Grain depot conveying motor load drift identification method and system based on internet of things

Abstract

The application relates to the field of online monitoring of motor states, and particularly discloses a grain depot conveying motor load drift identification method and system based on the Internet of Things, which comprises the following steps: S1, discrete operation data of a grain depot conveying motor are acquired, and effective operation sections are screened; S2, an angle domain alignment characteristic sequence is constructed in the effective operation sections; S3, an observation vector containing operation temperature, three-axis acceleration, vibration main frequency and the angle domain alignment characteristic sequence is constructed, the observation vector is clustered by using a Gaussian mixture model, and a plurality of working condition subspaces are divided; S4, slow-changing characteristic components are extracted in each working condition subspace, cross-modal phase lags and load drift indexes are obtained; and S5, mutation detection is performed on the load drift indexes in the working condition subspaces, and an alarm signal is output based on the mutation detection result. The application can locate the starting point of a gradual abnormality, and realizes timely identification of early drift.

Description

Technical Field

[0001] This application relates to the field of online monitoring of motor status, and specifically discloses a method and system for identifying load drift of grain depot conveyor motors based on the Internet of Things. Background Technology

[0002] Grain depot conveyors are the main equipment for grain storage and handling. The load on their drive motors is significantly affected by factors such as conveying volume, material moisture content, and belt tension, exhibiting time-varying characteristics. To ensure equipment safety, wireless vibration-operating temperature composite sensors are often deployed at grain depot sites. These sensors achieve periodic data collection near the motor housing or bearing housing with low power consumption and report the data to a monitoring platform via a wireless network.

[0003] Currently, there are two main types of monitoring methods for conveyor motors in grain depots: The first type is based on single-point thresholds and rule-based alarms, which typically involves setting upper limits for operating temperature and vibration on the monitoring platform and triggering alarms based on data reported at fixed intervals when the thresholds are exceeded; the second type is based on rotating machinery fault diagnosis methods using continuous vibration waveforms, which typically rely on installing wired accelerometers on the motor bearing housing to obtain continuous high-frequency waveforms and extracting spectral features through order tracking or angular domain resampling, and then comparing them with calibrated thresholds.

[0004] However, the aforementioned existing technologies generally suffer from the following technical problems in practical applications: The first type of method directly sets fixed thresholds for the reported operating temperature and vibration amplitude. This often fails to exclude idle segments of sampling points marked as idle with zero vibration, and does not consider the calculation deviations caused by sparse sampling between adjacent feature points, nor the data interference caused by idle and missing data, resulting in a high false alarm rate for the thresholds. Furthermore, the conveying capacity of grain depot conveyors fluctuates with the operating rhythm; normal load changes and the initial stage of material blockage both show an upward trend in vibration and operating temperature, making early abnormal drift difficult to distinguish.

[0005] In the second category of methods, existing corner domain methods are not suitable for sparse burst vibration data from wireless sensors, and cross-modal feature extraction lacks temporal scale consistency, which can easily lead to false alarms when switching operating conditions. Summary of the Invention

[0006] In view of this, the present invention proposes a method for identifying load drift of grain depot conveyor motors based on the Internet of Things, comprising: S1: Based on the sensing unit, obtain discrete operating data of the grain depot conveyor motor. The discrete operating data includes operating status identifier, operating temperature, triaxial acceleration and vibration main frequency. Based on the operating status identifier, filter to obtain the effective operating segment. S2: Within the effective operating segment, based on the vibration dominant frequency, the cumulative operating phase is calibrated, and the cumulative operating phase, triaxial acceleration, and vibration dominant frequency within the effective operating segment are mapped to a preset equal rotation angle interval grid to construct an angle domain aligned feature sequence; S3: Construct an observation vector containing the operating temperature, triaxial acceleration, vibration dominant frequency and the angular domain aligned feature sequence, and use a Gaussian mixture model to cluster the observation vector to obtain multiple working condition subspaces; S4: Within each of the aforementioned operating condition subspaces, extract the slow-varying characteristic components corresponding to each mode in the discrete operating data, perform cross-modal phase coupling analysis based on the slow-varying characteristic components to obtain the cross-modal phase lag, and normalize the cross-modal phase lag based on the historical statistics of the operating condition subspaces to obtain the load drift index. S5: Within the operating condition subspace, perform abrupt change detection on the load drift index and output an alarm signal based on the abrupt change detection result.

