Method and system for monitoring degradation state of industrial equipment based on parameter variation

Abstract

The application discloses a parameter change-based industrial equipment degradation state monitoring method and system, and particularly relates to the technical field of industrial equipment monitoring, and comprises the following steps: S1, collecting multi-source signals during the operation of an industrial equipment, and outputting synchronous signals after pretreatment; S2, extracting signal features from the synchronous signals to form a feature set; S3, mapping the feature set into a comprehensive degradation index through a state mapping function with time self-adaptive weights, and constructing a nonlinear dynamic model to output an instantaneous change rate and a future trajectory sequence; S4, outputting a basic state grade through an adaptive threshold decision device, obtaining a modified state grade through a trend persistence verifier, calculating a cross-domain coupling strength, identifying a dominant coupling mode, and outputting a final degradation state grade through a coupling mode arbitrator; and S5, inputting the final degradation state grade and the future trajectory sequence into a decision device to output a maintenance instruction. The application improves the accuracy and reliability of degradation state determination and reduces the risk of failure shutdown.

Description

Method and System for Monitoring the Degradation Status of Industrial Equipment Based on Parameter Changes Technical Field

[0001] This invention relates to the field of industrial equipment monitoring technology, and more specifically, to a method and system for monitoring the degradation status of industrial equipment based on parameter changes. Background Technology

[0002] In industrial production scenarios, the continuous and stable operation of industrial equipment is a core prerequisite for ensuring production efficiency and reducing safety risks. Equipment degradation monitoring, as a key link in preventive maintenance, has become an important research direction in the field of industrial intelligence. Existing industrial equipment degradation monitoring technologies mainly collect single physical signals or a combination of a few types of signals during equipment operation, and combine them with fixed feature extraction methods and threshold judgment rules to achieve a preliminary judgment of the equipment's operating status, providing a basic reference for equipment maintenance.

[0003] However, existing technologies still have significant shortcomings in practical applications, making it difficult to meet the monitoring requirements of high precision and high reliability: First, the monitoring signal dimension is singular, only reflecting the local operating status of the equipment, and cannot comprehensively capture the comprehensive degradation process caused by multiple factors such as mechanical wear, electrical faults, and abnormal temperatures, resulting in one-sided monitoring results and easy omission of potential fault hazards; Second, the use of fixed thresholds for status determination does not consider the differences in operating conditions under different operating modes such as no-load, rated load, and overload, and the fixed thresholds do not match the actual operating conditions, which is prone to misjudgment or omission; Third, it ignores the coupling relationship between the characteristics of different physical domains such as mechanical domain, temperature domain, and current domain, and fails to identify the cross-domain coupling mode that dominates equipment degradation, resulting in a lack of scientific basis for status level determination and insufficient reliability; Fourth, it lacks the ability to accurately predict the future degradation trajectory of the equipment, and maintenance instructions are mostly remedial after the fact, unable to plan maintenance strategies in advance, making it difficult to fundamentally avoid the risk of downtime due to failure, resulting in problems such as excessive maintenance costs and shortened equipment lifespan. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method and system for monitoring the degradation status of industrial equipment based on parameter changes. The following solutions address the problems mentioned in the background art, such as single monitoring signal, poor threshold adaptability, neglect of cross-domain coupling correlation, and lack of predictive maintenance support.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] Methods for monitoring the degradation status of industrial equipment based on parameter changes include:

[0007] S1: Collects vibration, temperature and current signals from industrial equipment during operation, preprocesses all raw signals, and outputs a preprocessed synchronization signal.

[0008] S2: Based on the synchronization signal, extract mechanical vibration features from the vibration signal, extract temperature change features from the temperature signal, extract current change features from the current signal, and output a feature set containing multiple types of original features;

[0009] S3: Based on the feature set, a state mapping function is constructed to map the feature set into a comprehensive degradation index; a nonlinear dynamic model is constructed based on the index, and the instantaneous rate of change of the comprehensive degradation index and the predicted future trajectory sequence are calculated and output through the model.

[0010] S4: Input the comprehensive degradation index into the adaptive threshold decision maker based on the operating mode partition, and output the basic state level; input the instantaneous rate of change into the trend persistence verifier to verify and pre-upgrade the basic state level, and output the corrected state level; based on the feature set, calculate the cross-domain coupling strength by quantifying the coherence of the feature change trajectories of different physical domains, and identify the dominant coupling mode; finally, the coupling mode arbitrator performs final arbitration on the corrected state level according to the dominant coupling mode, generates the state confidence related to the mode, and outputs the final degradation state level;

[0011] S5: Input the final degradation state level and the future trajectory sequence into the decision-maker, and output the corresponding predictive maintenance instructions.

[0012] An industrial equipment degradation status monitoring system based on parameter changes includes:

[0013] Signal acquisition module: used to acquire vibration signals, temperature signals and current signals of industrial equipment during operation;

[0014] Signal preprocessing module: used to perform synchronization alignment, noise reduction filtering and amplitude normalization on all raw signals, and output the preprocessed synchronization signal;

[0015] Feature extraction module: used to extract mechanical vibration features, temperature change features and current change features based on synchronization signals, and output a feature set containing multiple types of original features;

[0016] The comprehensive index and prediction module is used to construct a time-adaptive weighted state mapping function based on the feature set, mapping the feature set to a comprehensive degradation index; based on this index, a nonlinear dynamic model is constructed, outputting the instantaneous rate of change of the comprehensive degradation index and the predicted future trajectory sequence;

[0017] State level determination module: includes an adaptive threshold decision unit, a trend persistence verifier, a coupling pattern recognition unit and a coupling pattern arbitrator, which are used to sequentially complete the basic state level determination, correct the state level output, the dominant coupling pattern recognition and the final state level arbitration, and output the final degraded state level with confidence.

