Fault monitoring method and early warning system for loading and unloading crane pipe at bottom

Abstract

The invention relates to the field of electrical measurement and testing, in particular to a fault monitoring method and early warning system for a bottom loading and unloading crane pipe, and the method comprises the steps: obtaining and preprocessing the current, voltage and vibration sequence of a servo motor, and obtaining a corresponding IMF component and an energy spectrum; and in combination with historical normal operation data and to-be-monitored data, the importance degree of the IMF component is evaluated, and the time domain anomaly degree is obtained based on the standard deviation of the sliding window variance and the DTW distance. Odd harmonic information is extracted from the energy spectrum, and the time domain anomaly degree is combined to obtain the comprehensive anomaly degree. And obtaining a fault degree through weighted integration, and inputting the fault degree into a control system to judge a fault. According to the method, the energy spectrum difference of the IMF components, the first-order difference sequence difference and the time-frequency domain abnormal degree are comprehensively analyzed, the operation state of the servo motor is comprehensively evaluated, key fault information loss and single time domain feature monitoring limitation caused by elimination of high noise components in a traditional method are avoided, and the accuracy and sensitivity of fault monitoring are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electrical measurement and testing. In particular, it relates to a method for monitoring and early warning system for bottom loading and unloading crane. BACKGROUND

[0002] Bottom loading and unloading crane is a device for loading and unloading oil and gas, liquefied natural gas (LNG), chemicals and other media, widely used in wharf, tank station and railway loading and unloading site. This device drives the crane through servo motor for precise positioning and control of material flow, to ensure the safety and efficiency of the loading and unloading process. The running state of the servo motor is directly related to the accuracy and safety of the loading and unloading, once a fault occurs, it may cause serious consequences.

[0003] The normal operation of the servo motor is essential for the safe and efficient operation of the bottom loading and unloading crane. The failure of the servo motor, such as winding short circuit, bearing damage, encoder failure, etc., may cause material leakage, equipment downtime and even explosion and other serious consequences. Therefore, effective fault monitoring of the servo motor, timely detection and handling of potential faults, is of great significance to ensure the safe operation of the entire loading and unloading system.

[0004] In the existing servo motor fault monitoring technology, signal decomposition method is usually used to eliminate components with high noise content. However, this method may cause key fault information to be eliminated or ignored. In addition, relying on single time domain feature for fault monitoring cannot effectively monitor the slight abnormalities when the fault occurs. This may cause fault monitoring not timely or misjudgment, and thus cause material leakage, equipment downtime and even explosion and other serious consequences. SUMMARY

[0005] In order to solve the problem that the existing servo motor fault monitoring technology often eliminates high noise components through signal decomposition, which may cause key fault information to be lost, and single time domain feature monitoring cannot capture slight abnormalities, which may cause monitoring not timely or misjudgment, the present application provides solutions in the following aspects.

[0006] In a first aspect, a failure monitoring method of a bottom loading and unloading swivel joint includes: obtaining a running feature sequence of a servo motor and preprocessing to obtain corresponding IMF components and energy spectrum; wherein the running feature sequence includes: a current sequence, a voltage sequence, and a vibration sequence; for a plurality of historical normal running processes and a to-be-monitored running process, according to the energy difference in the energy spectrum of each IMF component sequence and the feature sequence and the difference of the first-order difference sequence, the importance of the IMF components of each running feature is evaluated, the time domain abnormality degree is obtained based on the standard deviation of the sliding window variance and the DTW distance; the frequency and energy information of the odd harmonics are extracted from the energy spectrum, the comprehensive abnormality degree is obtained according to the frequency and energy changes between the odd harmonics and the energy changes of the odd harmonics in combination with the time domain abnormality degree, the importance of the IMF components and the comprehensive abnormality degree are weighted and integrated based on the importance of the IMF components, the failure degree of the servo motor is obtained, the failure degree is input to the control system of the servo motor, and it is judged whether there is a failure, and the servo motor failure monitoring is completed.

