor.

[0018] Furthermore, the signal conditioning module includes: a vibration input module and a compact DAQ chassis.

[0019] Further, the control platform includes an intelligent diagnosis module guided by ISO standard 10816-3 and a set of predetermined parameters to provide intelligent diagnosis.

[0020] Further, the intelligent diagnostic module generates automatic responses to anomalies in the vibration signal and their corresponding failure modes.

[0021] Furthermore, the control platform includes an online user interface that provides automatic abnormal diagnosis information, displays the instantaneous risk status of the vibration velocity RMS value, ISO-compliant risk summary tables, real-time visualization of vibration velocity signals and trends, failure modes and threshold settings.

[0022] Furthermore, the user interface further includes a data storage module and a file merging setting module for subsequent analysis.

[0023] Further, the threshold setting includes customization to merge manual settings of failure modes with data stored in storage devices for historical backup, and merging files for historical trend analysis.

[0024] Furthermore, the subsequent analysis is performed offline.

[0025] Furthermore, the storage device is an external hard disk.

[0026] Another object of the present invention is to provide a method for monitoring the state of an induction motor, comprising the following steps:

[0027] Installing at least one vibration sensor on at least one surface of the induction motor to collect vibration signals in real time;

[0028] transmitting the vibration signal to a computing device via an IoT-enabled signal conditioning module;

[0029] The control platform analyzes the vibration signal according to the fault mode and automatically determines the fault type of the induction motor. If an abnormality exists, an alarm is issued to take corrective measures. Otherwise, the online monitoring module continues to scan for abnormalities and stores and manages the vibration signal and analysis results in a storage device for subsequent offline analysis.

[0030] Furthermore, the step of automatically determining the fault type of the induction motor further includes establishing an intelligent diagnosis module.

[0031] Furthermore, the step of establishing the intelligent diagnosis module further includes:

[0032] Construct standard failure modes and their corresponding fault modes based on ISO standard 10816-3 and a set of predetermined parameters;

[0033] Benchmarking the standard failure modes based on historical data and collected data containing anomalies;

[0034] Based on historical data, state-guided data collection is performed on the standard failure mode;

[0035] The historical data and the collected data are statistically analyzed to obtain the root mean square value to determine the threshold value.

[0036] Further, standard failure modes and their corresponding fault modes are constructed, including a denoising step by filtering interference using a random forest algorithm.

[0037] Further, determining the threshold value includes updating the setting of the threshold value in the control platform, corresponding to at least one failure mode.

[0038] Furthermore, the setting of the threshold in the update control platform includes the following steps:

[0039] Identify at least one critical failure mode;

[0040] Extract the corresponding failure mode from historical data and real-time collected data;

[0041] Perform statistical analysis of failure mode related features; and,

[0042] The filter frequency is determined based on standard fault modes to determine the condition of the induction motor.

[0043] Further, when using the random forest algorithm to filter interference, the original vibration signal is divided into a training set, a validation set, and a test set with training setting model parameters and validation set fine-tuning hyperparameters.

[0044] Furthermore, the subsequent offline analysis further comprises the following steps:

[0045] Analyze historical data containing anomalies with medium and high risk conditions to identify failure modes;

[0046] Analyze trend-related information, including the rate of change of RMS values, first harmonic order, second harmonic order, third harmonic order, and the frequency of manually selected Fast Fourier Transform spectra;

[0047] Generate offline reports of historical event backups.

