A method for predicting the life of industrial rotating equipment by fusing multi-dimensional vibration signal analysis

Abstract

The present application relates to the technical field of equipment health management, and more particularly to an industrial rotating equipment life prediction method fusing multi-dimensional vibration signal analysis, which uses the vibration signals and operating condition parameters of the industrial rotating equipment in the historical health operation stage to fit a nonlinear health surface equation of the speed and load to the vibration response, uses the nonlinear health surface equation to analyze the vibration response deviation at the current sampling time, and combines with the sensitivity compensation analysis to obtain an unexplainable vibration degradation coefficient at the current sampling time; uses the unexplainable vibration degradation coefficient at each sampling time in the current analysis time window to cooperatively regulate and control the unexplainable vibration degradation coefficient at the current sampling time, and obtains a trend-enhanced degradation coefficient to predict the remaining life of the industrial rotating equipment, effectively separates the coupling effect of the internal degradation and external operating condition change of the equipment, and makes the remaining life prediction more in line with the actual degradation rate of the equipment.

Description

Technical Field

[0001] This invention relates to the field of equipment health management technology, and in particular to a method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis. Background Technology

[0002] Industrial rotating equipment is widely used in energy, manufacturing, and chemical industries, and its operating status directly affects the stability and safety of industrial production processes. With the development of equipment towards larger scale, higher speed, and continuous operation, industrial rotating equipment typically needs to operate for extended periods in complex environments, such as speed regulation, load changes, start-stop switching, and temperature fluctuations. Therefore, current methods typically involve collecting state parameters such as vibration, temperature, and current signals during equipment operation, and utilizing time-domain analysis, frequency-domain analysis, feature extraction, and data-driven models to establish a mapping relationship between equipment health status and remaining life, thereby achieving the purpose of equipment fault early warning and remaining life assessment.

[0003] Existing methods for equipment fault early warning and remaining life assessment can effectively identify equipment performance degradation processes and improve equipment maintenance efficiency under relatively stable operating conditions and continuous degradation processes, based on the trend of vibration signal changes. However, during the operation of large rotating equipment with rapid switching between multiple operating conditions, the vibration response is simultaneously affected by both internal degradation factors and external operating condition changes. This makes it impossible for vibration amplitude changes to accurately represent the true degradation state of the equipment. For example, when the equipment load increases or the speed increases, the equipment vibration amplitude may naturally increase. Existing methods may easily mistake this change for increased equipment damage, causing health indicators to deviate from the actual degradation process and reducing the accuracy of life prediction. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis, in order to solve the problems of stability and accuracy in predicting the lifespan of industrial rotating equipment under rapid switching of multiple operating conditions.

[0005] This invention provides a method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis. The method includes the following steps:

[0006] The vibration signals and operating parameters of the industrial rotating equipment are collected synchronously during operation, and the vibration value, rotational speed and load at each sampling moment are obtained accordingly.

[0007] Using vibration signals and operating parameters from historical healthy operating phases, a nonlinear health surface equation for the vibration response of rotational speed and load is fitted to characterize the baseline of the healthy operating state of industrial rotating equipment. The vibration response deviation analysis is performed on the vibration value at the current sampling time using the nonlinear health surface equation to obtain the relative vibration deviation ratio. Sensitivity compensation analysis is performed on the relative vibration deviation ratio based on the rotational speed and load at the current sampling time to obtain the unexplainable vibration degradation coefficient at the current sampling time.

[0008] A current analysis time window of a preset length is constructed with the current sampling time as the cutoff time. The unexplainable vibration degradation coefficient at each sampling time within the current analysis time window is used to coordinately regulate the unexplainable vibration degradation coefficient at the current sampling time with consistent long and short time scales, thereby obtaining the trend enhancement degradation coefficient at the current sampling time.

[0009] The remaining life of industrial rotating equipment is predicted by using the trend enhancement degradation coefficient at the current sampling time.

[0010] Preferably, the step of fitting the nonlinear health surface equation of rotational speed and load response to vibration using vibration signals and operating condition parameters from historical healthy operating phases includes:

[0011] For any historical sampling moment during the historical healthy operation phase, a time window of a preset length is constructed with the sampling moment as the cutoff point. Based on the vibration value at each sampling moment within the time window, the RMS value of the vibration value is calculated and recorded as the effective vibration value at that historical sampling moment. The effective vibration value, rotational speed, and load at that historical sampling moment are combined into a sample to obtain a sample set for the historical healthy operation phase. Nonlinear fitting is performed on the sample set to obtain the nonlinear health surface equations of the rotational speed and load response to vibration.

[0012] ;

[0013] in, This represents the effective vibration value at the i-th historical sampling time. Represents the reference response coefficient. This represents the rotational speed at the i-th historical sampling moment. The load at the i-th historical sampling time, Indicates the power exponent of the rotational speed response. This represents the power exponent of the load response.