[0007] By comprehensively analyzing the IMF components and energy spectrum of the current, voltage and vibration and other running feature sequences of the servo motor, the importance of each IMF component and the time domain and frequency domain abnormality degree can be accurately evaluated, and then the failure degree of the servo motor can be accurately determined and timely warning is given. The problem of loss of key fault information caused by direct elimination of high noise components in the traditional technology is effectively avoided, and the limitation of relying on only a single time domain feature for fault monitoring is overcome, the accuracy, sensitivity and reliability of fault monitoring are significantly improved, and a strong guarantee is provided for the safe operation of the servo motor.

[0008] Preferably, the calculation method of the importance of the IMF components of each running feature includes: Taking any running feature as a target feature, the ratio between the average value of all frequency energies in the energy spectrum of the IMF components in the target feature and the average value of all energies in the energy spectrum of the feature sequence of the target feature is calculated to obtain an energy ratio; The data difference and waveform change rate difference of the IMF components and the target feature at each time point are calculated and multiplied and summed to obtain the similarity degree of the corresponding sequence of the target feature, the similarity degree is exponentially mapped using a negative exponential function, and the product of the mapping result and the energy ratio is taken as the importance of each IMF component of the target feature.

[0009] Preferably, the time domain abnormality degree includes: The reciprocal of the frequency with the maximum average energy value in all feature sequence energy spectrum is selected as the length of the sliding window; each IMF component is divided by the sliding window, the variance of the sequence in each window is calculated, the standard deviation based on the sequence variance of all windows in the IMF component is calculated, and the difference between the standard deviation in the historical normal operation process is summed up; the product of the difference sum, the standard deviation and the dynamic time warping distance is taken as the time domain abnormality degree of each IMF component.

[0010] Preferably, the calculation method of the comprehensive abnormality degree comprises: The energy change of the odd harmonic of each IMF component and the distribution difference of the frequency energy between the odd harmonics are calculated, and the distribution difference is smoothed by using an exponential function to obtain a frequency domain abnormality degree; and the sum of the frequency domain abnormality degree and the time domain abnormality degree is taken as the comprehensive abnormality degree.

[0011] Preferably, the calculation method of the fault degree comprises: The product of the importance degree and the comprehensive abnormality degree of each IMF component in each feature sequence is calculated respectively, the product results of all IMF components of all feature sequences are accumulated and summed up, and the sum result is divided by the product of the number of feature sequences and the number of IMF components to obtain the fault degree of the servo motor.

[0012] Preferably, the judging whether there is a fault comprises the steps of: If the fault degree is less than or equal to a preset fault threshold, the servo motor is in a normal operation state, otherwise, if the fault degree is greater than the preset fault threshold, the servo motor is in a fault state, and a fault warning is triggered.

[0013] Preferably, the preprocessing step comprises: The wavelet transform is used on the operation feature sequence to remove high-frequency noise, the denoised data is normalized, the normalized data is segmented into a plurality of time sequence segments, the empirical mode decomposition algorithm is applied to the time sequence segments, the time sequence segments are decomposed into a plurality of IMF components, Fourier transform is performed on each IMF component to convert from time domain to frequency domain, the corresponding frequency spectrum is obtained, and the energy spectrum of each IMF component is calculated to identify the main frequency component and the energy change thereof in the signal.

[0014] In the second aspect, a fault warning system of a bottom loading and unloading crane pipe comprises a processor and a memory, and the memory stores computer program instructions; when the computer program instructions are executed by the processor, the fault monitoring method of the bottom loading and unloading crane pipe is realized.

[0015] The present application has the following effects: 1、The application can more comprehensively evaluate the running state of the servo motor by comprehensively considering the energy spectrum difference of the IMF component, the first-order difference sequence difference, the time domain abnormality degree and the frequency domain abnormality degree. The loss of key fault information caused by directly removing high-noise components in the traditional method is avoided, and the limitation of relying on only a single time domain feature for fault monitoring is also overcome; through the combination of time domain and frequency domain multi-feature analysis, the fault signal can be more accurately identified, the misjudgment and missed detection are reduced, and the accuracy of fault monitoring is significantly improved.

[0016] 2、The application can capture the slight abnormality when the fault occurs by analyzing the energy change of the odd harmonic and the frequency energy distribution difference between the odd harmonics. It is particularly suitable for servo motors of bottom loading and unloading cranes with high start-stop frequency and high noise interference. Through the combination of the standard deviation of the sliding window variance and the DTW distance, periodic waveform distortion in the time domain can be effectively detected, even in the case of high noise interference, the fault signal can be found in time, thereby improving the sensitivity of fault monitoring, ensuring that the fault can be detected in the early stage, and avoiding further deterioration of the fault. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a method flowchart of steps S1-S3 in the fault monitoring method of the bottom loading and unloading crane of the embodiment of the application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments of the application.