[0018] Decision module: It receives the final degradation state level and future trajectory sequence, extracts and integrates parameters, determines validity and makes decisions based on scenario-specific logic, and outputs corresponding predictive maintenance instructions.

[0019] Preferably, the original signals include: vibration signals, temperature signals, and current signals during the operation of the industrial equipment;

[0020] The data collection method is as follows:

[0021] Vibration signal: A piezoelectric accelerometer is selected with a range of ±50g, a sensitivity of 200mV / g, and a frequency response range of 10Hz to 10kHz. It is installed in the radial and axial positions of the bearing housings at the drive end and non-drive end of the equipment.

[0022] Temperature signal: PT100 platinum resistance temperature sensor, accuracy class A, measurement range -20℃ to 200℃, accuracy ±0.15℃, matching temperature transmitter converts the resistance signal into a 4 to 20mA standard current signal; the sensor is arranged in two places, one is pre-embedded at the end of the motor stator winding, one for each of the U, V, and W phases, and the other is arranged on the surface of the bearing housing.

[0023] Current signal: Select a switchable Hall current sensor, model ACS758, with a measurement range of 0 to 200A, accuracy of ±1%, and response time ≤5μs. Install one sensor at the output of the device driver, one sensor for each of the U, V, and W phases.

[0024] Preferably, the feature set includes: mechanical vibration features, temperature change features, and current change features;

[0025] The mechanical vibration characteristics include kurtosis and root mean square value;

[0026] The temperature change characteristics include the winding temperature rise rate and the bearing housing temperature rise rate;

[0027] The current variation characteristics include the total harmonic distortion rate of the U-phase current, the total harmonic distortion rate of the V-phase current, and the total harmonic distortion rate of the W-phase current.

[0028] Preferably, the comprehensive degradation index is obtained by: constructing a time-adaptive weighted state mapping function, with the weight coefficient dynamically adjusted according to the equipment's operating time, emphasizing current and temperature characteristics in the early operating stage and vibration characteristics in the later stage; obtaining the comprehensive degradation index through multi-feature weighted fusion, with the comprehensive degradation index ranging from 0 to 1, where 0 represents no degradation and 1 represents complete failure; calculating each set of feature data group by group to obtain a continuous time series of comprehensive degradation index.

[0029] Preferably, the nonlinear dynamic model is constructed as follows: a composite structure of trend term and periodic fluctuation term is adopted; the model parameters are fitted by least squares method based on the historical operating data of the equipment and verified by 3 sets of independent fault samples; the comprehensive degradation index at the current moment is used as the initial input to complete the model initialization.

[0030] Preferably, the determination method of the basic status level is as follows: the equipment operation mode is divided into no-load mode, rated load mode and overload mode based on the feature set, and the current operation mode is determined by real-time feature parameter matching; for different operation modes, differentiated degradation thresholds are set based on historical fault data statistics, and the thresholds are dynamically adjusted with the running time; the comprehensive degradation index is compared with the adaptive threshold in the current mode, and three basic status levels, namely normal level, early warning level and alarm level, are output.

[0031] Preferably, the determination method for the corrected state level is as follows: setting the trend duration and degradation acceleration threshold; real-time monitoring of the instantaneous change rate, and statistical analysis of the change in the comprehensive degradation index within the trend duration; when the basic state level is normal and the change reaches the degradation acceleration threshold, pre-upgrading to the warning level; when the basic state level is warning and the change reaches the degradation acceleration threshold, pre-upgrading to the alarm level; when the basic state level is alarm or the change does not reach the degradation acceleration threshold, maintaining the original level, and outputting the corrected state level.

[0032] Preferably, the dominant coupling mode is determined as follows: the time trajectories of three types of physical domain features—mechanical domain, temperature domain, and current domain—are extracted from the feature set; the coherence of the feature trajectories of different physical domains is quantified using the Pearson correlation coefficient, and the coupling strength of the mechanical-temperature domain, mechanical-current domain, and temperature-current domain are calculated respectively; the combination of physical domains with the largest absolute value of coupling strength is selected as the dominant coupling mode, which is divided into three categories: mechanical-temperature coupling dominant, mechanical-current coupling dominant, and temperature-current coupling dominant; when there are two types of physical domains with equal absolute values ​​of coupling strength, they are judged according to a preset priority, with mechanical-temperature coupling dominant having the highest priority, followed by mechanical-current coupling dominant, and temperature-current coupling dominant having the lowest priority.