[0007] Furthermore, in step S1, obtaining the triaxial acceleration, vibration dominant frequency, and operating temperature includes: The sampling period corresponding to the preset equal rotation interval grid is used as the length of the sliding window, and the vibration waveform within the sliding window is cached by the sensing unit. For each sliding window, the triaxial acceleration and vibration main frequency are calculated based on the vibration waveform within the sliding window, and the average operating temperature within the sliding window is used as the operating temperature.

[0008] Furthermore, in step S1, the process of filtering and obtaining valid running segments based on the running status identifier includes: The continuous monitoring interval in which the operating status is identified as normal operating status and the continuous running time reaches a first preset threshold is marked as the effective operating segment.

[0009] Furthermore, S2 includes: S21: Within the effective operating segment, the sampling points where the effective value of the triaxial acceleration is greater than zero and the dominant vibration frequency is greater than zero are marked as effective vibration points; S22: Calculate the instantaneous angular velocity based on the dominant vibration frequency of the effective vibration point within the preset frequency band. For sampling point k, its instantaneous angular velocity... The calculation formula is ,in The dominant frequency of vibration is indicated; for effective vibration points that are not within the preset frequency band, linear interpolation is performed using the dominant frequencies of adjacent effective vibration points. S23: Based on the instantaneous angular velocity, phase accumulation is performed to obtain the cumulative operating phase of each sampling point. The formula for calculating the cumulative operating phase is as follows: , in, This represents the cumulative running phase of sampling point k. This indicates the cumulative running phase of the previous sampling point. This represents the time interval between sampling point k and the previous sampling point; S24: Map the accumulated running phase, triaxial acceleration and vibration dominant frequency to a preset equal rotation angle interval grid to construct an angle domain aligned feature sequence.

[0010] Furthermore, the Gaussian mixture model in S3 is:

[0011] in, This represents an observation vector containing the operating temperature, triaxial acceleration, vibration dominant frequency, and the angular domain alignment feature sequence; K is the preset number of operating condition clustering components. Let be the Gaussian mixing coefficient of the j-th component. Let j be the Gaussian probability density function of the j-th component. Let be the mean vector of the j-th component. Let be the covariance matrix of the j-th component; The Gaussian mixture model uses the component with the highest posterior probability corresponding to each sampling point as the working condition subspace to which that sampling point belongs.

[0012] Furthermore, the extraction of slowly varying feature components corresponding to each mode in the discrete running data in S4 includes: S41a: The operating temperature sequence and the corner-domain aligned feature sequence within the effective operating segment are resampled uniformly over time at preset intervals; S41b: Apply a moving average or exponentially weighted moving average to the resampled operating temperature sequence to obtain the slowly varying characteristic components of the operating temperature; apply an exponentially weighted moving average to the resampled corner-aligned sequence to obtain the corner-aligned slowly varying characteristic components.

[0013] Furthermore, in step S4, cross-modal phase coupling analysis is performed based on the slowly varying characteristic components to obtain the cross-modal phase lag. The cross-modal phase lag is then normalized based on historical statistics of the operating condition subspace to obtain the load drift index, including: S42: Construct analytical signals for the slowly varying characteristic components of the operating temperature and the slowly varying characteristic components aligned to the corner domain, respectively, and extract the corresponding instantaneous phase of the operating temperature. Aligning instantaneous phase with angular domain The difference between the instantaneous phase at the operating temperature and the instantaneous phase at angular domain alignment is calculated to obtain the cross-modal phase hysteresis. : ; S43: Based on cross-modal phase hysteresis Calculate the mean cross-modal phase lag in the current operating condition subspace. ; S44: Obtain the mean cross-modal phase lag of the current operating condition subspace under historical normal operating conditions. and standard deviation The load drift index was calculated. : .