[0019] Referring to Figure 1 The fault monitoring method of the bottom loading and unloading crane includes steps S1-S3, and specifically as follows: S1: Obtain the running feature sequence of the servo motor and perform preprocessing to obtain corresponding IMF components and energy spectrum; wherein the running feature sequence includes: current sequence, voltage sequence and vibration sequence.

[0020] It should be noted that in the fault monitoring of the servo motor of the bottom loading and unloading crane, the servo motor may have multiple faults, including but not limited to winding short circuit, bearing damage, encoder failure, etc. When a fault occurs, abnormal characteristics will usually appear in the current, voltage and vibration signals.

[0021] Specifically, the historical normal operation process of multiple servo motors and the data acquisition of a servo motor to be monitored. The current sequence of the servo motor is collected through the current sensor arranged at the input end of the motor power supply; the voltage sequence of the servo motor is collected through the voltage sensor arranged at the output end of the frequency converter; and the vibration sequence of the servo motor is collected through the accelerometer arranged at the bearing shell of the motor.

[0022] The acquired current, voltage and vibration sequence data are subjected to wavelet transform to remove high-frequency noise, the denoised data are subjected to normalization processing, the normalized data are segmented into a plurality of time sequence segments, the time sequence segments are subjected to empirical mode decomposition algorithm to be decomposed into a plurality of IMF (Intrinsic Mode Functions) components, each IMF component is subjected to Fourier transform to convert from time domain to frequency domain to acquire the corresponding energy spectrum, the energy distribution in the frequency domain, i.e. the energy spectrum, can be obtained by Fourier transform of the operation characteristic sequence and the components, so as to identify the main frequency components and the energy change in the signal. These IMF components have the same length as the original sequence and can effectively reflect the different frequency components and characteristic information in the original signal, thereby providing a basis for subsequent fault feature extraction and analysis.

[0023] Specifically, taking the current sequence as an example, a current sequence can obtain a plurality of IMF components after EMD (Empirical Mode Decomposition). The original current sequence is a time sequence representing the current values measured at different times, and each IMF component is also a time sequence, and the length of each IMF component is the same as that of the original current sequence.

[0024] S2: According to the energy difference in the energy spectrum of each IMF component sequence and the characteristic sequence and the difference of the first-order difference sequence, the importance of each operation characteristic IMF component is evaluated, and the time domain abnormality degree is acquired based on the standard deviation of the sliding window variance and the DTW distance.

[0025] In the energy spectrum, an arbitrary one is selected as a target characteristic for subsequent fault monitoring analysis. The energy spectrum of the characteristic sequence and the IMF component of the servo motor under different operating states is analyzed, thereby providing basic data for fault monitoring.

[0026] In servo motor fault monitoring for bottom loading arms, the complex coordinated motion of this equipment often results in significant noise associated with fault signals. Traditional fault monitoring methods typically eliminate high-frequency components containing noise, which may overlook fault information contained within these components, leading to untimely fault monitoring. Conversely, if these components are not eliminated, their noise content may be mistakenly identified as fault signals, resulting in misdiagnosis. Therefore, analyzing the importance of each IMF component is crucial for improving the accuracy and timeliness of fault monitoring. The specific steps are as follows: The steps for determining the importance of each IMF component of the target feature include: First, any operational feature is selected as the target feature. Next, the average energy of all frequencies in the energy spectrum of each IMF component in the target feature is calculated and compared with the average energy of all frequencies in the energy spectrum of the target feature's characteristic sequence to obtain the energy ratio. This ratio reflects the importance of the IMF component relative to the original feature sequence in terms of energy. Subsequently, the data differences and waveform change rate differences between each IMF component and the target feature at various time points are calculated, and the two are multiplied and summed to obtain the similarity degree of the target feature's corresponding sequence. The smaller the similarity degree, the more similar the IMF component is to the target feature in terms of numerical and waveform changes, and the richer the fault information it may contain. A negative exponential function is used to exponentially map the similarity degree, and the mapping result is multiplied by the aforementioned energy ratio to finally obtain the importance of each IMF component of the target feature. By comprehensively considering energy contribution and signal similarity, the importance of each IMF component in fault monitoring can be assessed more accurately.