[0033] Preferably, the final degradation state level is determined as follows: Differentiated arbitration weights are set for different dominant coupling modes. Under the mechanical-temperature coupling dominant mode, the co-degradation weight of vibration features and temperature features accounts for 60%; under the mechanical-current coupling dominant mode, the co-degradation weight of vibration features and current features accounts for 60%; under the temperature-current coupling dominant mode, the co-degradation weight of temperature features and current features accounts for 60%; the state level is re-evaluated and corrected based on the arbitration weights. If all co-degradation features under the dominant coupling mode reach the threshold corresponding to the corrected state level, the level is maintained; if only a single feature meets the threshold, the level is downgraded by one level, with the lowest being the normal level; the state confidence is calculated based on the coupling strength of the dominant coupling mode, with the confidence value ranging from 0.7 to 1.0; the final degradation state level is output by integrating the arbitrated state level and confidence.

[0034] The technical effects and advantages of this invention are as follows:

[0035] 1. This invention integrates multiple core features from the mechanical domain (kurtosis, root mean square value), temperature domain (temperature rise rate of windings and bearing housings), and current domain (total harmonic distortion rate of three phases) by synchronously collecting vibration, temperature, and current signals of industrial equipment during operation. It also constructs a time-adaptive weighted state mapping function to dynamically adjust the weight ratio of each feature in the early and late stages of the equipment, thus fusing multi-dimensional features into a comprehensive degradation index. This avoids the one-sidedness of monitoring a single signal and can comprehensively and accurately depict the overall degradation state of the equipment, adapting to the monitoring needs of the entire equipment life cycle.

[0036] 2. This invention divides the operating modes into three categories: no-load, rated load, and overload, and sets differentiated adaptive thresholds for different modes. It combines a trend persistence verifier to monitor the degradation acceleration trend in real time, and uses the Pearson correlation coefficient to quantify the cross-domain coupling strength and identify the dominant coupling mode. The coupling mode arbitrator completes the state level verification and confidence generation. The multi-stage layer-by-layer verification mechanism effectively solves the problems of poor adaptability of fixed thresholds and neglect of cross-domain correlation, significantly reduces the risk of misjudgment and omission, and improves the reliability and credibility of degradation state level determination.

[0037] 3. This invention constructs a nonlinear dynamic model consisting of a "trend term + periodic fluctuation term," fits parameters based on the equipment's historical operating data over the past 12 months, and uses the fourth-order Runge-Kutta method to accurately predict the future trajectory sequence of comprehensive degradation indicators. It then combines this with scenario-based decision-making logic based on the final degradation state level to output targeted predictive maintenance instructions. This achieves accurate prediction of the equipment's future degradation trend, breaking the limitations of traditional monitoring technologies that rely on "post-event remediation." It enables advance planning of maintenance strategies, effectively avoiding downtime risks, reducing maintenance costs, and extending equipment lifespan. Attached Figure Description

[0038] Figure 1 is a schematic diagram of the overall method steps of the present invention;

[0039] Figure 2 is a schematic diagram of the signal acquisition and preprocessing process of the present invention;

[0040] Figure 3 is a schematic diagram of the feature extraction process of the present invention;

[0041] Figure 4 is a schematic diagram of the indicator modeling and prediction process of the present invention;

[0042] Figure 5 is a schematic diagram of the status determination and arbitration process of the present invention;

[0043] Figure 6 is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0044] 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 embodiments of the present invention, and not all embodiments. 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.

[0045] The industrial equipment degradation monitoring method based on parameter changes, as shown in Figures 1 to 5, includes:

[0046] S1: Collects vibration, temperature and current signals from industrial equipment during operation, preprocesses all raw signals, and outputs a preprocessed synchronization signal.

[0047] The process of collecting vibration, temperature, and current signals from industrial equipment during operation is as follows:

[0048] Vibration signal: A piezoelectric accelerometer, model PCB352C33, is selected, with a range of ±50g, sensitivity of 200mV / g, and frequency response range of 10Hz to 10kHz. It is installed on the radial and axial positions of the bearing housing at the drive end and non-drive end of the equipment. Magnetic installation is used under normal working conditions, and bolt fixing is used under severe vibration conditions. Before installation, the surface of the bearing housing should be cleaned of oil and rust. The sampling frequency is 10kHz, the sampling time per cycle is 10s, and a set of continuous sampling data is generated every 200ms.

[0049] Temperature signal: A PT100 platinum resistance temperature sensor with accuracy class A is selected, measuring range from -20℃ to 200℃ and accuracy ±0.15℃. The matching temperature transmitter converts the resistance signal into a standard current signal of 4 to 20mA. The sensor is arranged in two places: one is pre-embedded at the end of the motor stator winding, one for each of the U, V, and W phases, and is fixed in close contact with the winding wire by binding; the other is arranged on the surface of the bearing housing, using threaded installation, with a mounting hole depth ≥10mm; the sampling frequency is 1kHz, and the sampling period is synchronized with the vibration signal.

[0050] Current signal: A switchable Hall current sensor, model ACS758, is selected, with a measurement range of 0 to 200A, an accuracy of ±1%, and a response time of ≤5μs. It is installed at the output end of the device driver, with one sensor for each of the U, V, and W phases. It is fixed to the output cable using a switchable clamping method to ensure that the center of the sensor coincides with the cable axis. The sampling frequency is 5kHz, and it is triggered synchronously with vibration and temperature signals.