[0014] Preferably, the step of using the nonlinear health surface equation to perform vibration response deviation analysis on the vibration value at the current sampling time to obtain the relative vibration deviation ratio includes:

[0015] Substituting the rotational speed and load at the current sampling moment into the nonlinear health surface equation yields the theoretical effective vibration value at the current sampling moment, which characterizes the vibration value of the industrial rotating equipment when it is in a healthy operating state at the current sampling moment. A time window of a preset length is constructed with the current sampling moment as the cutoff moment. Based on the vibration value at each sampling moment within the time window, the RMS value of the vibration value is calculated and recorded as the effective vibration value at the current sampling moment. The difference between the effective vibration value at the current sampling moment and the theoretical effective vibration value is calculated, and the maximum value between the difference and the constant 0 is taken as the unidirectional nonlinear deviation. The relative deviation ratio of the vibration at the current sampling moment is obtained by using the sum of the theoretical effective vibration value at the current sampling moment and the smallest real number as the denominator and the unidirectional nonlinear deviation as the numerator.

[0016] Preferably, the step of performing sensitivity compensation analysis on the relative deviation ratio of the vibration based on the rotational speed and load at the current sampling time to obtain the unexplainable vibration degradation coefficient at the current sampling time includes:

[0017] Obtain the product of the rotational speed and load at the current sampling moment. Use the sum of the product and a minimum real number as the denominator, and the product of the rated rotational speed and rated load of the industrial rotating equipment as the numerator to obtain the corresponding ratio. Use the ratio as the base and a preset sensitivity coefficient as the exponent to obtain the adaptive compensation factor for the operating condition sensitivity at the current sampling moment. Use the product of the adaptive compensation factor for the operating condition sensitivity and the relative deviation ratio of the vibration as the unexplainable vibration degradation coefficient at the current sampling moment.

[0018] Preferably, the step of using the unexplained vibration degradation coefficient at each sampling moment within the current analysis time window to perform consistent coordinated control of the unexplained vibration degradation coefficient at the current sampling moment across both long and short time scales, to obtain the trend-enhancing degradation coefficient at the current sampling moment, includes:

[0019] Based on the unexplained vibration degradation coefficient at each sampling moment within the current analysis time window, calculate the mean and standard deviation of the unexplained vibration degradation coefficient, which are denoted as the mean of the local degradation coefficient and the standard deviation of the local degradation coefficient, respectively.

[0020] Based on the unexplained vibration degradation coefficient at each historical sampling moment during the historical healthy operation phase, the mean of the unexplained vibration degradation coefficient is calculated and recorded as the health baseline value; based on the ratio between the health baseline value and the mean of the local degradation coefficient, the degradation magnitude accumulation factor at the current sampling moment is obtained.

[0021] Based on the mean and standard deviation of the local degradation coefficients, a fluctuation dispersion analysis is performed on the unexplained vibration degradation coefficients within the current analysis time window to obtain the linear fluctuation dispersion penalty factor at the current sampling time; based on the difference in unexplained vibration degradation coefficients between adjacent sampling times within the current analysis time window, the monotonic degradation direction consistency index at the current sampling time is obtained.

[0022] By using the cumulative factor of degradation magnitude, the penalty factor of linear fluctuation dispersion and the consistency index of monotonic degradation direction at the current sampling time, the unexplainable vibration degradation coefficient at the current sampling time is synergistically controlled to obtain the trend-enhancing degradation coefficient at the current sampling time.

[0023] Preferably, the step of performing a fluctuation dispersion analysis on the unexplainable vibration degradation coefficient within the current analysis time window based on the mean and standard deviation of the local degradation coefficient to obtain the linear fluctuation dispersion penalty factor at the current sampling time includes:

[0024] The coefficient of variation of the unexplained vibration degradation coefficient within the current analysis time window is calculated based on the mean and standard deviation of the local degradation coefficient. The negative of the coefficient of variation is used as the independent variable of an exponential function with the natural constant as the base to obtain the linear fluctuation dispersion penalty factor at the current sampling time.

[0025] Preferably, the step of obtaining the monotonic degradation direction consistency index at the current sampling moment based on the difference in unexplained vibration degradation coefficients between adjacent sampling moments within the current analysis time window includes:

[0026] ;

[0027] in, This represents the consistency index of the monotonic degradation direction at the current sampling moment. This represents the function that takes the maximum value. This represents the unexplained vibration degradation coefficient at the (i+1)th sampling time within the current analysis time window. This represents the uninterpretable vibration degradation coefficient at the i-th sampling time within the current analysis time window, where || denotes the absolute value sign, and 0 represents a constant. Represents a very small real number. This indicates the number of sampling moments within the current analysis time window.

[0028] Preferably, the step of using the cumulative factor of degradation magnitude, the linear fluctuation dispersion penalty factor, and the monotonic degradation direction consistency index at the current sampling time to coordinately regulate the unexplainable vibration degradation coefficient at the current sampling time to obtain the trend-enhancing degradation coefficient at the current sampling time includes:

[0029] The product of the cumulative factor of degradation magnitude, the linear fluctuation dispersion penalty factor, and the monotonic degradation direction consistency index at the current sampling time is obtained as the control coefficient. The product of the control coefficient and the unexplained vibration degradation coefficient at the current sampling time is obtained as the trend enhancement degradation coefficient at the current sampling time.