[0027] Specifically, the importance level satisfies the following relationship: ; In the formula, The first characteristic represents the feature of the target during the detection process. The importance of each IMF component Indicates the first The average energy of all frequencies in the energy spectrum of each IMF component. The average value of all frequency energies in the energy spectrum of the feature sequence representing the target characteristics. Indicates the length of the feature sequence. Indicates the first The first IMF component One element, The first characteristic sequence represents the... One element, Indicates the first The first-order difference sequence of the IMF components. One element, The first difference value of the characteristic sequence is represented by the... One element, Represented by natural numbers An exponential function with base 0.

[0028] In other words, The feature sequence that measures the relationship between IMF components and target features in the th... The numerical difference of each element, The feature sequence that measures the relationship between IMF components and target features in the th... The waveform change rate difference of each element. It is necessary to consider both the amplitude information of the signal and the dynamic change information of the signal, as the amplitude of the signal directly reflects the strength of the signal.

[0029] In fault monitoring, changes in amplitude can indicate changes in system state. For example, a sudden increase or decrease in the amplitude of current or voltage may indicate an anomaly in the system. Amplitude differences can help identify signal components whose amplitudes differ significantly from normal operating conditions, thereby improving the sensitivity of fault detection. Waveform changes (i.e., first-order differences) reflect the dynamic characteristics of a signal. In many practical applications, dynamic changes in a signal are more indicative of the system's operating state than static amplitude changes. For example, frequency changes in a vibration signal may indicate mechanical faults more clearly than amplitude changes. Waveform variation differences can help identify signal components whose dynamic characteristics differ significantly from normal operating conditions, thereby improving the accuracy of fault diagnosis. Analyzing both numerical differences and waveform variation differences allows for a more comprehensive assessment of the importance of signal components, thus improving the sensitivity, accuracy, and robustness of fault monitoring. This comprehensive analysis method is highly effective in practical applications, enabling better extraction of fault characteristics and reducing false positives and false negatives.

[0030] When evaluating the importance of each IMF component, this invention considers not only its correlation with the original signal but also the energy information it contains, thus enabling a more accurate assessment of each component's contribution to fault identification. Specifically, for IMF components that are dissimilar to the original signal but have high energy, this invention does not simply discard them but assigns them appropriate importance, avoiding the possibility of overlooking crucial fault information through direct removal and ensuring timely fault monitoring. Simultaneously, for IMF components similar to the original signal but with lower energy, this invention assigns a smaller weight to reduce the misleading influence of noise on fault monitoring and effectively lower the risk of misjudgment. This comprehensive evaluation strategy significantly improves the accuracy and reliability of fault monitoring.

[0031] When a servo motor is operating normally, the time series corresponding to its current, voltage, and vibration characteristics exhibit significant periodicity. In the time domain, these operational characteristic sequences display periodic waveforms with relatively small fluctuations. However, while noise can cause changes in the sequence's fluctuations, it does not disrupt its inherent periodicity. Conversely, when a fault signal occurs, the periodic waveform in the time domain is distorted due to impacts at specific frequencies and the amplification of harmonics. Therefore, analyzing whether the periodicity of the characteristic sequences is disrupted and whether the waveforms maintain similarity is crucial. The specific steps are as follows: The methods for obtaining the temporal anomaly degree of each IMF component include: First, the length of the sliding window is determined. Specifically, by analyzing the energy spectrum of all characteristic sequences, the frequency with the highest average energy value is selected, and its reciprocal is used to determine the length of the sliding window. This method ensures that the window length matches the main frequency components of the signal, thus effectively capturing the periodic characteristics of the signal. Next, for multiple historical normal operation processes and one monitoring process, based on the above sliding window length, all IMF components are divided into sliding windows. Within each sliding window, the variance of the sequence is calculated. Variance is an important indicator for measuring the degree of signal fluctuation. By calculating the variance of each window, we can understand the changes in the signal within a local time. Then, the standard deviation is calculated based on the sequence variance of all windows within each IMF component. The standard deviation further quantifies the distribution of variance, reflecting the fluctuation characteristics of the signal throughout the time series. The calculated standard deviation is compared with the corresponding standard deviation during historical normal operation processes, and the differences between them are calculated and summed. The sum of differences reflects the degree of deviation between the current IMF component and the historical normal state. Finally, the product of the sum of differences, the standard deviation of the current IMF component, and the DTW (Dynamic Time Warping) distance is defined as the temporal anomaly degree of the IMF component. The DTW distance measures the similarity between the current IMF component and historical normal components over time. By combining the standard deviation and the DTW distance, the temporal anomaly degree of the IMF component can be more comprehensively assessed, thus providing a more accurate basis for fault monitoring.