[0051] All raw signals are preprocessed to output a preprocessed synchronization signal. The specific process is as follows:

[0052] Synchronization alignment processing: Using the timestamp generated by FPGA hardware trigger as a reference, the raw vibration, temperature, and current signals are subjected to 16-bit AD digitization conversion; using the vibration signal timestamp as a reference axis, linear interpolation calibration is performed on the temperature and current digitized signals to correct sensor response delays. The PT100 sensor delay is 200μs, and the Hall current sensor delay is 150μs. The correction formula is: t sync =t raw +Δt where t raw The original timestamp of the sensor is Δt, where Δt is the calibration delay value. sync This is the timestamp after synchronization.

[0053] Noise reduction and filtering: The vibration signal is processed using a wavelet thresholding noise reduction algorithm, with a db4 wavelet basis and 5 decomposition levels. The threshold is calculated using: σ is the noise standard deviation, estimated by one layer of detail coefficients, and N is the signal length. The decomposed detail coefficients are subjected to soft thresholding. The temperature signal uses a moving average filtering algorithm with a window length of 100 sampling points. The current signal uses a combination algorithm of 50Hz notch filtering and Kalman filtering.

[0054] Amplitude normalization: The min-max normalization algorithm is used to normalize the amplitude of the three types of signals after noise reduction. The normalization formula is as follows: Where x raw x represents the original signal value after noise reduction. min x maxThese are the minimum and maximum values ​​of the signal's normal operating data over the past 30 days, respectively, and all signal amplitudes are mapped to the range of 0 to 1.

[0055] S2: Based on the synchronization signal, extract mechanical vibration features from the vibration signal, extract temperature change features from the temperature signal, extract current change features from the current signal, and output a feature set containing multiple types of original features;

[0056] The specific process for extracting mechanical vibration features from vibration signals is as follows:

[0057] Based on the preprocessed synchronous vibration signal, kurtosis and root mean square value are selected as the core mechanical vibration characteristics; specifically including:

[0058] Kurtosis calculation: The fourth-order central moment formula is used, and the formula is as follows: Where, x v,i For the i-th sampling point of the synchronous vibration signal, μ v σ is the mean of the synchronous vibration signal. v N represents the standard deviation of the synchronous vibration signal. v This represents the total number of sampling points for a single set of synchronous vibration signals.

[0059] Root mean square (RMS) value calculation: The effective value formula is used, and the formula is as follows: This formula reflects the energy intensity of the vibration signal and characterizes the overall severity of the mechanical vibration of the equipment.

[0060] For each 200ms set of synchronous vibration signals, the corresponding kurtosis and root mean square value are calculated, and finally a mechanical vibration feature set containing kurtosis and root mean square value is obtained.

[0061] The specific process for extracting temperature change features from a temperature signal is as follows:

[0062] Based on the preprocessed synchronous temperature signal, the temperature rise rate is selected as the core temperature change characteristic; the temperature rise rate is calculated using the sliding time window method, specifically including:

[0063] Time window parameter settings: Set the sliding time window length to 10,000 sampling points, corresponding to a duration of 10 seconds;

[0064] Temperature rise rate calculation: It is calculated by the ratio of the average temperature difference between two adjacent sliding time windows to the duration of the time window interval. The formula is as follows: in, The average temperature within the k-th sliding time window. The average temperature within the (k-1)th sliding time window, Δt T The interval between two adjacent sliding time windows is 10 seconds.

[0065] The temperature rise rate is calculated for the collected winding temperature signal and bearing housing temperature signal respectively, and finally a temperature change feature set including the winding temperature rise rate and bearing housing temperature rise rate is obtained.

[0066] The specific process for extracting current change features from a current signal is as follows:

[0067] Based on the preprocessed synchronous current signal, the total harmonic distortion rate (THD) is selected as the core current variation characteristic; specifically including:

[0068] Signal frequency domain decomposition: Perform fast Fourier transform on the three-phase synchronous current signal to decompose it into the fundamental component and harmonic components of each phase current.

[0069] Total harmonic distortion (THD) calculation: The formula is as follows: Among them, I f,1 I is the effective value of the fundamental component of a single-phase signal. h,2 to I h,n These are the effective values ​​of the 2nd to nth harmonic components of the phase signal, respectively.

[0070] For each group of synchronous three-phase current signals every 200ms, the corresponding total harmonic distortion rate is calculated, and finally the current change feature set including the total harmonic distortion rates of the U-phase, V-phase, and W-phase currents is obtained.

[0071] By integrating the aforementioned mechanical vibration feature set, temperature change feature set, and current change feature set, a feature set containing multiple types of original features is formed.

[0072] S3: Based on the feature set, a state mapping function is constructed to map the feature set into a comprehensive degradation index; a nonlinear dynamic model is constructed based on the index, and the instantaneous rate of change of the comprehensive degradation index and the predicted future trajectory sequence are calculated and output through the model.