[0030] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0031] In this invention, a nonlinear health surface equation for the vibration response of rotational speed and load to vibration is constructed using vibration signals and operating condition parameters from historical healthy operation phases. This equation is then used to perform vibration response deviation analysis on real-time acquired vibration signals. Combined with adaptive sensitivity compensation for operating conditions, an unexplainable vibration degradation coefficient is obtained. This effectively separates the coupling effect between internal equipment degradation and external operating condition changes, solving the problem that traditional methods easily misjudge vibration caused by increased load as increased damage under multiple operating condition switching. A statistical stability assessment mechanism based on a local historical window (a pre-defined analysis time window with the current sampling time as the cutoff time) is introduced. By utilizing mean deviation, fluctuation dispersion penalty, and monotonic direction consistency index, transient shocks and noise are strongly suppressed, while enhancing the credibility of steady and continuous degradation. This significantly improves the time consistency and anti-interference ability of degradation indicators, resulting in a trend-enhancing degradation coefficient at the current sampling time. Finally, the life prediction of industrial rotating equipment is performed based on the trend-enhancing degradation coefficient, making the remaining life prediction more consistent with the current actual degradation rate of the equipment. This improves the accuracy of life prediction for industrial rotating equipment under rapid switching of multiple operating conditions and the reliability of maintenance decisions. Detailed Implementation

[0032] The embodiments of this disclosure are described in detail below. The embodiments described below are exemplary and intended to explain this disclosure, and should not be construed as limiting this disclosure.

[0033] It should be noted that the terms "first," "second," etc., used in this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those described herein. The implementations described in the following exemplary embodiments do not represent all implementations consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure.

[0034] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0035] Specifically, this invention provides a method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis. This method includes the following specific steps:

[0036] Step S101: Simultaneously collect vibration signals and operating parameters of the industrial rotating equipment during operation, and obtain the vibration value, rotational speed and load at each sampling moment.

[0037] The vibration response of industrial rotating equipment during operation is not only affected by internal damage factors, but also by external factors such as changes in rotational speed, load, and operating environment. Therefore, in this embodiment of the invention, vibration signals and corresponding operating parameters at key locations of the industrial rotating equipment are collected, and different types of data are processed for time synchronization, noise suppression, and scale unification, so that all data are established on the same time reference, providing a reliable data foundation for subsequent analysis.

[0038] Specifically, vibration signals and operating parameters of industrial rotating equipment are collected synchronously during operation. Vibration signals are acquired using accelerometers, velocity sensors, or displacement sensors to reflect changes in the mechanical response generated during equipment operation. Operating parameters include rotational speed and load. Rotational speed is obtained from a speed sensor installed on the rotating shaft or in the control system, while load is obtained from a torque sensor or motor current conversion. Both are used together to describe changes in the mechanical excitation state. After collecting vibration signals, rotational speed, and load, they are mapped to the same time series to establish a unified time reference, ensuring that vibration changes in subsequent calculations accurately correspond to specific changes in operating state. Furthermore, methods such as low-pass filtering, wavelet denoising, or empirical mode decomposition are used to remove interference components unrelated to equipment degradation, resulting in denoised vibration signals, which serve as input for subsequent degradation analysis. To eliminate the order-of-magnitude differences between different physical quantities and to provide a consistent numerical basis for subsequent processing, the Norm function was used to normalize the collected vibration signals, rotational speed, and load, mapping all data to a unified scale range [0, 1], thus obtaining the vibration value, rotational speed, and load at each sampling time. It is worth noting that the vibration values, rotational speed, and load used in subsequent data analysis are all normalized data and have no dimensions.

[0039] Step S102: Using vibration signals and operating condition parameters from historical healthy operation phases, fit a nonlinear health surface equation for the vibration response of rotational speed and load to characterize the baseline of the healthy operation state of industrial rotating equipment. Use the nonlinear health surface equation to perform vibration response deviation analysis on the vibration value at the current sampling time to obtain the relative deviation ratio of vibration. Based on the relative deviation ratio of rotational speed and load at the current sampling time, perform sensitivity compensation analysis to obtain the unexplainable vibration degradation coefficient at the current sampling time.

[0040] Existing methods for predicting the lifespan of industrial rotating equipment based on vibration signal analysis typically use indicators such as vibration amplitude or root mean square value to evaluate the health status of the equipment. However, in multi-condition switching scenarios, there is a highly nonlinear dynamic coupling relationship between vibration response and rotational speed and load. Simple linear models cannot accurately fit the normal vibration benchmark under multiple conditions, and early degradation under light load is easily masked. Therefore, in this embodiment of the invention, based on the characteristic that the vibration excitation of rotating machinery exhibits a nonlinear power law relationship with the changes in rotational speed (centrifugal force and alternating stress) and load (contact stress and structural deformation), the vibration signal, rotational speed, and load of the industrial rotating equipment during the historical normal operation phase (historical healthy operation phase) are used to construct a nonlinear response benchmark surface based on rotational speed and load under multi-dimensional operating conditions, so as to accurately characterize the theoretical healthy vibration response under different operating conditions.