[0014] Furthermore, in step S5, performing abrupt change detection on the load drift index and outputting an alarm signal based on the abrupt change detection result includes: The load drift index is assumed to be a piecewise constant mean model: , in, The load drift index is the observed value. For runtime variable, Indicates normal operating status. Indicates an abnormal drift state; Within the operating condition subspace, when the newly observed load drift index causes the running length variable to... When the posterior probability of a change point exceeds the second preset threshold, the time of the change point is recorded and the first alarm signal is output.

[0015] Furthermore, this method also includes step S6: obtaining the average effective value of the three-axis acceleration in the current working condition subspace under historical normal operating conditions. and standard deviation ; Based on the current effective value of triaxial acceleration Mean value of triaxial acceleration and standard deviation The vibration surge rate was calculated as follows: ; When the vibration surge rate is greater than the third preset threshold and the current operating temperature does not exceed the upper limit of the historical operating temperature baseline of the operating condition subspace, a second alarm signal is output.

[0016] This invention also proposes an IoT-based load drift identification system for grain depot conveyor motors, comprising: The sensing unit is used to collect discrete operating data of the grain depot conveyor motor. The discrete operating data includes operating status indicators, operating temperature, triaxial acceleration, and vibration frequency. A computing unit is used to receive the operating data of the sensing unit and execute the above method; The execution unit is used to receive the alarm signal output by the computing unit and report it to the grain depot storage equipment IoT platform.

[0017] This invention filters out continuously valid operating segments based on operating status identifiers, eliminating standby states and invalid zero-value vibration points, thus ensuring the purity of the basic data. Furthermore, by constructing an angle-domain aligned feature sequence, it alleviates the incomparability of features caused by sparse and non-uniform reporting and the signal non-stationarity issues caused by speed fluctuations. By dividing the operating condition subspace using a Gaussian mixture model and combining it with phase coupling of slowly varying feature components extracted within the operating condition subspace, and normalizing them to a load drift exponent, it effectively reduces the false alarm rate of the global threshold during operating condition switching. Simultaneously, the extraction of slowly varying components solves the problem of inconsistent vibration and temperature timescales, enabling the location of the starting point of gradual anomalies and achieving timely identification of early drift. Attached Figure Description

[0018] Figure 1 This is a flowchart of a method for identifying load drift of a grain depot conveyor motor based on the Internet of Things, according to the present invention. Figure 2 This is a diagram illustrating the components of an IoT-based load drift identification system for grain depot conveyor motors according to the present invention. Detailed Implementation

[0019] To address the numerous problems existing in current monitoring methods for grain depot conveyor motors, this invention proposes an Internet of Things-based method for identifying load drift in grain depot conveyor motors.

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been presented in the various embodiments of this application to enable the reader to better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments.

[0021] In the following description, certain specific details are set forth for the purpose of illustrating various disclosed embodiments in order to provide a thorough understanding of the various disclosed embodiments. However, those skilled in the art will recognize that embodiments may be practiced without one or more of these specific details. In other instances, well-known apparatuses, structures, and techniques associated with this application may not have been shown or described in detail to avoid unnecessarily obscuring the description of the embodiments.

[0022] Unless the context requires otherwise, throughout the specification and claims, the word “comprising” and its variations, such as “including” and “having”, shall be understood to have an open, inclusive meaning, that is, to be interpreted as “including, but not limited to”.

[0023] The embodiments of this application will be described in detail below with reference to the accompanying drawings to provide a clearer understanding of the purpose, features, and advantages of this application. It should be understood that the embodiments shown in the drawings are not intended to limit the scope of this application, but are merely for illustrating the essential spirit of the technical solution of this application.

[0024] Throughout this specification, references to "an embodiment" or "an embodiment" indicate that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Therefore, the appearance of "in an embodiment" or "an embodiment" in various places throughout the specification does not necessarily refer to the same embodiment. Furthermore, a particular feature, structure, or characteristic may be combined in any manner in one or more embodiments.