[0073] Based on the aforementioned feature set, a state mapping function is constructed to map the feature set into a comprehensive degradation index. The specific process is as follows:

[0074] Based on the aforementioned feature set (including kurtosis K, root mean square value RMS, winding temperature rise rate θ', bearing housing temperature rise rate θ'', and total harmonic distortion (THD) of U-phase current), U Total Harmonic Distortion (THD) of Phase V Current V Total Harmonic Distortion (THD) of Phase W Current W Constructing a time-adaptive weighted state mapping function, specifically including:

[0075] Adaptive weight setting: The weight coefficient is dynamically adjusted according to the equipment's operating time, based on the current operating time t and the equipment's designed rated lifespan T. total The ratio is the basis for stage determination - when t / Ttotal When the efficiency is ≤70%, it is considered the early operating stage, focusing on current and temperature characteristics (reflecting initial operating condition stability); when t / T total When the percentage is >70%, it is considered the later stage of operation, focusing on vibration characteristics (reflecting accumulated mechanical wear); the weighting calculation reference formula is: Where, ω i (t) represents the weight coefficient of the i-th feature at time t, α i The feature base weight coefficient (reference value: α) K =0.2、α RMS =0.18、α θ' =0.16、α θ'' =0.16、α THD_U =0.09、α THD_V =0.09、α THD_W =0.12), β i The weighted attenuation / enhancement coefficient (vibration characteristic β) K =β RMS =0.8, temperature and current characteristics β i =-0.5), t is the current runtime, T total Design the equipment to have a rated lifespan;

[0076] Comprehensive degradation index fusion: Multi-feature weighted fusion is achieved through a state mapping function, the formula is as follows: Where D(t) is the comprehensive degradation index at time t (range 0~1, 0 represents no degradation, 1 represents complete failure), F i (t) is the standardized value of the i-th feature at time t; the feature set data is calculated group by group every 200ms to finally obtain the continuous comprehensive degradation index time series.

[0077] A nonlinear dynamic model is constructed based on the time series of comprehensive degradation indicators. The specific process is as follows:

[0078] Model structure design: A composite structure of "trend term + periodic fluctuation term" is adopted to closely reflect the actual characteristics of equipment degradation accumulation and periodic fluctuations in operating conditions. The core form of the model is as follows: in, The instantaneous change rate of the comprehensive degradation index is given by , where a and b are degradation trend coefficients (a>0, b>0, indicating that degradation accumulates and increases over time), c is the periodic fluctuation amplitude, and ω is the fluctuation angular frequency (calculated from the equipment operating cycle T, ω=2π / T). Let d be the initial phase, and d be the basic degradation rate constant.

[0079] Model parameter fitting: Historical operating data of the equipment over the past 12 months (covering normal, slight degradation, and moderate degradation stages) were selected, and the least squares method was used to fit the model parameters a, b, c, ω. The fitting error was controlled within 5%; after fitting, the model was validated through 3 sets of independent fault samples to ensure that the prediction deviation of the degradation trend was ≤8%.

[0080] Model initialization: The comprehensive degradation index D(t0) at the current time (denoted as t0) is used as the initial input and substituted into the fitted model to complete the model initialization.

[0081] The instantaneous rate of change of the comprehensive degradation index and the predicted future trajectory sequence are calculated and output through the model. The specific process is as follows:

[0082] Instantaneous rate of change calculation: Substitute D(t0) at time t0 into the model to directly solve for the instantaneous rate of change of the comprehensive degradation index at the current time. This parameter reflects the current degradation rate of the equipment; the larger the value, the faster the degradation process.

[0083] Future trajectory sequence prediction: The model is numerically solved using the fourth-order Runge-Kutta method, with a prediction step size of 200ms. The prediction duration can be set as needed (usually 72 hours in the future). Through iterative solution, the predicted value of the comprehensive degradation index corresponding to each 200ms in the future is obtained and arranged in chronological order to form the future trajectory sequence.

[0084] S4: Input the comprehensive degradation index into the adaptive threshold decision maker based on the operating mode partition, and output the basic state level; input the instantaneous rate of change into the trend persistence verifier to verify and pre-upgrade the basic state level, and output the corrected state level; based on the feature set, calculate the cross-domain coupling strength by quantifying the coherence of the feature change trajectories of different physical domains, and identify the dominant coupling mode; finally, the coupling mode arbitrator performs final arbitration on the corrected state level according to the dominant coupling mode, generates the state confidence related to the mode, and outputs the final degradation state level;

[0085] The comprehensive degradation index is input into an adaptive threshold decision maker based on operating mode partitioning, and the basic state level is output. The specific process is as follows:

[0086] Operating mode partitioning: Based on the feature set, the equipment operating mode is divided into three categories: no-load mode (current value ≤ 20% of rated current), rated load mode (current value between 20% and 80% of rated current), and overload mode (current value ≥ 80% of rated current). The current operating mode is determined by real-time feature parameter matching.

[0087] Adaptive threshold setting: Different degradation thresholds are set based on historical fault data statistics for different operating modes. The thresholds are dynamically adjusted over operating time (the adjustment coefficient is based on the degree of equipment aging). The basic threshold reference values ​​for each mode are: under no-load mode, the comprehensive degradation index thresholds D1=0.65, D2=0.80; under rated load mode, the thresholds D1=0.60, D2=0.75; under overload mode, the thresholds D1=0.55, D2=0.70.

[0088] Basic status level determination: The comprehensive degradation index D(t) is compared with the adaptive threshold in the current mode, and three basic status levels are output: normal level D(t) < D1, warning level D1 ≤ D(t) < D2, and alarm level D(t) ≥ D2, so as to obtain the basic status level of the current device.