[0048] Compared with the prior art, the present invention has the following advantages:

[0049] The implementation is simple, and includes: a vibration sensor, at least one of which is mounted on at least one surface of the induction motor for collecting vibration signals; a signal conditioning module, which supports the Internet of Things and is configured to receive the vibration signal and transmit the signal to a computing device; a control platform, which is configured to analyze the vibration signal based on the fault mode and automatically determine the type of fault in the induction motor; a storage device for storing and managing the vibration signal and the analysis results for subsequent analysis. It is capable of accurately identifying the root cause of the anomaly based on multiple sets of data. It is capable of automatically and instantly diagnosing the induction motor failure mode. It is possible to identify and classify data without heavy reliance on statistical calculations and complex algorithmic models and without the need for experts in the field. It can not only speed up the identification of abnormal conditions, but also simplify the process of determining the specific way in which the induction motor fails. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The features of the present invention will be more readily understood and appreciated when the following detailed description is read in conjunction with the accompanying drawings of preferred embodiments of the present invention, in which:

[0051] Figure 1 An overall schematic diagram of an embodiment of the present invention is shown;

[0052] Figure 2 A smart diagnostic module with other interfaces to help analyze vibration signals is shown;

[0053] Figure 3 The user interface in the control platform of the present invention is shown;

[0054] Figure 4 The threshold setting interface of one embodiment of the present invention is shown;

[0055] Figure 5 A guide table of induction motor conditions extracted based on a set of predetermined parameters and ISO standard 10816-3 is shown;

[0056] Figure 6 The overall flow chart of establishing the intelligent diagnosis module is shown;

[0057] Figures 7A-7Dshows vibration graphs and data collected from different components of an induction motor;

[0058] Figure 8 A random forest algorithm for denoising vibration signals is shown;

[0059] Fig. 9 The training and validation process of signal denoising using the (RF) algorithm is shown;

[0060] Fig.10 shows a fault tree for an induction motor, which corresponds failure modes to its failure modes; and,

[0061] Fig.11 The state of an induction motor in one of the embodiments of the present invention is shown, where "X" is a value used to determine the state of the motor. DETAILED DESCRIPTION

[0062] As required, specific embodiments of the present invention are disclosed herein. However, it should be understood that the disclosed embodiments are only examples of the present invention, which can be implemented in a variety of different forms. Therefore, the specific structural and functional details disclosed herein should not be interpreted as limitations, but only as the basis of the claims. It should be understood that the drawings and their detailed description are not used to limit the present invention to the specific forms disclosed herein. On the contrary, the present invention covers all modifications, equivalents and alternatives that fall within the scope defined by the claims. As used throughout this application, the word "may" means optional (i.e., it means possible), rather than mandatory (i.e., it means must). Similarly, the words "include" and "comprise" are meant to include but are not limited to. In addition, unless otherwise mentioned, the word "one" means "at least one" and the word "multiple" means one or more. When using abbreviations or technical terms, these refer to the generally accepted meanings known in the art.

[0063] Figure 1 An overall schematic diagram of an embodiment of the present invention is shown, and the present invention provides a condition monitoring system (10) for an induction motor, the system comprising: at least one vibration sensor (12) mounted on at least one surface of the induction motor, for collecting vibration signals; a signal conditioning module (14) supporting the Internet of Things, configured to receive the vibration signal and transmit the signal to a computing device (16); a control platform (18), configured to automatically determine the type of fault in the induction motor based on the fault mode; and a storage device, for storing and managing the vibration signal and analysis results for subsequent analysis.

[0064] In one embodiment of the present invention, the vibration sensor (12) is an accelerator sensor mounted on at least one surface of the induction motor for real-time acquisition of vibration signals. To demonstrate the effectiveness of the present invention, the present invention has been implemented on flue gas desulfurization (FGD) and train traction motors, wherein at least four accelerators have been placed in the induction motor. Although the present invention can work on any type of induction motor, the number of accelerators required will depend on the size and capacity of the induction motor.

[0065] In one embodiment of the present invention, accelerometers are strategically mounted in the induction motor to collect and analyze real-time vibration signals. The IoT-enabled signal conditioning module (14) further includes a vibration input module and a compact DAQ chassis for transmitting the signal to a modem and a computer device, including but not limited to a desktop or laptop computer. The modem router acts as a network modem for remote system control via the Internet, and in one embodiment, the laptop runs a control platform (18), wherein the control platform (18) is guided by ISO standard 10816-3 and a set of predetermined parameters to provide a smart diagnostic module (20). Finally, an external portable hard drive stores the collected data for subsequent analysis and diagnosis.

