Centrifuge fault diagnosis method and system based on data analysis

Abstract

This invention relates to the field of fault diagnosis methods, and particularly to a centrifuge fault diagnosis method and system based on data analysis. The method includes: acquiring vibration signals, rotational speed signals, and feed valve opening signals of the centrifuge; performing time-frequency analysis on the vibration signals to obtain a time-frequency spectrum; extracting energy within the 1X frequency band of the time-frequency spectrum to obtain a vibration energy sequence; calculating a condition fluctuation index, wherein the condition fluctuation index is positively correlated with the absolute value of the rate of change of the rotational speed signal and the absolute value of the rate of change of the feed valve opening signal; calculating a synchronization correlation degree, wherein the synchronization correlation degree is correlated with the vibration energy and the condition fluctuation index; and determining that the centrifuge has a fault when the synchronization correlation degree is greater than a diagnostic threshold. This invention improves the accuracy and reliability of centrifuge imbalance fault detection.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis methods, and in particular to a method and system for diagnosing centrifuge faults based on data analysis. Background Technology

[0002] Centrifuges are key equipment in chemical, pharmaceutical, and bioengineering fields for solid-liquid or liquid-liquid separation. Their core component, the rotor, requires extremely high uniformity of mass distribution during high-speed rotation. Even minute imbalances are amplified dramatically at high speeds, causing severe vibrations that not only affect separation efficiency but can also lead to equipment damage and safety accidents. Therefore, early identification and diagnosis of rotor dynamic imbalances are crucial.

[0003] Currently, wavelet transform (WT), as an effective time-frequency analysis tool, is being applied to the analysis of centrifuge vibration signals. By converting a one-dimensional vibration signal into a two-dimensional time-spectrum graph, it can effectively capture transient and non-stationary features in the signal, thereby identifying abnormal vibrations caused by imbalance. By analyzing the energy distribution pattern at specific frequencies (usually the 1X frequency of the rotational frequency) on the time-spectrum graph, the existence of imbalance can be preliminarily determined. For example, a briefly appearing strong energy region may correspond to a transient imbalance caused by uneven material distribution, while a persistent energy band may point to a structural problem.

[0004] The complete operating cycle of a centrifuge includes multiple stages such as startup and acceleration, stable operation, and deceleration and unloading. During production, it also undergoes operations such as feeding and slag removal. These normal operating condition transitions introduce severe transient impacts into the vibration signal, such as the impact of material on the drum during feeding and the dynamic response during speed changes. The characteristics formed by these normal operations in the wavelet transform time-spectrum diagram are highly similar in morphology to the characteristics of early faults such as uneven material distribution or small component detachment. Because existing fault detection methods cannot distinguish whether transient vibration characteristics originate from normal operating condition transitions or abnormal faults, normal operating condition fluctuations are easily misjudged as faults, generating alarms and affecting the reliability of the diagnostic system. Summary of the Invention

[0005] To address the problem that existing fault detection methods are prone to misjudging fault occurrences, resulting in low accuracy of detection results, this invention provides a centrifuge fault diagnosis method and system based on data analysis.

[0006] In a first aspect, the present invention provides a centrifuge fault diagnosis method based on data analysis, employing the following technical solution:

[0007] A data analysis-based method for centrifuge fault diagnosis includes the following steps:

[0008] Acquire the vibration signal, rotation speed signal, and feed valve opening signal of the centrifuge; perform time-frequency analysis on the vibration signal to obtain the time-frequency spectrum; extract the vibration energy within the 1X frequency band in the time-frequency spectrum.

[0009] The operating condition fluctuation index is calculated and is positively correlated with the absolute value of the rate of change of the speed signal and the absolute value of the rate of change of the feed valve opening signal.

[0010] The synchronization correlation degree is calculated. The synchronization correlation degree is related to the vibration energy and the operating condition fluctuation index. When the operating condition fluctuation index is greater than a set threshold, the synchronization correlation degree is equal to the difference between the vibration energy and the correction term. The correction term is the product of the operating condition fluctuation index and the fluctuation energy conversion coefficient. When the operating condition fluctuation index is less than or equal to the set threshold, the difference between 1 and the operating condition fluctuation index is calculated. The synchronization correlation degree is equal to the product of the difference and the vibration energy. In response to the synchronization correlation degree being greater than the diagnostic threshold, the centrifuge is determined to have a fault.