[0025] A method for identifying load drift of grain depot conveyor motors based on the Internet of Things, such as Figure 1 As shown, it includes the following steps: S1: Based on the sensing unit, obtain discrete operating data of the grain depot conveyor motor. The discrete operating data includes operating status identifier, operating temperature, triaxial acceleration and vibration main frequency. Based on the operating status identifier, filter to obtain the effective operating segment. S2: Within the effective operating segment, based on the vibration dominant frequency, the cumulative operating phase is calibrated, and the cumulative operating phase, triaxial acceleration, and vibration dominant frequency within the effective operating segment are mapped to a preset equal rotation angle interval grid to construct an angle domain aligned feature sequence; S3: Construct an observation vector containing the operating temperature, triaxial acceleration, vibration dominant frequency and the angular domain aligned feature sequence, and use a Gaussian mixture model to cluster the observation vector to obtain multiple working condition subspaces; S4: Within each of the aforementioned operating condition subspaces, extract the slow-varying characteristic components corresponding to each mode in the discrete operating data, perform cross-modal phase coupling analysis based on the slow-varying characteristic components to obtain the cross-modal phase lag, and normalize the cross-modal phase lag based on the historical statistics of the operating condition subspaces to obtain the load drift index. S5: Within the operating condition subspace, perform abrupt change detection on the load drift index and output an alarm signal based on the abrupt change detection result.

[0026] This invention filters out continuously valid operating segments based on operating status identifiers, eliminating standby states and invalid zero-value vibration points, thus ensuring the purity of the basic data. Furthermore, by constructing an angle-domain aligned feature sequence, it alleviates the incomparability of features caused by sparse and non-uniform reporting and the signal non-stationarity issues caused by speed fluctuations. By dividing the operating condition subspace using a Gaussian mixture model and combining it with phase coupling of slowly varying feature components extracted within the operating condition subspace, and normalizing them to a load drift exponent, it effectively reduces the false alarm rate of the global threshold during operating condition switching. Simultaneously, the extraction of slowly varying components solves the problem of inconsistent vibration and temperature timescales, enabling the location of the starting point of gradual anomalies and achieving timely identification of early drift.

[0027] Furthermore, in step S1, obtaining the triaxial acceleration, vibration dominant frequency, and operating temperature includes: The sampling period corresponding to the preset equal rotation interval grid is used as the length of the sliding window, and the vibration waveform within the sliding window is cached by the sensing unit. For each sliding window, the triaxial acceleration and vibration main frequency are calculated based on the vibration waveform within the sliding window, and the average operating temperature within the sliding window is used as the operating temperature.

[0028] Furthermore, in step S1, the process of filtering and obtaining valid running segments based on the running status identifier includes: The continuous monitoring interval in which the operating status is identified as normal operating status and the continuous running time reaches a first preset threshold is marked as the effective operating segment.

[0029] Through the above screening process, invalid or interfering data reported in standby, alarm, or dangerous states are effectively eliminated. For example, in standby mode, vibration acceleration may be close to zero due to the equipment being stationary; directly including such data in the analysis would severely interfere with the judgment of the load characteristics of the operating segment. Simultaneously, limiting the minimum continuous duration to a first preset threshold prevents occasional instantaneous state changes from being misjudged as valid operating segments, ensuring the selected data possesses continuity and statistical significance, and providing reliable data support for subsequent operating condition clustering and change point detection.

[0030] Furthermore, S2 includes: S21: Within the effective operating segment, the sampling points where the effective value of the triaxial acceleration is greater than zero and the dominant vibration frequency is greater than zero are marked as effective vibration points; S22: Calculate the instantaneous angular velocity based on the dominant vibration frequency of the effective vibration point within the preset frequency band. For sampling point k, its instantaneous angular velocity... The calculation formula is ,in The dominant frequency of vibration is indicated; for effective vibration points that are not within the preset frequency band, linear interpolation is performed using the dominant frequencies of adjacent effective vibration points. In order to eliminate noise interference, angular velocity estimation is limited to a preset reasonable frequency band. S23: Based on the instantaneous angular velocity, phase accumulation is performed to obtain the cumulative operating phase of each sampling point. The formula for calculating the cumulative operating phase is as follows: , in, This represents the cumulative running phase of sampling point k. This indicates the cumulative running phase of the previous sampling point. This represents the time interval between sampling point k and the previous sampling point; S24: Map the accumulated running phase, triaxial acceleration and vibration dominant frequency to a preset equal rotation angle interval grid to construct an angle domain aligned feature sequence.