[0066] In another embodiment of the present invention, the intelligent diagnostic module (20) generates failure modes (21) that automatically correspond to anomalies in the vibration signal and their corresponding failure modes. Figure 2 An intelligent diagnostic module (20) is shown with other interfaces to aid in the analysis of the vibration signal, in conjunction with other displays, namely a Fast Fourier Transform (FFT) of the vibration velocity with a sampling rate of 4,000 Hz and a continuous Short Time Fourier Transform (STFT) of the vibration velocity. The purpose of this functionality is to automatically identify failure modes based on a set of predetermined parameters and to immediately notify the operator or user of the failure mode.

[0067] In another embodiment of the present invention, the system further includes an online user interface that provides automatic abnormal diagnosis information, including but not limited to instantaneous risk conditions (30) with vibration velocity root mean square (RMS) values, ISO-compliant risk summary tables (32) as references, real-time visualization of vibration velocity signals (36a) and trends (36b), failure modes (21) and threshold settings (22). The user interface further provides information including but not limited to data storage (40) and file merging settings (42) for subsequent analysis, wherein the subsequent analysis can be performed offline for further diagnosis. Figure 3 and Figure 4 This further illustrates the point.

[0068] In one embodiment of the present invention, the threshold setting (22) (see Figure 4) can be customized to merge manual settings of fault modes (44), store data for historical backup in storage devices, and merge files for historical trend analysis. More specifically, threshold setting (22) is one of the interfaces for customizing fault mode thresholds. It includes four components: manual fault mode setting, FFT visualization of auxiliary fault mode setting, data storage (40), and file merging for offline trend analysis. The interface sets fault modes and thresholds through visualization of the intelligent diagnosis module (20), stores data for historical backup, and merges files for historical trend analysis.

[0069] The present invention provides a method for monitoring the condition of an induction motor, comprising the following steps: installing at least one vibration sensor (12) on at least one surface of the induction motor for real-time collection of vibration signals; transmitting the vibration signals to a computing device (16) through a signal conditioning module (14) supporting the Internet of Things; a control platform (18) automatically determining the fault type of the induction motor based on the fault mode; if an abnormality exists, alerting an operator to take corrective measures, or if no abnormality is detected, an online monitoring module continues to scan for abnormalities; and storing and managing the vibration signals and analysis results in a storage device for subsequent analysis.

[0070] As a specific embodiment, the storage device of the present invention is an external hard disk.

[0071] In another embodiment of the present invention, when determining the fault type in the control platform (18), the step further includes establishing an intelligent diagnostic module (20). The step of establishing the intelligent diagnostic module (20) further includes according to ISO standard 10816-3 (see Figure 5 ) and a set of predetermined parameters to prepare a standard failure mode and its corresponding standard fault mode; benchmark the fault mode based on historical data and collected data containing anomalies; extract the fault mode from the historical data and the collected data guided by the standard failure mode and state; and statistically analyze the historical data and the collected data to obtain a root mean square (RMS) value (30) to help determine the threshold of the intelligent diagnosis module (20).

[0072] Reference Figure 6 , data filtering, historical data benchmarking, extraction of characteristic patterns or health indicators guided by a set of predetermined parameters, and statistical analysis based on failure modes. Data filtering removes abnormal network data, while benchmarking uses historical data to guide the design of optimal filters or algorithms. Characteristic pattern extraction identifies features based on a set of predetermined parameters that reflect the health status or failure mode of the induction motor. Statistical analysis determines the distribution, average or maximum value of each health indicator to help determine the rule threshold of the control platform (18).