[0032] Specifically, the degree of temporal anomaly satisfies the following relationship: ; In the formula, Indicates the first The degree of temporal anomaly of each IMF component, Indicates the first The standard deviation of the sliding window variance of each IMF component This indicates the number of normal historical processes. Indicates the first The first historical normal operation process The standard deviation of the sliding window variance of each IMF component Indicates the process to be monitored and the first The first historical normal operation process The average dynamic time bending distance between the IMF components.

[0033] The variance within a sliding window reflects the trend of IMF components within that window. When the variance values ​​of multiple windows are similar, i.e., the standard deviation of the variance is small, this indicates that the IMF components have strong periodicity. This is because a smaller standard deviation means that the degree of signal fluctuation is relatively consistent across different windows, thus reflecting the periodic characteristics of the signal.

[0034] By calculating the difference between the standard deviation of the sliding window variance of the current IMF component and the corresponding standard deviation during historical normal operation, it can be determined whether the periodic waveform of the IMF component has been distorted. The larger the difference, the higher the degree of waveform distortion, and thus the greater the probability of anomalies. The standard deviation of the sliding window variance mainly measures the difference in local periodicity, that is, the change in the periodic characteristics of the signal within a local time range.

[0035] To more comprehensively assess the differences between IMF components, the DTW distance is also introduced. The DTW distance measures the global, overall difference between two IMF components, rather than just local periodic variations. By combining the standard deviation of the sliding window variance with the DTW distance, anomalies in IMF components can be identified and assessed more accurately, thereby improving the accuracy and reliability of fault monitoring.

[0036] S3: Extract the frequency and energy information of odd harmonics from the energy spectrum, combine it with the time domain anomaly degree, obtain the comprehensive anomaly degree based on the frequency and energy changes between odd harmonics and the energy changes of odd harmonics, perform weighted integration based on the importance of IMF components and the comprehensive anomaly degree to obtain the fault degree of the servo motor, input the fault degree into the servo motor control system, and determine whether a fault exists, thus completing the servo motor fault monitoring.

[0037] During the operation of the bottom loading arm, the servo motor starts and stops at a high frequency, resulting in significant noise interference in the operating environment. When a fault first occurs, its anomaly in the time domain is relatively weak and difficult to detect effectively. In the frequency domain, the fault mainly manifests as an increase in odd harmonic energy, exhibiting multiple peaks in the energy spectrum. However, the presence of noise makes these peaks less pronounced, further increasing the difficulty of fault detection. Therefore, a combined frequency domain fault detection approach is adopted, with the specific steps as follows: The methods for obtaining the overall anomaly level of each IMF component include: First, the time-domain anomaly degree of each component is calculated based on the standard deviation of the sliding window variance and the DTW distance, used to measure the anomalousness of signal fluctuations and morphological changes in the time domain. Next, the energy variations of the odd harmonics of each IMF component and the differences in frequency energy distribution among the odd harmonics are analyzed. Specifically, the difference between the energy of each odd harmonic of the current IMF component and the corresponding odd harmonic energy during historical normal operation is calculated, considering the average difference in frequency energy among these odd harmonics. These differences are smoothed using an exponential function to avoid values ​​that are too large or too small, thus obtaining the frequency-domain anomaly degree. Finally, the frequency-domain anomaly degree is added to the time-domain anomaly degree to obtain the comprehensive anomaly degree. This approach considers not only anomalies in the time domain but also characteristic changes in the frequency domain, thus providing a more comprehensive and accurate evaluation index for fault monitoring.