[0041] Specifically, the vibration value, rotational speed, and load of the industrial rotating equipment at each historical sampling moment during its historical healthy operation phase are obtained. For any historical sampling moment in the historical healthy operation phase, a time window of a preset length is constructed with the specified historical sampling moment as the cutoff moment. The preset length is either a physical period or a fixed time, such as the time it takes for the shaft to rotate 10 times or a fixed duration of 1 second. Then, based on the vibration value at each sampling moment within the time window, the RMS value of the vibration value is calculated and recorded as the effective vibration value at that historical sampling moment. The effective vibration value, rotational speed, and load at that historical sampling moment are combined into a sample to obtain a sample set for the historical healthy operation phase. Nonlinear least multiplication or log-linear regression is used to perform nonlinear fitting on the sample set to obtain the nonlinear health surface equation of the rotational speed and load response to vibration.

[0042]

[0043] in, This represents the effective vibration value at the i-th historical sampling time. Represents the reference response coefficient. This represents the rotational speed at the i-th historical sampling moment. The load at the i-th historical sampling time, Indicates the power exponent of the rotational speed response. This represents the power exponent of the load response.

[0044] It should be noted that, These are all fixed values ​​obtained through nonlinear fitting. When rolling bearings and gears are subjected to load, their contact deformation conforms to Hertzian contact theory. Nonlinear stiffness leads to a nonlinear power-law relationship between vibration response and load. Therefore, the nonlinear health surface equation can serve as a "baseline for healthy operating state" of industrial rotating equipment, representing the theoretical normal vibration response of healthy industrial rotating equipment under different speed and load combinations.

[0045] After determining the "baseline of healthy operating state" for industrial rotating equipment, which is the nonlinear health surface equation of vibration response to speed and load, in this embodiment of the invention, the current sampling time t is used as the real-time time for predicting the lifespan of the industrial rotating equipment. The vibration value, speed, and load at the current sampling time t are obtained, and the vibration response deviation analysis of the vibration value at the current sampling time is performed using the nonlinear health surface equation to obtain the relative vibration deviation ratio:

[0046] Substituting the rotational speed and load at the current sampling moment into the nonlinear health surface equation yields the theoretical effective vibration value at the current sampling moment, which characterizes the vibration value of the industrial rotating equipment when it is in a healthy operating state at the current sampling moment. A time window of a preset length is constructed with the current sampling moment as the cutoff moment. Based on the vibration value at each sampling moment within the time window, the RMS value of the vibration value is calculated and recorded as the effective vibration value at the current sampling moment. The difference between the effective vibration value at the current sampling moment and the theoretical effective vibration value is calculated, and the maximum value between the difference and the constant 0 is taken as the unidirectional nonlinear deviation. The relative deviation ratio of the vibration at the current sampling moment is obtained by using the sum of the theoretical effective vibration value at the current sampling moment and the smallest real number as the denominator and the unidirectional nonlinear deviation as the numerator.

[0047] In one embodiment, the formula for calculating the relative deviation ratio of vibration at the current sampling time is:

[0048]

[0049] in, This indicates the relative deviation of the vibration at the current sampling time. This represents the function that takes the maximum value, and 0 represents a constant. This represents the effective value of the vibration at the current sampling time. This represents the effective vibration value obtained by substituting the rotational speed and load at the current sampling moment into the nonlinear health surface equation, which is also the theoretical effective vibration value. Indicates the reference response parameters. Indicates the power exponent of the rotational speed response. Indicates the power exponent of load response. This indicates the rotational speed at the current sampling moment. Indicates the load at the current sampling time. To represent extremely small real numbers, prevent the denominator from being 0.

[0050] It should be noted that, The deviation is characterized by subtracting the theoretical value (theoretical vibration effective value) calculated by the nonlinear health surface equation from the actual value (effective vibration value) at the current sampling time. When the actual value is less than or equal to the theoretical value, it indicates that there is no abnormality in the industrial rotating equipment, and the difference is taken as 0. When the actual value is greater than the theoretical value, the difference represents the absolute increment of abnormal vibration that cannot be explained by the nonlinear change of the operating condition, and then the health theoretical benchmark is used. The absolute increment of abnormal vibration is standardized into a relative increment ratio, thereby making it comparable under different working conditions.

[0051] To further address the issue of weak and easily masked degradation signals under light loads or low speeds, this embodiment of the invention applies an adaptive adjustment factor based on operating condition sensitivity to the relative vibration deviation ratio. This constructs a dimensionless, unexplainable vibration degradation coefficient with consistent degradation sensitivity across operating conditions, thereby achieving precise separation between equipment degradation factors and complex nonlinear operating condition responses. Specifically, based on the speed and load at the current sampling time, a sensitivity compensation analysis is performed on the relative vibration deviation ratio at the current sampling time to obtain the unexplainable vibration degradation coefficient at the current sampling time.

[0052] Obtain the product of the rotational speed and load at the current sampling moment. Use the sum of the product and a minimum real number as the denominator, and the product of the rated rotational speed and rated load of the industrial rotating equipment as the numerator to obtain the corresponding ratio. Use the ratio as the base and a preset sensitivity coefficient as the exponent to obtain the adaptive compensation factor for the operating condition sensitivity at the current sampling moment. Use the product of the adaptive compensation factor for the operating condition sensitivity and the relative deviation ratio of the vibration as the unexplainable vibration degradation coefficient at the current sampling moment.