[0073] Vibration graphs and data collected for different parts of the induction motor (see Fig. 7A -D) is provided to aid in the extraction process. More specifically, Fig. 7A Fault models for "mass imbalance" of induction motors are shown, where (A) shows force imbalance, where 1X RPM dominates the spectrum and the phase difference between the unloading and load side horizontal lines is about 0 degrees; (B) shows coupling imbalance, where 1X RPM dominates the spectrum and there is a phase difference of about 180 degrees between the unloading and load side horizontal lines; (C) illustrates dynamic imbalance, where it is a combination of force and coupling imbalance, 1XRPM dominates the spectrum, and the phase difference between the unloading and load side bearings ranges from 0 to 180 degrees, (D) illustrates cantilever rotor imbalance, where both axial and radial are high at 1X RPM. The axial readings are in phase, while the radial phase readings may be unstable.

[0074] Figure 7B Another embodiment of the invention is depicted, wherein the figures show the results of fault models for rotor friction and journal bearings. Rotor friction typically produces a range of frequencies, typically exciting one or more resonant frequencies and integer fractional speed harmonics (1 / 2, 1 / 3, 1 / 4, 1 / 5...), while journal bearing wear / clearance issues are evidenced by the presence of multiple integer speed harmonics (up to 10 or 20 times) (see Figure A). Oil instability (see Figure B) occurs within 0.4-0.48 times RPM and is considered severe.

[0075] Figure 7C Another embodiment of the invention is depicted in FIG, wherein the figure shows the failure spectrum of a rolling element bearing in relation to its specifications. The failure mode of a rolling element bearing can be determined by the following formula:

[0076]

[0077] Fig.7D Another embodiment of the present invention is depicted, wherein the figure shows a fault model in an AC induction motor. In the figure, (A) describes a stator eccentricity, where a stator problem will occur at 2X line frequency (2F L ) produces high vibration at 2X line frequency; (B) describes an eccentric rotor which creates a rotating variable air gap between the rotor and stator, causing pulsating vibration; (C) describes a rotor problem where damaged rotor bars will produce sidebands around the 2nd, 3rd, 4th and 5th operating speed harmonics, and (D) describes a phase problem due to a damaged connector, which can cause excessive vibration at 2X line frequency with sidebands around 2F L At 1 / 3 line frequency horizontal spacing.

[0078] In one embodiment of the present invention, the pre-trained artificial intelligence module performs a denoising step on the collected data by filtering interference using a random forest (RF) algorithm with predetermined hyperparameter values ​​(see Figure 8 ). More specifically, in order to accurately monitor the health of the induction motor signal, the peaks caused by other electromagnetic interference must be filtered out. In this method, the random forest (RF) algorithm with set hyperparameter values ​​can be used to filter out abnormal data. The raw vibration acceleration signal from the channel is used as input. The dataset is divided into training set, validation set, and test set in a ratio of 8:1:1. The training set determines the model parameters, and the validation set selects the hyperparameters. Fig. 9 The training and validation process of signal denoising using the (RF) algorithm is illustrated.

[0079] Another embodiment of the present invention is that when using the random forest algorithm to filter interference, the original vibration signal is divided into a training set, a validation set and a test set with training set model parameters and validation set fine-tuning hyperparameters.

[0080] In another embodiment of the present invention, the subsequent analysis performed offline further includes the following steps: analyzing historical data containing anomalies with medium and high risk conditions to find out failure modes; analyzing trend-related information, including but not limited to the rate of change of RMS value (30), first harmonic order, second harmonic order, third harmonic order, and frequency of manually selected fast Fourier transform (FTT) spectrum; and generating an offline report for historical event backup.

[0081] In another embodiment of the present invention, the fault tree of the induction motor is summarized as follows: Fig.10 After summarizing the failure modes of the induction motor in the form of a fault tree, it is necessary to understand the corresponding failure mode of each state. Based on the vibration diagnosis chart shown and the motor specifications, the failure mode summary of each state is shown in Table 1. Different failure modes can be added if necessary.