[0011] By introducing the operating condition fluctuation index and synchronization correlation, the normal operating condition fluctuation is quantified by comprehensively analyzing the changes in speed and feed valve. During diagnosis, the vibration generated by this normal fluctuation is subtracted from or suppressed from the total vibration energy, thereby effectively separating the excessive vibration signal caused by the real imbalance fault. This can accurately distinguish between normal operation interference and real faults, greatly reducing the false alarm rate of the diagnostic system and improving the accuracy and reliability of centrifuge imbalance fault detection.

[0012] The preferred expression for the operating condition fluctuation index is:

[0013] ;

[0014] in, for Operating condition fluctuation index at any given time. for Rate of change of rotational speed at time t. for The rate of change of the opening of the feed valve at any given time. , These are the maximum values ​​of the absolute values ​​of the rate of change of rotational speed and the absolute values ​​of the rate of change of the feed valve opening, respectively, within the monitoring period. for Weighting coefficients for each time step.

[0015] By weighted summing the rate of change of rotational speed and the rate of change of feed valve opening, a clear and quantifiable mathematical model is provided for the fluctuations in operating conditions. This allows for precise calculation of the intensity of fluctuations in normal operating conditions, laying the foundation for accurate isolation of normal vibration interference and enhancing the feasibility and objectivity of the method.

[0016] Preferably, the weighting coefficient is calculated as follows: the real-time power signal of the drive motor is obtained; the material load index is calculated based on the power signal, and the material load index is positively correlated with the power signal; the weighting coefficient is negatively correlated with the material load index.

[0017] By introducing a material load index based on motor power calculation, the weighting coefficients are dynamically adjusted. Compared to using fixed weights, this approach can adaptively determine whether the current operating condition fluctuations are mainly caused by changes in rotational speed (such as no-load start-stop) or feeding operations (such as full-load feeding) based on the actual material load inside the centrifuge, and assign higher weights to the corresponding fluctuation sources. This makes the calculation of the operating condition fluctuation index closer to physical reality, improving the model's adaptability and accuracy under different operating conditions.

[0018] Preferably, the expression for the material load index is:

[0019] ;

[0020] in, This represents the normalized material load index; This indicates the real-time collected motor power; This represents the baseline value of the motor power of the centrifuge during stable no-load operation; This indicates the rated motor power of the centrifuge when it is running stably under full load.

[0021] By utilizing readily available real-time motor power, no-load baseline power, and full-load rated power, a simple and effective method is provided to quantify the material load inside a centrifuge, providing reliable data input for the dynamic adjustment of weighting coefficients.

[0022] The preferred method for calculating the weighting coefficients is as follows:

[0023] ;

[0024] in, for Weighting coefficients at time points, This represents the normalized material load index; This represents an exponential function with base e. This represents the center point parameter.

[0025] Preferably, the method for obtaining the fluctuation energy conversion coefficient includes: collecting historical data of the centrifuge under healthy conditions, including the start-up acceleration, stable operation, and deceleration unloading processes; identifying all windows of pure operating condition fluctuations in the historical data, calculating the ratio of transient vibration energy to operating condition fluctuation index at all times within each window; and using the average of all ratios as the fluctuation energy conversion coefficient.

[0026] By collecting historical data under healthy conditions to calibrate the fluctuation energy conversion coefficient, this solution learns from the equipment's own healthy operating data to establish a quantitative relationship between operating condition fluctuations and normal vibration energy on a specific piece of equipment. This allows for more accurate estimation and elimination of normal vibrations, further improving the accuracy of fault identification.

[0027] Preferably, the time-frequency analysis includes performing continuous wavelet transform on the vibration signal.

[0028] Continuous wavelet transform can better capture and analyze transient and non-stationary features in vibration signals. It is particularly effective in identifying sudden vibrations caused by unbalanced faults or changes in operating conditions, ensuring the accuracy of vibration energy extraction.