[0053] In one embodiment, the formula for calculating the adaptive compensation factor for operating condition sensitivity at the current sampling time is:

[0054]

[0055] The formula for the unexplained vibration degradation coefficient at the current sampling time is:

[0056]

[0057] in, This represents the unexplained vibration degradation coefficient at the current sampling time. This represents the adaptive compensation factor for operating condition sensitivity at the current sampling time. This indicates the rotational speed at the current sampling moment. Indicates the load at the current sampling time. To represent extremely small real numbers, in order to prevent the denominator from being 0, This represents the reference excitation product under rated operating conditions, obtained by multiplying the rated speed and the rated load. It is the sensitivity coefficient, which ranges from 0 to 1. The larger the value, the greater the compensation.

[0058] It should be noted that under low speed or light load conditions ( or In smaller cases, the absolute vibration increment caused by physical degradation is often weak, and the compensation factor... The compensation factor will automatically increase, appropriately increasing the weight of abnormal signals under light load conditions to prevent early, minor damage from being masked under these conditions; while under high-speed, heavy-load conditions, the compensation factor will... Keep the value close to 1 to maintain the stability of the indicator.

[0059] Step S103: Construct a current analysis time window of a preset length with the current sampling time as the cutoff time. Utilize the unexplained vibration degradation coefficient of each sampling time within the current analysis time window to perform coordinated control of the unexplained vibration degradation coefficient of the current sampling time with consistent long and short time scales, thereby obtaining the trend-enhancing degradation coefficient of the current sampling time.

[0060] Unexplained vibration degradation coefficient Although it can characterize abnormal vibration responses caused by internal equipment degradation and reduce the impact of normal operating condition changes on health indicators, rapid changes in vibration response can still occur in a short period of time due to equipment start-up and shutdown, sudden load changes, and control adjustments. Such transient disturbances may still generate a large unexplained vibration degradation coefficient. This does not necessarily correspond to the actual performance degradation process of the device; it relies solely on the unexplained vibration degradation coefficient at the current sampling moment. Short-term shocks may still be misjudged as persistent degradation. Therefore, in this embodiment of the invention, for short-term severe shock disturbances caused by equipment start-up, shutdown, or sudden load increases, a statistical stability assessment mechanism based on a local historical window is further introduced. This mechanism extracts the local mean and local volatility of the unexplainable vibration degradation coefficient sequence, penalizes and suppresses isolated shocks with high volatility, and adds credibility to stable persistent degradation with low volatility, thus forming a trend-enhancing degradation coefficient with high temporal consistency.

[0061] Specifically, since equipment performance degradation inherently possesses fluctuating and discrete characteristics and irreversible monotonic directionality, this embodiment of the invention constructs a three-dimensional evaluation model. Through the coordinated modulation of arithmetic and nonlinear operators, it strongly suppresses highly discrete, bidirectional oscillating random impacts and enhances the credibility of low-discrete, monotonically increasing real physical degradation, thereby forming a trend-enhancing degradation coefficient with high physical self-consistency and temporal consistency. A current analysis time window of a preset length is constructed with the current sampling time as the cutoff time. This preset length can be set according to the equipment sampling period and the time scale of degradation changes; for example, setting the preset length to all historical sampling times within one day. Then, according to the aforementioned unexplainable vibration degradation coefficient... The acquisition scheme obtains the unexplained vibration degradation coefficient at each sampling moment within the current analysis time window, and uses the unexplained vibration degradation coefficient at each sampling moment within the current analysis time window to perform consistent co-regulation of the unexplained vibration degradation coefficient at the current sampling moment across both long and short time scales, thus obtaining the trend-enhancing degradation coefficient at the current sampling moment.

[0062] (1) Based on the unexplained vibration degradation coefficient at each sampling time within the current analysis time window, calculate the mean and standard deviation of the unexplained vibration degradation coefficient, and record them as the mean of the local degradation coefficient. and the standard deviation of the local degradation coefficient .

[0063] (2) Based on the unexplained vibration degradation coefficient at each historical sampling time during the historical healthy operation phase, calculate the mean of the unexplained vibration degradation coefficient and record it as the health baseline value. Based on the ratio between the health baseline value and the mean local degradation coefficient, the degradation magnitude accumulation factor at the current sampling time is obtained:

[0064]

[0065] in, This represents the cumulative factor of the degradation magnitude at the current sampling time. This represents the mean of the local degradation coefficient. Indicates health baseline values, For a very small real number, we need to prevent the denominator from being 0.

[0066] It should be noted that, Used to assess the fold deviation of the average abnormality level within a recent historical window (the current analysis time window) from a healthy baseline. When true degradation has occurred recently, the mean is greater than the typical level. The value will be greater than 1, giving the indicator a historical cumulative gain; if the device is healthy, The value will approach 1.