[0089] The instantaneous rate of change is input into the trend persistence verifier to verify and pre-upgrade the basic state level, and output the corrected state level. The specific process is as follows:

[0090] Verification parameter settings: Set the duration Δt for trend persistence determination. verify =5min, degradation acceleration threshold ΔD threshold =0.05 (i.e., an increase of ≥0.05 in the comprehensive degradation index within 5 minutes is considered accelerated degradation);

[0091] Trend persistence detection: Real-time monitoring of instantaneous rate of change Statistical Δt verify The change in the comprehensive degradation index over the time period is ΔD = D(t + Δt). verify )-D(t);

[0092] State level pre-upgrade: If the base state level is normal, but ΔD ≥ ΔD threshold The pre-upgrade is to the warning level; if the basic status level is the warning level and ΔD ≥ ΔD threshold The status is pre-escalated to an alarm level; if the base status level is an alarm level, or ΔD < ΔD threshold The original level remains unchanged, and the final output is the corrected status level.

[0093] Based on the aforementioned feature set, the cross-domain coupling strength is calculated by quantifying the coherence of the feature change trajectories in different physical domains, and the dominant coupling mode is identified. The specific process is as follows:

[0094] Feature trajectory extraction: Extracting time trajectories of three types of physical domain features from the feature set—mechanical domain (kurtosis K, root mean square value RMS), temperature domain (winding temperature rise rate θ', bearing housing temperature rise rate θ''), and current domain (U / V / W phase total harmonic distortion rate THD). U / THD V / THDW The trajectory data is taken from continuous feature values ​​within the last 10 minutes;

[0095] Coupling strength calculation: The Pearson correlation coefficient is used to quantify the coherence of characteristic trajectories in different physical domains, and is used as an evaluation index for coupling strength. The formula is as follows: Where, r xy Let X(t) represent the coupling strength between physical domain X and physical domain Y (ranging from -1 to 1, with the absolute value closer to 1 indicating stronger coupling), and let X(t) and Y(t) represent the characteristic trajectory values ​​of the two physical domains, respectively. , where are the mean values ​​of the two trajectories, and m is the number of trajectory data points; calculate the coupling strengths r1, r2, and r3 in the mechanical-temperature domain, mechanical-current domain, and temperature-current domain, respectively;

[0096] Dominant Coupling Mode Identification: The combination of physical domains with the largest absolute value of coupling strength is selected as the dominant coupling mode, which is divided into three typical modes: mechanical-temperature coupling dominant (|r1| maximum), mechanical-current coupling dominant (|r2| maximum), and temperature-current coupling dominant (|r3| maximum). If the absolute values ​​of coupling strength of two types of physical domains are equal, "mechanical-temperature coupling dominant" is determined first; if temperature-current coupling and mechanical-current coupling are equal, "mechanical-current coupling dominant" is determined first. Finally, the dominant coupling mode of the current device is determined.

[0097] The coupled-mode arbitrator performs final arbitration on the modified state level based on the dominant coupled mode, generates the state confidence related to the mode, and outputs the final degenerate state level. The specific process is as follows:

[0098] Arbitration rules are set as follows: Different arbitration weights are set for different dominant coupling modes. Under the mechanical-temperature coupling dominant mode, the weight of the co-degradation of vibration characteristics and temperature characteristics accounts for 60%; under the mechanical-current coupling dominant mode, the weight of the co-degradation of vibration characteristics and current characteristics accounts for 60%; under the temperature-current coupling dominant mode, the weight of the co-degradation of temperature characteristics and current characteristics accounts for 60%. Among them, "co-degradation characteristics" refers to the core characteristics of each domain (kurtosis in the mechanical domain, winding temperature rise rate in the temperature domain, and three-phase total harmonic distortion rate in the current domain).

[0099] State level arbitration: The state level is reassessed and corrected based on the arbitration weight. If all the collaborative degradation features in the dominant coupling mode reach the threshold corresponding to the corrected state level, the corrected state level is maintained; if only a single feature meets the standard, the level is downgraded by one level (the lowest being the normal level).

[0100] State confidence generation: The confidence level is calculated based on the coupling strength of the dominant coupling mode, using the formula: C = 0.7 + 0.3 × |r max |where r maxis the maximum coupling strength, and the confidence level ranges from 0.7 to 1.0;

[0101] Final result output: Integrate the status level and confidence level after arbitration, and output the final degradation status level (including the level name and confidence level), such as "Warning level (confidence level: 0.92)".

[0102] S5: Input the final degradation status level and the future trajectory sequence into the decision maker together, and output the corresponding predictive maintenance instruction.

[0103] First, preprocess and verify the input final degradation status level and future trajectory sequence, specifically including:

[0104] Parameter extraction and integration: Extract the level type (normal level, warning level, alarm level) and status confidence level C from the final degradation status level; extract key prediction information from the future trajectory sequence, including the peak value D of the comprehensive degradation index within the prediction duration (usually 72 hours) max 、the prediction time t when reaching D1 (warning threshold) and D2 (alarm threshold) under the corresponding mode D1 、t D2 ,and the value D of the comprehensive degradation index at the end of the prediction, end to form a unified decision input data set;

[0105] Validity determination: If the status confidence level C < 0.7, recalculate the final degradation status level; if the proportion of outliers (outside the 0 - 1 interval) in the future trajectory sequence ≥ 5%, regenerate the future trajectory sequence; after both types of parameters pass the verification, enter the decision logic determination link to finally obtain a valid decision input data set.