[0038] Specifically, the overall anomaly level satisfies the following relationship: ; In the formula, Indicates the first The overall anomaly degree of each IMF component Indicates the first The degree of temporal anomaly of each IMF component, This indicates the number of normal historical processes. Indicates the number of odd harmonics. Indicates the first step in the monitoring process. The first IMF component The energy value of an odd harmonic Indicates the first The first historical normal operation process The first IMF component The energy value of an odd harmonic Indicates the first step in the monitoring process. The first IMF component The average energy of all frequencies between an odd harmonic and the previous odd harmonic. Indicates the first The first historical normal operation process The first IMF component The average energy of all frequencies between an odd harmonic and the previous odd harmonic. Represented by natural numbers An exponential function with base 0.

[0039] In the fault detection of the servo motor of the bottom loading and unloading crane pipe, the abnormality can be judged by analyzing the odd harmonic energy value of the IMF component. Specifically, if the odd harmonic energy value of the IMF component is significantly higher than the energy value during normal operation, it indicates that the probability of abnormality is relatively large. However, due to the wideband noise generated by the servo motor during operation, such noise will cause the overall energy of the energy spectrum to increase in the wideband. Therefore, the change of the odd harmonic energy value alone may be disturbed by the noise, resulting in misjudgment. In order to avoid the misjudgment of the fault caused by the influence of the noise, the abnormality degree of the odd harmonic needs to be adjusted appropriately, so that the abnormality probability is correspondingly reduced; the exponential function is introduced to avoid the small denominator. Due to the influence of the noise, the small does not mean that the abnormality probability is infinite. Through the adjustment of the exponential function, the abnormality probability can be more reasonably reflected, thereby improving the accuracy and reliability of the fault detection.

[0040] The fault degree is obtained by: First, for each IMF component in each feature sequence, the product of its importance and comprehensive abnormality degree is calculated. The multiplication operation aims to combine the importance of each IMF component with the abnormal situation, so as to more accurately reflect its contribution in fault monitoring. Subsequently, the product results of all IMF components in all feature sequences are accumulated and summed to obtain an overall abnormality degree index. Finally, the sum result is divided by the product of the number of feature sequences and the number of IMF components to realize normalization processing and obtain the final servo motor fault degree. This process not only considers the individual characteristics of each IMF component, but also ensures fair comparison between different feature sequences and IMF components through normalization processing, thereby providing a quantitative and reliable evaluation index for the fault monitoring of the servo motor.

[0041] Not only the energy change of the odd harmonic itself is investigated, but also the frequency energy distribution change between the odd harmonics is analyzed in depth, which can more accurately evaluate the degree of noise influence on the IMF component. Further combining the energy change of the odd harmonic, the frequency domain abnormality is comprehensively evaluated, which effectively distinguishes the energy change caused by noise and the energy change caused by fault, thereby significantly improving the accuracy and reliability of the frequency domain abnormality evaluation.

[0042] By combining the frequency domain abnormality evaluation result with the time domain abnormality evaluation result, dual-mode abnormality detection is realized. This dual-mode detection method can fully utilize the advantages of time domain and frequency domain information, overcome the shortcomings of single-mode detection method, and thus more comprehensively and accurately monitor and diagnose the fault of the servo motor.

[0043] Specifically, the fault degree satisfies the following relationship: ; In the formula, a fault degree of the servo motor, a number of features, a number of EMD decomposition IMF components, a normalized importance degree of the i-th IMF component of the j-th feature, a normalized importance degree of the i-th IMF component of the j-th feature, a normalized importance degree of the i-th IMF component of the j-th feature, a normalized importance degree of the i-th IMF component of the j-th feature.

[0044] In response to the fault degree being less than or equal to a preset fault threshold, the servo motor is in a normal operation state, otherwise, the servo motor is in a fault state, triggering a fault warning.

[0045] The application further provides a fault warning system for a bottom loading and unloading crane. The system comprises a processor, a memory and a warning module, the fault degree is input to the warning module, and it is determined whether a fault exists, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the fault monitoring method for the bottom loading and unloading crane according to the first aspect of the application is realized. The system further comprises a communication bus and a communication interface and other components familiar to those skilled in the art, the settings and functions of which are known in the art, and thus will not be described here.