[0067] (3) Based on the mean and standard deviation of the local degradation coefficients, perform fluctuation dispersion analysis on the unexplainable vibration degradation coefficients within the current analysis time window to obtain the linear fluctuation dispersion penalty factor at the current sampling time: Calculate the coefficient of variation of the unexplainable vibration degradation coefficients within the current analysis time window based on the mean and standard deviation of the local degradation coefficients, and use the negative of the coefficient of variation as the independent variable of an exponential function with the natural constant as the base to obtain the linear fluctuation dispersion penalty factor at the current sampling time:

[0068]

[0069] in, This represents the linear fluctuation dispersion penalty factor at the current sampling time. This represents an exponential function with the natural constant as its base. This represents the standard deviation of the local degradation coefficient. This represents the mean of the local degradation coefficient. For a very small real number, we need to prevent the denominator from being 0.

[0070] It should be noted that, The coefficient of variation is represented by an exponential function that applies a negative feedback penalty. When isolated transient shocks or severe noise occur, the data dispersion of the unexplained vibration degradation coefficient within the current analysis time window is extremely high (the numerator is large, but the denominator has a small impact, resulting in a large coefficient of variation). The coefficient of variation is amplified, thus increasing the linear fluctuation dispersion penalty factor. Approaching 0, it forcibly suppresses false alarms caused by shocks; when actual physical degradation occurs in the equipment, the unexplained vibration degradation coefficient within the current analysis time window operates at a high steady-state level with small dispersion and the coefficient of variation approaching 0, thus reducing the linear fluctuation dispersion penalty factor. Approaching 1, it preserves the true degradation signal.

[0071] (4) Based on the difference in unexplained vibration degradation coefficients between adjacent sampling times within the current analysis time window, obtain the monotonic degradation direction consistency index at the current sampling time:

[0072]

[0073] in, This represents the consistency index of the monotonic degradation direction at the current sampling moment. This represents the function that takes the maximum value. This represents the unexplained vibration degradation coefficient at the (i+1)th sampling time within the current analysis time window. This represents the uninterpretable vibration degradation coefficient at the i-th sampling time within the current analysis time window, where || denotes the absolute value sign, and 0 represents a constant. Represents a very small real number. This indicates the number of sampling moments within the current analysis time window.

[0074] It should be noted that because mechanical damage is irreversible, the degradation process exhibits a monotonically increasing trend on a macroscopic scale, while noise or operational vibrations oscillate in both directions. Therefore, In the calculation formula, the numerator is activated by a function to extract only the positive upward increment, while the denominator accumulates the absolute values ​​of adjacent changes. In this case, if it's a random shock or jump disturbance, because a rise will inevitably be followed by a fall, the upward and downward amplitudes essentially cancel each other out. The numerator only accumulates the upward segment, while the denominator accumulates both the upward and downward segments, resulting in... The value drops sharply to around 0.5 or even lower, further suppressing false alarms from the perspective of directional consistency; if it is a true monotonic degradation, the unexplained vibrational degradation coefficient within the current analysis time window increases steadily, adjacent differences are all positive, and the numerator and denominator are completely equal, making... The value is exactly 1, and the degradation gain is unaffected.

[0075] (5) Using the cumulative factor of degradation magnitude, the penalty factor for linear fluctuation dispersion, and the consistency index of monotonic degradation direction at the current sampling time, the unexplainable vibration degradation coefficient at the current sampling time is synergistically regulated in the long term, short term, and consistency, so as to enhance accuracy and robustness while preserving the current degradation amplitude: the product of the cumulative factor of degradation magnitude, the penalty factor for linear fluctuation dispersion, and the consistency index of monotonic degradation direction at the current sampling time is obtained as the regulation coefficient, and the product of the regulation coefficient and the unexplainable vibration degradation coefficient at the current sampling time is obtained as the trend enhancement degradation coefficient at the current sampling time:

[0076]

[0077] in, This represents the trend enhancement degradation coefficient at the current sampling time.

[0078] Step S104: Use the trend enhancement degradation coefficient at the current sampling time to predict the remaining life of industrial rotating equipment.

[0079] Trend enhancement degradation coefficient While this indicator can reliably reflect the true degree of equipment degradation under rapid switching across multiple operating conditions, it only represents the numerical trend of health status changes and cannot directly indicate the remaining operational time of the equipment. Furthermore, in multi-condition scenarios, the equipment degradation rate is not constant; the rate of early, slow degradation differs significantly from the rate of accelerated degradation in the later stages. Equal-weighted fitting throughout the entire lifespan can lead to a lower estimation of the extrapolation slope based on early, slow data, resulting in overly optimistic predictions during the accelerated degradation phase and an inability to promptly respond to sudden changes in the degradation rate. Therefore, in this embodiment of the invention, based on... Based on historical data, a degradation trend fitting and extrapolation model is established. By identifying the starting point of accelerated degradation and processing it in segments, the degradation state values ​​are transformed into predicted remaining service life results. The equipment maintenance cycle is then dynamically adjusted based on these predictions. The steps for predicting equipment life using the trend-enhancing degradation coefficient are roughly as follows:

[0080] (1) Degradation index input: Trend enhancement degradation coefficient at the current sampling time Through operating condition separation and time consistency enhancement processing, the impact of speed and load changes during rapid operating condition switching on degradation judgment can be reduced, and the internal degradation trend of the equipment can be more accurately reflected. Therefore, the degradation coefficient is enhanced by the trend at the current sampling time. This is added to the trend enhancement degradation coefficient sequence (composed of trend enhancement degradation coefficients at historical sampling times) to form a new trend enhancement degradation coefficient sequence. .