[0106] Based on the difference in the final degradation status level, combined with the future trajectory prediction trend, construct decision logics for different scenarios and match the corresponding predictive maintenance instructions, specifically including:

[0107] Scenario 1: The final degradation status level is the normal level:

[0108] Core determination basis: Whether there is a potential risk in the degradation trend in the next 72 hours;

[0109] If D max < D1 and D end - D(t0) ≤ 0.03 (D(t0) is the current comprehensive degradation index), it is determined that there is no short - term degradation risk; Output "Regular inspection instruction": Perform inspections according to the original cycle (such as once a week), record the basic data of vibration and temperature, and no additional maintenance operations are required;

[0110] If D1 < D max < D2 and t D1If the time is ≤72h, it is determined that there is a potential early warning risk; output "enhanced monitoring instruction": increase the data acquisition frequency to twice the original frequency (vibration 20kHz, temperature 2kHz, current 10kHz), increase the inspection frequency to once a day, and focus on tracking the changes in vibration kurtosis and temperature rise rate;

[0111] If D max ≥D2 and t D2 If the time is ≤72h, it is considered an emergency degradation risk; output "Status Review Instruction": immediately re-execute the full process assessment, check for sensor failures or data anomalies, and simultaneously arrange for technical personnel to conduct on-site investigation.

[0112] Scenario 2: The final degradation state level is the warning level.

[0113] The core judgment criteria are: the level of confidence and the urgency of reaching the alarm threshold (matching sub-scenes from high to low urgency, and no longer matching low-priority scenarios when high-priority conditions are met).

[0114] If C ≥ 0.8 and t D2 If the time limit is ≤48h, it is considered an emergency degradation risk; output "time-limited maintenance instruction": immediately reduce the equipment load to below 50% of the rated load, complete the shutdown maintenance within 24 hours, and prioritize the investigation of mechanical wear (bearings, rotors) and electrical circuit faults;

[0115] If 0.7 ≤ C < 0.9 and 48 <t D2 If the time is ≤72h, it is considered a risk of excessive degradation; an "urgent preparation instruction" is issued: complete the maintenance plan and spare parts preparation within 24 hours, and be ready for shutdown maintenance within 48 hours;

[0116] If C ≥ 0.9 and t D2 >72h, judged as moderate degradation risk; output "planned maintenance instruction": complete the maintenance plan within 3 days, arrange shutdown maintenance within 7 days, maintenance content includes bearing lubrication inspection, winding insulation test, and current circuit tightening;

[0117] Scenario 3: The final degradation state level is the alarm level:

[0118] The core criterion for judgment is whether there is a risk of rapid failure in the future trajectory.

[0119] If t D2 If the time limit is ≤24h (the critical value for complete failure), it is judged as an extremely urgent failure risk; an "emergency shutdown maintenance command" is output: immediately stop the machine and prohibit further operation. The technical team shall arrive within 2 hours to carry out a comprehensive overhaul, focusing on checking for fatal faults such as bearing seizure, winding short circuit, and current overload.

[0120] If 24 <t D2If the time is ≤72h, it is judged as a high risk of emergency degradation; output "priority maintenance instruction": suspend non-essential production tasks, complete shutdown preparation within 4 hours, start emergency maintenance process, and conduct no-load and load test runs for verification after maintenance.

[0121] The industrial equipment degradation status monitoring system based on parameter changes, as shown in Figure 6, includes:

[0122] Signal acquisition module: used to acquire vibration signals, temperature signals and current signals of industrial equipment during operation;

[0123] Signal preprocessing module: used to perform synchronization alignment, noise reduction filtering and amplitude normalization on all raw signals, and output the preprocessed synchronization signal;

[0124] Feature extraction module: used to extract mechanical vibration features, temperature change features and current change features based on synchronization signals, and output a feature set containing multiple types of original features;

[0125] The comprehensive index and prediction module is used to construct a time-adaptive weighted state mapping function based on the feature set, mapping the feature set to a comprehensive degradation index; based on this index, a nonlinear dynamic model is constructed, outputting the instantaneous rate of change of the comprehensive degradation index and the predicted future trajectory sequence;

[0126] State level determination module: includes an adaptive threshold decision unit, a trend persistence verifier, a coupling pattern recognition unit and a coupling pattern arbitrator, which are used to sequentially complete the basic state level determination, correct the state level output, the dominant coupling pattern recognition and the final state level arbitration, and output the final degraded state level with confidence.

[0127] Decision module: It receives the final degradation state level and future trajectory sequence, extracts and integrates parameters, determines validity and makes decisions based on scenario-specific logic, and outputs corresponding predictive maintenance instructions.