[0031] Effective value of triaxial acceleration That is, the square root of the sum of the square components of the accelerations of the three orthogonal axes, which represents the root mean square amplitude of the synthesized triaxial vibration signal vector within this time window.

[0032] In this step, sampling points where the effective value of triaxial acceleration is greater than zero and the dominant vibration frequency is greater than zero are marked as effective vibration points. This is because some reported points may have dominant vibration frequencies as high as hundreds or even thousands of hertz, but the effective value of triaxial acceleration is extremely low or even zero. These high-frequency dominant frequencies do not have corresponding mechanical displacement support and are purely environmental electronic noise or resonance spurious peaks. Therefore, they are eliminated to ensure that subsequent instantaneous angular velocity calculations and phase accumulation have a real physical motion basis.

[0033] Furthermore, the Gaussian mixture model in S3 is:

[0034] in, This represents an observation vector containing the operating temperature, triaxial acceleration, vibration dominant frequency, and the angular domain alignment feature sequence; K is the preset number of operating condition clustering components. Let be the Gaussian mixing coefficient of the j-th component. Let j be the Gaussian probability density function of the j-th component. Let be the mean vector of the j-th component. Let be the covariance matrix of the j-th component; The Gaussian mixture model uses the component with the highest posterior probability corresponding to each sampling point as the working condition subspace to which that sampling point belongs.

[0035] The observation vector constructed by this method not only includes temperature, dominant frequency, and acceleration values, but also an angular domain aligned sequence, ensuring that the feature values ​​can simultaneously reflect the thermal and mechanical vibration states of the equipment. The mathematical expression of the Gaussian mixture model consists of a weighted superposition of multiple single Gaussian probability density functions, and the model parameters are iteratively estimated using the expectation-maximization algorithm. After model convergence, for each time-series sampling point, the posterior probability of its belonging to each Gaussian component is calculated, and the index corresponding to the component with the highest posterior probability is used as the subspace label of the sampling point.

[0036] Furthermore, to address the severe mismatch in sampling timescales between slow temperature variation and rapid vibration variation, the extraction step for the slow-varying feature components employs a two-stage process. Step S41a first performs time-uniform resampling on the temperature sequence and the corner-domain aligned sequence within the effective operating segment, mapping the originally non-fixed interval reporting time points onto a fixed, equally time-interval grid, establishing a unified temporal alignment. Step S41b, based on the resampling, performs filtering and smoothing. A moving average or exponentially weighted moving average is applied to the temperature sequence to filter out isolated jumps in the measured values, extracting the underlying slow-varying temperature component reflecting the thermal inertia trend. Simultaneously, an exponentially weighted moving average is applied to the corner-domain aligned sequence to extract the slow-varying trend of the vibration envelope. Through this processing, although vibration and temperature occur simultaneously, the high-frequency carrier wave of vibration and the low-frequency thermal process of temperature are uniformly normalized to the same time scale, providing a physically consistent signal basis for subsequent phase comparison.

[0037] Furthermore, in step S4, cross-modal phase coupling analysis is performed based on the slowly varying characteristic components to obtain the cross-modal phase lag. The cross-modal phase lag is then normalized based on historical statistics of the operating condition subspace to obtain the load drift index, including: S42: Construct analytical signals for the slowly varying characteristic components of the operating temperature and the slowly varying characteristic components aligned to the corner domain, respectively, and extract the corresponding instantaneous phase of the operating temperature. Aligning instantaneous phase with angular domain The difference between the instantaneous phase at the operating temperature and the instantaneous phase at angular domain alignment is calculated to obtain the cross-modal phase hysteresis. : ; S43: Based on cross-modal phase hysteresis Calculate the mean cross-modal phase lag in the current operating condition subspace. ; S44: Obtain the mean cross-modal phase lag of the current operating condition subspace under historical normal operating conditions. and standard deviation The load drift index was calculated. : .