[0081] (2) Identification of the starting point of accelerated degradation: Since the damage and degradation of industrial rotating equipment is a process of slow development followed by accelerated development, and is not a constant uniform change, it is necessary to identify the new trend-enhanced degradation coefficient sequence. Continuous monitoring of changes is conducted to determine whether industrial rotating equipment has transitioned from a stable degradation phase to a rapid degradation phase. Specifically, this involves monitoring the degradation coefficient sequence in the new trend. In this study, the changes in the trend-enhancing degradation coefficient between adjacent time points are detected. Simultaneously, the normal range of variation of the trend-enhancing degradation coefficient during the normal operation phase of the industrial rotating equipment is statistically analyzed, and this normal range is used as a reference for the industrial rotating equipment under normal operating conditions. When a new trend-enhancing degradation coefficient sequence... When the degradation coefficient changes beyond the normal range between multiple consecutive adjacent moments, it is determined that the internal damage development rate of the equipment has begun to increase significantly, and the first moment among multiple consecutive adjacent moments is taken as the degradation stage change node.

[0082] (3) Degradation trend fitting and extrapolation: Based on the judgment results of step (2), the degradation coefficient sequence of the new trend is enhanced. Trend analysis is performed on the changes in the equipment. When the equipment has entered the rapid degradation stage, the trend enhancement degradation coefficient after the degradation stage change node is used first for trend analysis, so that the analysis results are closer to the performance change state of the industrial rotating equipment at the current sampling time. When the equipment has not yet entered the rapid degradation stage, a new trend enhancement degradation coefficient sequence is used. Trend analysis is performed to ensure that lifespan calculations are always based on the latest equipment condition. This allows us to obtain the degradation trend of the industrial rotating equipment at the current sampling time and predict future equipment condition changes based on this degradation trend (i.e., predicting normal lifespan based on historical data during stable degradation, and predicting lifespan based only on accelerated degradation data during accelerated degradation).

[0083] (4) Failure Threshold Determination and Remaining Life Calculation: Based on historical operating data, fault data, or operating standards, determine the degradation threshold corresponding to when the equipment reaches an unacceptable operating state. Based on the degradation trend of the industrial rotating equipment at the current sampling time, predict the time required for the equipment to reach the degradation threshold, and use this time as the remaining life of the equipment. Finally, adjust the equipment maintenance cycle based on the obtained remaining life result to achieve predictive maintenance for industrial rotating equipment that can quickly switch between multiple operating conditions.

[0084] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis, characterized in that, The method includes: The vibration signals and operating parameters of the industrial rotating equipment are collected synchronously during operation, and the vibration value, rotational speed and load at each sampling moment are obtained accordingly. Using vibration signals and operating parameters from historical healthy operating phases, a nonlinear health surface equation for the vibration response of rotational speed and load is fitted to characterize the baseline of the healthy operating state of industrial rotating equipment. The vibration response deviation analysis is performed on the vibration value at the current sampling time using the nonlinear health surface equation to obtain the relative vibration deviation ratio. Sensitivity compensation analysis is performed on the relative vibration deviation ratio based on the rotational speed and load at the current sampling time to obtain the unexplainable vibration degradation coefficient at the current sampling time. A current analysis time window of a preset length is constructed with the current sampling time as the cutoff time. The unexplainable vibration degradation coefficient at each sampling time within the current analysis time window is used to coordinately regulate the unexplainable vibration degradation coefficient at the current sampling time with consistent long and short time scales, thereby obtaining the trend enhancement degradation coefficient at the current sampling time. The remaining life of industrial rotating equipment is predicted by using the trend enhancement degradation coefficient at the current sampling time.

2. The method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis according to claim 1, characterized in that, The process of fitting a nonlinear health surface equation for the vibration response of rotational speed and load to vibration using vibration signals and operating parameters from historical healthy operating phases includes: For any historical sampling moment during the historical healthy operation phase, a time window of a preset length is constructed with the sampling moment as the cutoff point. Based on the vibration value at each sampling moment within the time window, the RMS value of the vibration value is calculated and recorded as the effective vibration value at that historical sampling moment. The effective vibration value, rotational speed, and load at that historical sampling moment are combined into a sample to obtain a sample set for the historical healthy operation phase. Nonlinear fitting is performed on the sample set to obtain the nonlinear health surface equations of the rotational speed and load response to vibration. ； in, This represents the effective vibration value at the i-th historical sampling time. Represents the reference response coefficient. This represents the rotational speed at the i-th historical sampling moment. The load at the i-th historical sampling time, Indicates the power exponent of the rotational speed response. This represents the power exponent of the load response.