[0029] Preferably, the fault diagnosis method further includes: analyzing the detected fault; if the duration of the synchronization correlation degree being higher than the diagnostic threshold is within a set time range and does not recur in subsequent operation, it is determined to be a recoverable material distribution imbalance fault; if the synchronization correlation degree is consistently and stably higher than the diagnostic threshold, it is determined to be a structural imbalance fault and a maintenance warning is triggered.

[0030] By analyzing the duration and reproducibility of fault signals, it is possible to distinguish between temporary, recoverable material distribution imbalances, such as materials briefly clumping and then dispersing, and persistent structural imbalances, such as component wear. This makes the diagnostic results no longer simply a matter of whether there is a fault, but provides more instructive diagnostic conclusions, offering a more accurate basis for subsequent production operation adjustments or maintenance decisions.

[0031] Preferably, the fault diagnosis method further includes filtering the vibration signal, speed signal and feed valve opening signal of the centrifuge.

[0032] Secondly, the present invention provides a centrifuge fault diagnosis system based on data analysis, which adopts the following technical solution:

[0033] A centrifuge fault diagnosis system based on data analysis includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the centrifuge fault diagnosis method based on data analysis described above is implemented.

[0034] The aforementioned data analysis-based centrifuge fault diagnosis method is generated into a computer program and stored in a memory for loading and execution by a processor. Thus, a system is created based on the memory and processor for convenient use.

[0035] The present invention has the following technical effects:

[0036] This invention quantifies the impact of normal operations such as start-up, shutdown, and feeding in real time by constructing an operating condition fluctuation index. By calculating the synchronization correlation degree, it intelligently removes this normal interference from the total vibration signal, solving the problem of frequent false alarms caused by misjudging normal operations as faults in existing technologies. This improves the accuracy and reliability of centrifuge imbalance fault diagnosis and enables accurate identification and early warning of real faults. Attached Figure Description

[0037] Figure 1 This is a flowchart of the centrifuge fault diagnosis method based on data analysis according to the present invention. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] This invention discloses a centrifuge fault diagnosis method based on data analysis, referring to... Figure 1 The process includes the following steps, as detailed below:

[0040] S1: Acquire multi-source heterogeneous data during centrifuge operation.

[0041] Vibration sensors deployed at key locations in the centrifuge (such as bearing housings) acquire raw vibration signal sequences at a preset sampling frequency. Simultaneously, a sequence of operating parameters aligned with the vibration signal timestamps is obtained, primarily including rotational speed and feed valve opening signals. Kalman filters are then used to filter these signals and eliminate noise.

[0042] The acquired raw vibration signal is segmented, and a continuous wavelet transform is applied to each segment to obtain the initial time-frequency spectrum. The initial time-frequency spectrum represents the distribution of vibration energy in the time and frequency dimensions.

[0043] S2: Calculate the operating condition fluctuation index of the centrifuge during operation.

[0044] Rapid changes in rotational speed and the opening or closing of the feed valve are the main non-fault sources causing severe transient impacts in the vibration signal. Therefore, the rate of change of rotational speed, i.e., the rotational acceleration, is first calculated based on the rotational speed signal. Simultaneously, the rate of change of the feed valve opening is calculated based on the feed valve opening signal to capture the feeding action.

[0045] When the centrifuge is in a non-steady-state condition (such as start-up, shutdown, or feeding), the absolute value of the rate of change of its rotational speed or the absolute value of the rate of change of the feed valve state increases. Based on this situation, the operating condition fluctuation index of the centrifuge during operation is calculated, and the expression is as follows:

[0046] ;

[0047] in, for The operating condition fluctuation index of the centrifuge at all times. for Rate of change of rotational speed at time t. for The rate of change of the opening of the feed valve at any given time. , These are the maximum values ​​of the absolute rates of change of rotational speed and the absolute rates of change of feed valve opening over the entire monitoring period, respectively, used for normalization. for The weighting coefficient for each moment, ranging from 0 to 1, is used to adjust the contribution of speed changes and feeding actions to the index.

[0048] When the centrifuge is running stably, the speed is constant and there is no feeding action. and All are close to zero, making the operating condition fluctuation index The value also approaches zero, indicating that the centrifuge's operating state is relatively stable; when the centrifuge performs acceleration, deceleration, or feeding operations, or It will increase instantaneously, further leading to the operating condition fluctuation index. Increase.