[0082] Table 1

[0083]

[0084] In another embodiment, Fig.11 The "X" in refers to the value used to determine the characteristic mode (or health) of the motor. This value can be defined according to ISO standard 10816-3, which takes into account the basic type and rated power of the induction motor. For example, in an induction fan motor used in flue gas desulfurization (FGD), Fig.11The value of "X" in is equal to 2.3, which is rigid for the foundation type and rated power above 300kW. The corresponding characteristic frequencies of the critical failure modes can be determined, and a rule-based expert reasoning system with initial thresholds can be formulated in Table 2. Where "X" is the value used to determine the state of the motor.

[0085] Table 2

[0086]

[0087] In another embodiment, the threshold setting interface in the control platform (18) is customizable and corresponds to at least one failure mode. The setting of the threshold in the updated control platform (18) includes the following steps: determining at least one failure mode; extracting its corresponding fault mode from historical data and real-time collected data; performing statistical analysis on the fault mode related features; and determining the filter frequency according to the standard fault mode to determine the state of the induction motor. When the future fault mode meets the preset rules, fault identification (diagnosis) can be automatically performed. After consulting with experts, the filter frequency is determined according to a set of predetermined parameters for fault detection and identification of the induction motor.

[0088] In summary, the present invention has a significant breakthrough in induction motors, especially systems such as flue gas desulfurization (FGD) and train traction motors. All embodiments disclosed in the present invention are unique features, which realize automatic and immediate diagnosis of induction motor failure modes by providing a system and method guided by ISO standards and a set of predetermined parameters, so as to distinguish it from existing solutions or patents. This automatic abnormality diagnosis function is more efficient and immediate, and eliminates the possibility of human error.

[0089] In addition, the present invention integrates the expertise of induction motors to ensure that the system's diagnostic response to failure modes is accurate and reliable, and the abnormal diagnosis system is easier to operate, concise and accurate for specific applications, and can cover all components inside the induction motor.

[0090] Additionally, establishing failure mode thresholds is critical to the present invention. It enables users to customize failure mode thresholds based on data analysis results, ensuring that the system is optimized for their specific application needs. By customizing these thresholds, users can achieve optimal performance and minimize failures, which is critical to maintaining system integrity. Additionally, the ability to customize these thresholds gives users greater control and flexibility to adjust the system over time to adapt to changes in their needs.

[0091] The artificial intelligence algorithm designed specifically for induction motor signal denoising greatly improves the accuracy of the anomaly detection function of the present invention, making it a more reliable and effective solution for systems using induction motors.

[0092] Furthermore, vibration analysis is a method used to detect motor faults by revealing differences in the vibration spectrum. This technique can identify bearing defects, imbalance, misalignment, gear failures, and lubrication issues. In addition, it can detect structural problems and weaknesses in the motor. By monitoring the vibration signature, potential problems can be detected early so that corrective measures can be taken before the problem becomes serious.

[0093] The present invention uses only one type of sensor, an accelerometer that measures the appropriate acceleration. Accelerometers can detect mechanical and bearing failures by reporting vibration and shaft time changes, and users can detect machine failures before equipment failures by monitoring accelerometer vibration data. This single data type for diagnostics is more practical and cost-effective than multiple sensors.

[0094] In fact, the present invention collects data external to the motor and provides a non-invasive way to detect anomalies.

[0095] The above explanation of the present invention is not limited to the aforementioned embodiments and drawings, and it is obvious to those skilled in the art that various substitutions, modifications and changes may be made without departing from the scope of the present invention.

Claims

1. A condition monitoring system for an induction motor, characterized in that: include: A vibration sensor, at least one of which is mounted on at least one surface of the induction motor to collect vibration signals; a signal conditioning module, which supports the Internet of Things and is configured to receive the vibration signal and transmit the signal to a computing device; a control platform configured to analyze the vibration signal based on the fault mode and automatically determine the type of fault in the induction motor; The storage device is used to store and manage vibration signals and analysis results for subsequent analysis.

2. The condition monitoring system for an induction motor according to claim 1, characterized in that: The vibration sensor is an accelerometer sensor.