[0046] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the application, and these all belong to the protection scope of the application. Therefore, the protection scope of the patent of the application should be subject to the appended claims.​​

Claims

1. A fault monitoring method for bottom loading arms, characterized in that, include: The operating characteristic sequence of the servo motor is acquired and preprocessed to obtain the corresponding IMF components and energy spectrum; the operating characteristic sequence includes: current sequence, voltage sequence and vibration sequence; For multiple historical normal operation processes and one operation process to be monitored, the importance of the IMF components of each operation feature is assessed based on the energy differences in the energy spectra of each IMF component sequence and the characteristic sequence, as well as the differences in the first-order difference sequence. The degree of time-domain anomaly is obtained based on the standard deviation of the sliding window variance and the DTW distance. Frequency and energy information of odd harmonics are extracted from the energy spectrum. Combined with the degree of time-domain anomaly, the comprehensive anomaly degree is obtained based on the frequency and energy changes between odd harmonics and the energy changes of odd harmonics. The fault degree of the servo motor is obtained by weighted integration based on the importance of the IMF components and the comprehensive anomaly degree, thus completing the servo motor fault monitoring.

2. The fault monitoring method for the bottom loading arm according to claim 1, characterized in that, The methods for calculating the importance of the IMF components of each operational feature include: Taking any operational feature as the target feature, calculate the ratio between the average value of all frequency energies in the energy spectrum of the IMF component of the target feature and the average value of all energies in the energy spectrum of the feature sequence of the target feature, and obtain the energy ratio. The data differences and waveform change rate differences between the IMF components and the target feature at each time point are calculated and multiplied together to obtain the similarity of the corresponding sequence of the target feature. The similarity is then exponentially mapped using a negative exponential function, and the product of the mapping result and the energy ratio is taken as the importance of each IMF component of the target feature.

3. The fault monitoring method for the bottom loading arm according to claim 1, characterized in that, The degree of temporal anomaly includes: The reciprocal of the frequency with the largest average energy value in the energy spectrum of all characteristic sequences is selected as the sliding window length. The sliding window is divided for each IMF component, the variance of the sequence in each window is calculated, and the standard deviation is calculated based on the variance of the sequence in all windows in the IMF component. The difference between the standard deviation and the difference is summed with the standard deviation during the historical normal operation. The product of the difference summation result, the standard deviation and the dynamic time bending distance is used as the time domain anomaly degree of each IMF component.

4. The fault monitoring method for the bottom loading arm according to claim 1, characterized in that, The calculation method for the overall anomaly level includes: Calculate the energy variation of odd harmonics in each IMF component and the frequency energy distribution differences between odd harmonics, and use an exponential function to smooth the distribution differences to obtain the frequency domain anomaly degree. The sum of the frequency domain anomaly degree and the time domain anomaly degree is taken as the comprehensive anomaly degree.

5. The fault monitoring method for the bottom loading arm according to claim 1, characterized in that, The calculation method for the degree of failure includes: Calculate the product of the importance of each IMF component and the overall anomaly level in each feature sequence. Sum the products of all IMF components in all feature sequences. Divide the sum by the product of the number of feature sequences and the number of IMF components to obtain the fault level of the servo motor.

6. The fault monitoring method for the bottom loading arm according to claim 1, characterized in that, The process of determining whether a fault exists includes the following steps: If the fault level is less than or equal to the preset fault threshold, the servo motor is in normal operation; otherwise, if the fault level is greater than the preset fault threshold, the servo motor is in fault condition, triggering a fault warning.

7. The fault monitoring method for the bottom loading arm according to claim 1, characterized in that, The preprocessing steps include: Wavelet transform is used to remove high-frequency noise from the running feature sequence. The denoised data is then normalized and divided into several time series segments. Empirical Mode Decomposition (IMD) algorithm is applied to the time series segments to decompose them into several IMF components. Fourier transform is performed on each IMF component to convert it from the time domain to the frequency domain and obtain the corresponding spectrum. The energy spectrum of each IMF component is calculated to identify the main frequency components in the signal and their energy changes.

8. A fault early warning system for bottom loading arms, characterized in that, include: The system includes a processor, a memory, and an early warning module. The degree of fault is input to the early warning module, and the system determines whether a fault exists. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the system implements the fault monitoring method for bottom loading arms according to any one of claims 1-7.

Images