[0038] Treating the time-aligned slowly varying temperature and vibration components as two correlated signals within the same observation period, we perform Hilbert transforms on each signal to construct analytic signals, thereby obtaining the instantaneous phase of each signal in time. Since both signals have had their high-frequency carrier components filtered out, the physical meaning of their phase difference can directly represent the trend-hysteresis relationship between the thermal and vibration responses.

[0039] The load drift index reflects the statistical significance of the current coupling relationship deviating from the historical normal pattern. The larger the value, the higher the probability of load drift.

[0040] In step S5, performing abrupt change detection on the load drift index and outputting an alarm signal based on the abrupt change detection result includes: The load drift index is assumed to be a piecewise constant mean model: , in, The load drift index is the observed value. For runtime variable, Indicates normal operating status. Indicates an abnormal drift state; Within the operating condition subspace, when the newly observed load drift index causes the running length variable to... When the posterior probability of a change point exceeds the second preset threshold, the time of the change point is recorded and the first alarm signal is output.

[0041] The load drift index is treated as a piecewise constant-mean model, observing the posterior probability distribution of a run-length variable. This run-length variable indicates the duration for which the data has maintained the same distribution before the current time point. When a newly observed load drift index shows a significant statistical difference from historical model parameters, the probability of a change point is calculated. Once this probability exceeds a second preset threshold, the system marks a change point on the timeline. This provides the first warning signal before a gradual early anomaly evolves into a serious failure.

[0042] It also includes step S6: obtaining the average effective value of the three-axis acceleration in the current working condition subspace under historical normal operating conditions. and standard deviation ; Based on the current effective value of triaxial acceleration Mean value of triaxial acceleration and standard deviation The vibration surge rate was calculated as follows: ; When the vibration surge rate is greater than the third preset threshold and the current operating temperature does not exceed the upper limit of the historical operating temperature baseline of the operating condition subspace, a second alarm signal is output.

[0043] The calculation of the surge rate relies on the statistical mean and standard deviation of the historical effective values ​​of vibration acceleration in the subspace of this operating condition. When the vibration surge rate exceeds the third preset threshold, a preliminary judgment is made that a transient change in load has occurred. To distinguish this change from normal operating condition switching, it is determined in parallel whether the current operating temperature exceeds the upper limit of the historical temperature baseline for this operating condition. Only when the vibration surge is abnormal but the temperature has not yet produced a significant overheating response will a second alarm signal be output. This allows the system to capture the special physical scenario of a sudden surge in vibration within a few seconds at the beginning of material blockage, while the temperature remains unchanged, thus compensating for the response hysteresis of the load drift index under thermal inertia scenarios.

[0044] Furthermore, this invention also proposes an IoT-based load drift identification system for grain depot conveyor motors, such as... Figure 2 As shown, it includes: The sensing unit is used to collect discrete operating data of the grain depot conveyor motor. The discrete operating data includes operating status indicators, operating temperature, triaxial acceleration, and vibration frequency. A computing unit is used to receive the operating data of the sensing unit and execute the above method; The execution unit is used to receive the alarm signal output by the computing unit and report it to the grain depot storage equipment IoT platform.

[0045] This invention constructs a complete real-time monitoring and control system for the Internet of Things (IoT) encompassing sensing, communication, computing, and execution through the aforementioned units.

[0046] Taking the monitoring of the motor of a bucket elevator conveyor operating inside a single-story warehouse as an example, the specific implementation process of the present invention is described in detail: The motor is equipped with two wireless vibration and temperature composite sensors, which periodically collect vibration waveforms in a 10-second sliding window and report the temperature, acceleration RMS value and main frequency to the warehouse IoT platform.

[0047] During a certain day's operation, the conveyor switched from standby mode to full-load operation, and the reported operating status indicated normal operation. The computing platform first filtered the continuous running segments as valid operating segments based on this status indicator, excluding invalid points where the acceleration was zero during standby. At this time, the conveyor motor speed varied slightly due to fluctuations in the material loading. When processing the vibration data, the platform used the dominant vibration frequency reported by the sensor as the instantaneous angular velocity, performed phase accumulation calculations, and mapped the vibration characteristics onto a grid with equal rotation angle intervals. This allowed the vibration points, which were originally unevenly distributed over time, to be physically aligned to a fixed rotation angle, reducing the signal smearing effect caused by speed fluctuations.