3. The condition monitoring system for an induction motor according to claim 1, characterized in that: The signal conditioning module includes: a vibration input module and a compact DAQ chassis.

4. The condition monitoring system for an induction motor according to claim 1, characterized in that: The control platform includes an intelligent diagnostic module guided by ISO standard 10816-3 and a set of predetermined parameters to provide intelligent diagnostics.

5. The condition monitoring system for an induction motor according to claim 4, characterized in that: The intelligent diagnostic module generates automatic responses to anomalies in the vibration signal and their corresponding failure modes.

6. The condition monitoring system for an induction motor according to claim 1 or 4, characterized in that: The control platform includes an online user interface that provides automatic anomaly diagnostic information, displays instantaneous risk status of vibration velocity RMS values, ISO-compliant risk summary tables, real-time visualization of vibration velocity signals and trends, failure modes and threshold settings.

7. The condition monitoring system for an induction motor according to claim 6, characterized in that: The user interface further includes a data storage module and a file merging setting module for subsequent analysis.

8. The condition monitoring system for an induction motor according to claim 6, characterized in that: The threshold settings include customization to incorporate manual settings for failure modes and storage of data in storage devices for historical backup and merging files for historical trend analysis.

9. The condition monitoring system for an induction motor according to claim 7, characterized in that: The subsequent analysis was performed offline.

10. The condition monitoring system for an induction motor according to claim 8, characterized in that: The storage device is an external hard disk.

11. A method for monitoring the state of an induction motor, comprising the following steps: Installing at least one vibration sensor on at least one surface of the induction motor to collect vibration signals in real time; transmitting the vibration signal to a computing device via an IoT-enabled signal conditioning module; The control platform analyzes the vibration signal according to the fault mode and automatically determines the fault type of the induction motor. If an abnormality exists, an alarm is issued to take corrective measures. Otherwise, the online monitoring module continues to scan for abnormalities and stores and manages the vibration signal and analysis results in a storage device for subsequent offline analysis.

12. The method for monitoring the state of an induction motor according to claim 11, characterized in that: The step of automatically determining the fault type of the induction motor further includes establishing an intelligent diagnosis module.

13. The method for monitoring the state of an induction motor according to claim 12, characterized in that: The steps of establishing the intelligent diagnosis module further include: Construct standard failure modes and their corresponding fault modes based on ISO standard 10816-3 and a set of predetermined parameters; Benchmarking the standard failure modes based on historical data and collected data containing anomalies; Based on historical data, state-guided data collection is performed on the standard failure mode; The historical data and the collected data are statistically analyzed to obtain the root mean square value to determine the threshold value.

14. The method for monitoring the state of an induction motor according to claim 13, characterized in that: The standard failure modes and their corresponding fault modes are constructed, including a denoising step by filtering out interference using a random forest algorithm.

15. The method for monitoring the state of an induction motor according to claim 11, characterized in that: Determining the threshold value includes updating the setting of the threshold value in the control platform, corresponding to at least one failure mode.

16. The method for monitoring the state of an induction motor according to claim 15, characterized in that: The setting of the threshold in the update control platform includes the following steps: Identify at least one critical failure mode; Extract the corresponding failure mode from historical data and real-time collected data; Perform statistical analysis of failure mode related features; and, The filter frequency is determined based on standard fault modes to determine the condition of the induction motor.

17. The method for monitoring the state of an induction motor according to claim 14, characterized in that: When filtering disturbances using the random forest algorithm, the original vibration signal is divided into a training set, a validation set, and a test set with training setting model parameters and a validation set fine-tuning hyperparameters.

18. The method for monitoring the state of an induction motor according to claim 11, characterized in that: The subsequent offline analysis further comprises the following steps: Analyze historical data containing anomalies with medium and high risk conditions to identify failure modes; Analyze trend-related information, including the rate of change of RMS values, first harmonic order, second harmonic order, third harmonic order, and the frequency of manually selected Fast Fourier Transform spectra; Generate offline reports of historical event backups.

Images