[0049] The weighting coefficients are calculated by obtaining the real-time power signal of the drive motor. The material load index is calculated based on the power signal, and the expression is:

[0050] ;

[0051] in, This represents the normalized material load index, with a value between 0 and 1. 0 represents no load, and 1 represents full load. This indicates the real-time collected motor power; This represents the baseline value of the motor power of the centrifuge during stable no-load operation; This indicates the rated motor power of the centrifuge when it is running stably under full load.

[0052] ;

[0053] in, for Weighting coefficients at time points, This represents the normalized material load index; exp represents the exponential function with base e, and c represents the center point parameter, which has a value of 0.5, meaning that when the load reaches 50%, The value is 0.5.

[0054] When the equipment is idle ( ), When the value is relatively large, the operating condition fluctuation index is... It is mainly determined by the change in rotational speed; when the equipment is fully loaded ( ), When the value is small, the operating condition fluctuation index is... It is mainly determined by the feeding action, specifically by the rate of change of the feeding valve opening. During load changes, It will smoothly change between 1 and 0, dynamically adjusting the weights of the two sources of fluctuation.

[0055] S3: Calculate the degree of synchronization between vibration energy and operating condition fluctuation index at each moment.

[0056] From the initial time spectrum Extract the frequency band related to imbalance, i.e., use real-time frequency switching. The 1X frequency band centered on this point has a preset width of 1Hz. The energy integral within this band is calculated to obtain a time-varying vibration energy sequence, where each moment corresponds to a vibration energy level. It should be noted here that frequency conversion... It is calculated from the rotational speed signal, which is existing technology.

[0057] If a strong transient vibrational energy Peak value and a working condition fluctuation index If the peak values ​​are synchronized in time, then the vibration is most likely caused by fluctuations in normal operating conditions; conversely, if it occurs during a stable operating phase, i.e. If a significant vibration energy peak occurs at a low level, it is more likely to be a true fault signal.

[0058] Based on the above characteristics, the synchronization correlation between vibration energy and operating condition fluctuation index at each moment is calculated. The method is as follows: Calculate the fluctuation energy conversion coefficient: Collect historical data containing multiple healthy start-stop and feeding processes, identify all windows of pure operating condition fluctuation, i.e., the speed or feed rate is changing, but the equipment itself is healthy. Based on multiple time windows, calculate the average value of the ratio of vibration energy to operating condition fluctuation index within the 1X frequency band at each moment, and use the average value as the fluctuation energy conversion coefficient. The expression for the synchronization correlation is:

[0059]

[0060] in, for The degree of synchronization correlation at any given moment for Vibrational energy constantly within the 1X frequency band. for The operating condition fluctuation index at time t, where k is the fluctuation energy conversion coefficient. The threshold used to distinguish between stable and fluctuating operating conditions is set according to the actual situation, for example, its value is 0.1.

[0061] When the operating conditions are stable, that is , Approaching 0, at this time Approximately equal to Synchronization correlation can accurately reflect the magnitude of vibration energy; when operating conditions fluctuate drastically, i.e. At this point, based on the severity of the current operating condition fluctuations, the normal vibration energy generated by the operating condition fluctuations in a healthy device is obtained. This reflects the excess vibration energy obtained by subtracting the normal vibration energy caused by fluctuations in operating conditions from the total vibration energy measured in reality. It removes the vibration energy caused by fluctuations in normal operating conditions and retains the excess energy contributed by faults. The synchronization correlation degree can provide a preliminary reflection of the centrifuge's operating status; the higher the value, the greater the probability of the centrifuge malfunctioning.

[0062] S4: Determine whether the centrifuge has malfunctioned based on the synchronization correlation.

[0063] Set diagnostic thresholds At any time ,like If a real, non-operating condition-related abnormal imbalance event is detected at this moment, it is determined that an abnormal imbalance event was detected at this time. Further analysis of the detected abnormal event is performed. If the duration of the event exceeding the threshold is very short (e.g., only a few seconds) and does not recur in subsequent operations, it is determined to be a recoverable material distribution imbalance. Continuously and steadily above the diagnostic threshold If so, it is determined to be a structural imbalance, such as component wear or permanent scaling, and a corresponding maintenance warning is triggered.