3. The method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis according to claim 1, characterized in that, The step of using the nonlinear health surface equation to perform vibration response deviation analysis on the vibration value at the current sampling time to obtain the relative vibration deviation ratio includes: Substituting the rotational speed and load at the current sampling moment into the nonlinear health surface equation yields the theoretical effective vibration value at the current sampling moment, which characterizes the vibration value of the industrial rotating equipment when it is in a healthy operating state at the current sampling moment. A time window of a preset length is constructed with the current sampling moment as the cutoff moment. Based on the vibration value at each sampling moment within the time window, the RMS value of the vibration value is calculated and recorded as the effective vibration value at the current sampling moment. The difference between the effective vibration value at the current sampling moment and the theoretical effective vibration value is calculated, and the maximum value between the difference and the constant 0 is taken as the unidirectional nonlinear deviation. The relative deviation ratio of the vibration at the current sampling moment is obtained by using the sum of the theoretical effective vibration value at the current sampling moment and the smallest real number as the denominator and the unidirectional nonlinear deviation as the numerator.

4. The method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis according to claim 1, characterized in that, The sensitivity compensation analysis of the relative deviation ratio of the vibration based on the rotational speed and load at the current sampling time yields the unexplainable vibration degradation coefficient at the current sampling time, including: Obtain the product of the rotational speed and load at the current sampling moment. Use the sum of the product and a minimum real number as the denominator, and the product of the rated rotational speed and rated load of the industrial rotating equipment as the numerator to obtain the corresponding ratio. Use the ratio as the base and a preset sensitivity coefficient as the exponent to obtain the adaptive compensation factor for the operating condition sensitivity at the current sampling moment. Use the product of the adaptive compensation factor for the operating condition sensitivity and the relative deviation ratio of the vibration as the unexplainable vibration degradation coefficient at the current sampling moment.

5. The method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis according to claim 1, characterized in that, The method involves using the unexplained vibration degradation coefficient at each sampling moment within the current analysis time window to perform consistent coordinated control of the unexplained vibration degradation coefficient at the current sampling moment across both long and short time scales, thereby obtaining the trend-enhancing degradation coefficient at the current sampling moment, including: Based on the unexplained vibration degradation coefficient at each sampling moment within the current analysis time window, calculate the mean and standard deviation of the unexplained vibration degradation coefficient, which are denoted as the mean of the local degradation coefficient and the standard deviation of the local degradation coefficient, respectively. Based on the unexplained vibration degradation coefficient at each historical sampling moment during the historical healthy operation phase, the mean of the unexplained vibration degradation coefficient is calculated and recorded as the health baseline value; based on the ratio between the health baseline value and the mean of the local degradation coefficient, the degradation magnitude accumulation factor at the current sampling moment is obtained. Based on the mean and standard deviation of the local degradation coefficients, a fluctuation dispersion analysis is performed on the unexplained vibration degradation coefficients within the current analysis time window to obtain the linear fluctuation dispersion penalty factor at the current sampling time; based on the difference in unexplained vibration degradation coefficients between adjacent sampling times within the current analysis time window, the monotonic degradation direction consistency index at the current sampling time is obtained. By using the cumulative factor of degradation magnitude, the penalty factor of linear fluctuation dispersion and the consistency index of monotonic degradation direction at the current sampling time, the unexplainable vibration degradation coefficient at the current sampling time is synergistically controlled to obtain the trend-enhancing degradation coefficient at the current sampling time.

6. The method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis according to claim 5, characterized in that, The step involves performing a fluctuation dispersion analysis on the unexplainable vibration degradation coefficients within the current analysis time window based on the mean and standard deviation of the local degradation coefficients, to obtain the linear fluctuation dispersion penalty factor at the current sampling time, including: The coefficient of variation of the unexplained vibration degradation coefficient within the current analysis time window is calculated based on the mean and standard deviation of the local degradation coefficient. The negative of the coefficient of variation is used as the independent variable of an exponential function with the natural constant as the base to obtain the linear fluctuation dispersion penalty factor at the current sampling time.

7. The method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis according to claim 5, characterized in that, The step of obtaining the monotonic degradation direction consistency index at the current sampling moment based on the difference in unexplained vibration degradation coefficients between adjacent sampling moments within the current analysis time window includes: ； in, This represents the consistency index of the monotonic degradation direction at the current sampling moment. This represents the function that takes the maximum value. This represents the unexplained vibration degradation coefficient at the (i+1)th sampling time within the current analysis time window. This represents the uninterpretable vibration degradation coefficient at the i-th sampling time within the current analysis time window, where || denotes the absolute value sign, and 0 represents a constant. Represents a very small real number. This indicates the number of sampling moments within the current analysis time window.

8. The method for predicting the lifespan of industrial rotating equipment by integrating multidimensional vibration signal analysis according to claim 5, characterized in that, The method of using the cumulative factor of degradation magnitude, the linear fluctuation dispersion penalty factor, and the monotonic degradation direction consistency index at the current sampling time to coordinately regulate the unexplainable vibration degradation coefficient at the current sampling time, thereby obtaining the trend-enhancing degradation coefficient at the current sampling time, includes: The product of the cumulative factor of degradation magnitude, the linear fluctuation dispersion penalty factor, and the monotonic degradation direction consistency index at the current sampling time is obtained as the control coefficient. The product of the control coefficient and the unexplained vibration degradation coefficient at the current sampling time is obtained as the trend enhancement degradation coefficient at the current sampling time.

Images