[0048] Due to the large daily transport volume, the high-temperature and high-pressure data received by the platform led the Gaussian mixture model to determine that the current operating condition had entered the heavy-load subspace. Within this subspace, the system extracted the slowly varying trend components of the temperature sequence and vibration envelope sequence, respectively, and calculated the phase lag between them using Hilbert transform. At this point, the system read the historical normal statistical baseline for this heavy-load condition, normalized the phase lag difference of the current window, and obtained the load drift index. This index remained relatively stable near zero during the initial stage of full-load operation, indicating normal temperature-vibration coupling.

[0049] After approximately ten minutes of continuous operation, the motor load began to slowly increase due to the gradual accumulation of material at the tail end of the conveyor. The mean value of the load drift index gradually shifted upward, and the online change-point detection algorithm promptly identified this shift in the index. The posterior probability of the change-point rapidly increased, and the system determined that load drift had occurred and issued the first alarm signal. Simultaneously, in the initial stage of this process, although the effective value of vibration acceleration had increased to a certain extent, the motor winding temperature had not yet changed significantly due to thermal inertia. The system's vibration surge rate parallel criterion was also triggered when the vibration surge rate exceeded the threshold, avoiding the system's lag due to complete reliance on temperature response. Ultimately, the dual-channel alarm decision was reported to the execution unit, which automatically generated an alarm record and sent it to the maintenance personnel's mobile terminal, gaining valuable time for timely repairs of the early signs of material blockage.

[0050] The solutions provided by the embodiments of this application have been described in detail above. The various embodiments in the specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section. It should be noted that those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of the claims of this application.

[0051] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0052] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

Claims

1. A method for identifying load drift of a grain depot conveyor motor based on the Internet of Things, characterized in that, include: S1: Based on the sensing unit, obtain discrete operating data of the grain depot conveyor motor. The discrete operating data includes operating status identifier, operating temperature, triaxial acceleration and vibration main frequency. Based on the operating status identifier, filter to obtain the effective operating segment. S2: Within the effective operating segment, based on the vibration dominant frequency, the cumulative operating phase is calibrated, and the cumulative operating phase, triaxial acceleration, and vibration dominant frequency within the effective operating segment are mapped to a preset equal rotation angle interval grid to construct an angle domain aligned feature sequence; S3: Construct an observation vector containing the operating temperature, triaxial acceleration, vibration dominant frequency and the angular domain aligned feature sequence, and use a Gaussian mixture model to cluster the observation vector to obtain multiple working condition subspaces; S4: Within each of the aforementioned operating condition subspaces, extract the slowly varying characteristic components corresponding to each mode in the discrete operating data. Perform cross-modal phase coupling analysis based on the slowly varying characteristic components to obtain the cross-modal phase lag. Normalize the cross-modal phase lag based on the historical statistics of the operating condition subspaces to obtain the load drift index, including: S42: Construct analytical signals for the slowly varying characteristic components of the operating temperature and the slowly varying characteristic components aligned to the corner domain, respectively, and extract the corresponding instantaneous phase of the operating temperature. Aligning instantaneous phase with angular domain The difference between the instantaneous phase at the operating temperature and the instantaneous phase at angular domain alignment is calculated to obtain the cross-modal phase hysteresis. : ; S43: Based on cross-modal phase hysteresis Calculate the mean cross-modal phase lag in the current operating condition subspace. ; S44: Obtain the mean cross-modal phase lag of the current operating condition subspace under historical normal operating conditions. and standard deviation The load drift index was calculated. : ； S5: Within the operating condition subspace, perform abrupt change detection on the load drift index and output an alarm signal based on the abrupt change detection result.

2. The method according to claim 1, characterized in that, In step S1, obtaining the triaxial acceleration, dominant vibration frequency, and operating temperature includes: The sampling period corresponding to the preset equal rotation interval grid is used as the length of the sliding window, and the vibration waveform within the sliding window is cached by the sensing unit. For each sliding window, the triaxial acceleration and vibration main frequency are calculated based on the vibration waveform within the sliding window, and the average operating temperature within the sliding window is used as the operating temperature.