[0064] This invention also discloses a centrifuge fault diagnosis system based on data analysis, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the centrifuge fault diagnosis method based on data analysis according to the present invention.

[0065] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0066] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A centrifuge fault diagnosis method based on data analysis, characterized in that, Including the following steps: Acquire the vibration signal, rotation speed signal, and feed valve opening signal of the centrifuge; perform time-frequency analysis on the vibration signal to obtain the time-frequency spectrum; extract the vibration energy within the 1X frequency band in the time-frequency spectrum. The operating condition fluctuation index is calculated and is positively correlated with the absolute value of the rate of change of the speed signal and the absolute value of the rate of change of the feed valve opening signal. The synchronization correlation degree is calculated. The synchronization correlation degree is related to the vibration energy and the operating condition fluctuation index. When the operating condition fluctuation index is greater than the set threshold, the synchronization correlation degree is equal to the difference between the vibration energy and the correction term. The correction term is the product of the operating condition fluctuation index and the fluctuation energy conversion coefficient. When the operating condition fluctuation index is less than or equal to the set threshold, the difference between 1 and the operating condition fluctuation index is calculated, and the synchronization correlation is equal to the product of the difference between 1 and the operating condition fluctuation index and the vibration energy. If the synchronization correlation degree is greater than the diagnostic threshold, the centrifuge is determined to be faulty. The expression for the operating condition fluctuation index is: ； in, for Operating condition fluctuation index at any given time. for Rate of change of rotational speed at time t. for The rate of change of the opening of the feed valve at any given time. , These are the maximum values ​​of the absolute values ​​of the rate of change of rotational speed and the absolute values ​​of the rate of change of the feed valve opening, respectively, within the monitoring period. for Weighting coefficients at different times.

2. The centrifuge fault diagnosis method based on data analysis according to claim 1, characterized in that, The weighting coefficient is calculated as follows: obtain the real-time power signal of the drive motor; calculate the material load index based on the power signal, and the material load index is positively correlated with the power signal; the weighting coefficient is negatively correlated with the material load index.

3. The centrifuge fault diagnosis method based on data analysis according to claim 2, characterized in that, The expression for the material load index is: ； in, This represents the normalized material load index; This indicates the real-time collected motor power; This represents the baseline value of the motor power of the centrifuge during stable no-load operation; This indicates the rated motor power of the centrifuge when it is running stably under full load.

4. The centrifuge fault diagnosis method based on data analysis according to claim 2, characterized in that, The specific method for calculating the weighting coefficients is as follows: ； in, for Weighting coefficients at time points, This represents the normalized material load index; This represents an exponential function with base e. This represents the center point parameter.

5. The centrifuge fault diagnosis method based on data analysis according to claim 1, characterized in that, The method for obtaining the fluctuation energy conversion coefficient includes: collecting historical data of the centrifuge under healthy conditions, including the start-up acceleration, stable operation, and deceleration unloading processes; identifying all windows of pure operating condition fluctuations in the historical data; calculating the ratio of vibration energy to operating condition fluctuation index at all times within each window; and using the average of all ratios as the fluctuation energy conversion coefficient.

6. The centrifuge fault diagnosis method based on data analysis according to claim 1, characterized in that, Time-frequency analysis includes performing continuous wavelet transform on the vibration signal.

7. The centrifuge fault diagnosis method based on data analysis according to claim 1, characterized in that, The fault diagnosis method also includes: analyzing the detected faults; if the duration of the synchronization correlation degree being higher than the diagnostic threshold is within a set time range and does not recur in subsequent operations, it is determined to be a recoverable material distribution imbalance fault; if the synchronization correlation degree is consistently and stably higher than the diagnostic threshold, it is determined to be a structural imbalance fault and a maintenance warning is triggered.

8. The centrifuge fault diagnosis method based on data analysis according to claim 1, characterized in that, The fault diagnosis method also includes filtering the vibration signal, speed signal and feed valve opening signal of the centrifuge.

9. A centrifuge fault diagnosis system based on data analysis, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the centrifuge fault diagnosis method based on data analysis according to any one of claims 1-8.

Images