3. The method according to claim 1, characterized in that, In step S1, the valid running segments obtained by filtering based on the running status identifier include: The continuous monitoring interval in which the operating status is identified as normal operating status and the continuous running time reaches a first preset threshold is marked as the effective operating segment.

4. The method according to claim 1, characterized in that, S2 includes: S21: Within the effective operating segment, the sampling points where the effective value of the triaxial acceleration is greater than zero and the dominant vibration frequency is greater than zero are marked as effective vibration points; S22: Calculate the instantaneous angular velocity based on the dominant vibration frequency of the effective vibration point within the preset frequency band. For sampling point k, its instantaneous angular velocity... The calculation formula is ,in The dominant frequency of vibration is indicated; for effective vibration points that are not within the preset frequency band, linear interpolation is performed using the dominant frequencies of adjacent effective vibration points. S23: Based on the instantaneous angular velocity, phase accumulation is performed to obtain the cumulative operating phase of each sampling point. The formula for calculating the cumulative operating phase is as follows: ， in, This represents the cumulative running phase of sampling point k. This indicates the cumulative running phase of the previous sampling point. This represents the time interval between sampling point k and the previous sampling point; S24: Map the accumulated running phase, triaxial acceleration and vibration dominant frequency to a preset equal rotation angle interval grid to construct an angle domain aligned feature sequence.

5. The method according to claim 1, characterized in that, The Gaussian mixture model mentioned in S3 is: ， in, This represents an observation vector containing the operating temperature, triaxial acceleration, vibration dominant frequency, and the angular domain alignment feature sequence; K is the preset number of operating condition clustering components. Let be the Gaussian mixing coefficient of the j-th component. Let j be the Gaussian probability density function of the j-th component. Let be the mean vector of the j-th component. Let be the covariance matrix of the j-th component; The Gaussian mixture model uses the component with the highest posterior probability corresponding to each sampling point as the working condition subspace to which that sampling point belongs.

6. The method according to claim 1, characterized in that, The extraction of slowly varying feature components corresponding to each mode in the discrete running data in step S4 includes: S41a: The operating temperature sequence and the corner-domain aligned feature sequence within the effective operating segment are resampled uniformly over time at preset intervals; S41b: Apply a moving average or exponentially weighted moving average to the resampled operating temperature sequence to obtain the slowly varying characteristic components of the operating temperature; apply an exponentially weighted moving average to the resampled corner-aligned sequence to obtain the corner-aligned slowly varying characteristic components.

7. The method according to claim 1, characterized in that, In step S5, performing abrupt change detection on the load drift index and outputting an alarm signal based on the abrupt change detection result includes: The load drift index is assumed to be a piecewise constant mean model: ， in, The load drift index is the observed value. For runtime variable, Indicates normal operating status. Indicates an abnormal drift state; Within the operating condition subspace, when the newly observed load drift index causes the running length variable to... When the posterior probability of a change point exceeds the second preset threshold, the time of the change point is recorded and the first alarm signal is output.

8. The method according to claim 7, characterized in that, It also includes step S6: obtaining the average effective value of the three-axis acceleration in the current working condition subspace under historical normal operating conditions. and standard deviation ; Based on the current effective value of triaxial acceleration Mean value of triaxial acceleration and standard deviation The vibration surge rate was calculated as follows: ； When the vibration surge rate is greater than the third preset threshold and the current operating temperature does not exceed the upper limit of the historical operating temperature baseline of the operating condition subspace, a second alarm signal is output.

9. A load drift identification system for grain depot conveyor motors based on the Internet of Things, characterized in that, include: The sensing unit is used to collect discrete operating data of the grain depot conveyor motor. The discrete operating data includes operating status indicators, operating temperature, triaxial acceleration, and vibration frequency. A computing unit is configured to receive operational data from the sensing unit and execute the method as described in any one of claims 1 to 8; The execution unit is used to receive the alarm signal output by the computing unit and report it to the grain depot storage equipment IoT platform.

Images