[0128] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0129] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring the degradation status of industrial equipment based on parameter changes, characterized in that, include: S1: Collects vibration, temperature and current signals from industrial equipment during operation, preprocesses all raw signals, and outputs a preprocessed synchronization signal. S2: Based on the synchronization signal, extract mechanical vibration features from the vibration signal, temperature change features from the temperature signal, and current change features from the current signal, outputting a feature set containing multiple types of original features; S3: Based on the feature set, map the feature set to a comprehensive degradation index by constructing a state mapping function; construct a nonlinear dynamic model based on this index, and calculate and output the instantaneous change rate of the comprehensive degradation index and the predicted future trajectory sequence through the model; the state mapping function is constructed as follows: based on the feature set, construct a time-adaptive weighted state mapping function, the weight coefficients are dynamically adjusted with the equipment operating time, and the ratio of the current operating time to the equipment's designed rated lifespan is used as the stage determination criterion, and the formula is used to determine the stage. The weighting coefficients are calculated, where ω i (t) represents the weight coefficient of the i-th feature at time t, α i β represents the feature-based weight coefficients. i T is the weight attenuation or enhancement coefficient, t is the current running time of the device, and T is the weight attenuation or enhancement coefficient. total The equipment is designed with a rated lifespan; a multi-feature weighted fusion is performed through a state mapping function, mapping the feature set into a comprehensive degradation index, as shown in the formula. Where D(t) is the comprehensive degradation index at time t, F i (t) is the standardized value of the i-th feature at time t; S4: Input the comprehensive degradation index into the adaptive threshold decision-maker based on the operation mode partition, and output the basic state level; input the instantaneous rate of change into the trend persistence verifier to verify and pre-upgrade the basic state level, and output the corrected state level; based on the feature set, calculate the cross-domain coupling strength by quantifying the coherence of the feature change trajectories of different physical domains, and identify the dominant coupling mode; Finally, the coupled mode arbitrator performs final arbitration on the modified state level according to the dominant coupled mode, generates the state confidence related to the mode, and outputs the final degraded state level; S5: The final degraded state level and the future trajectory sequence are input into the decision-maker, and the corresponding predictive maintenance instructions are output.

2. The method for monitoring the degradation status of industrial equipment based on parameter changes according to claim 1, characterized in that: The raw signals include: vibration signals, temperature signals, and current signals during the operation of industrial equipment; the acquisition methods are as follows: Vibration signal: a piezoelectric accelerometer is selected, with a range of ±50g, sensitivity of 200mV / g, and frequency response range of 10Hz to 10kHz, installed in the radial and axial positions of the bearing housings at the drive end and non-drive end of the equipment; Temperature signal: a PT100 platinum resistance temperature sensor is used, with an accuracy class of A, a measurement range of -20℃ to 200℃, and an accuracy of ±0.15℃. A matching temperature transmitter converts the resistance signal into a standard current signal of 4 to 20mA; the sensor is arranged in two locations: one is pre-embedded at the end of the motor stator winding, with one sensor for each of the U, V, and W phases; the other is arranged on the surface of the bearing housing; Current signal: an ACS758 Hall effect current sensor is selected, with a measurement range of 0 to 200A, an accuracy of ±1%, and a response time ≤5μs, installed at the output end of the equipment driver, with one sensor for each of the U, V, and W phases.

3. The method for monitoring the degradation status of industrial equipment based on parameter changes according to claim 1, characterized in that: The feature set includes: mechanical vibration features, temperature change features, and current change features; the mechanical vibration features include kurtosis and root mean square value; the temperature change features include winding temperature rise rate and bearing housing temperature rise rate; the current change features include total harmonic distortion rate of U-phase current, total harmonic distortion rate of V-phase current, and total harmonic distortion rate of W-phase current.

4. The method for monitoring the degradation status of industrial equipment based on parameter changes according to claim 1, characterized in that: The method for obtaining the comprehensive degradation index is as follows: a time-adaptive weighted state mapping function is constructed, and the weight coefficients are dynamically adjusted with the equipment operating time. In the early operating stage, current and temperature characteristics are emphasized, and in the later stage, vibration characteristics are emphasized. The comprehensive degradation index is obtained by weighted fusion of multiple features. The comprehensive degradation index ranges from 0 to 1, where 0 represents no degradation and 1 represents complete failure. The comprehensive degradation index is calculated for each set of feature data to obtain a continuous time series of comprehensive degradation index.

5. The method for monitoring the degradation status of industrial equipment based on parameter changes according to claim 1, characterized in that: The nonlinear dynamic model is constructed as follows: a composite structure of trend term and periodic fluctuation term is adopted; based on the historical operating data of the equipment, the least squares method is used to fit the model parameters, and the model is verified by three sets of independent fault samples; the comprehensive degradation index at the current moment is used as the initial input to complete the model initialization.

6. The method for monitoring the degradation status of industrial equipment based on parameter changes according to claim 1, characterized in that: The determination method for the basic status level is as follows: Based on the feature set, the equipment operation mode is divided into no-load mode, rated load mode and overload mode, and the current operation mode is determined by real-time feature parameter matching; for different operation modes, differentiated degradation thresholds are set based on historical fault data statistics, and the thresholds are dynamically adjusted with the running time; the comprehensive degradation index is compared with the adaptive threshold in the current mode, and three basic status levels, namely normal level, early warning level and alarm level, are output.

7. The method for monitoring the degradation status of industrial equipment based on parameter changes according to claim 1, characterized in that: The determination method for the modified state level is as follows: set the trend duration and degradation acceleration threshold; monitor the instantaneous change rate in real time, and count the change of the comprehensive degradation index within the trend duration; when the basic state level is normal and the change reaches the degradation acceleration threshold, it is pre-upgraded to the warning level. When the basic status level is a warning level and the change reaches the degradation acceleration threshold, it is pre-escalated to an alarm level. When the base status level is an alarm level or the change amount has not reached the degradation acceleration threshold, the original level remains unchanged, and the corrected status level is output.

8. The method for monitoring the degradation status of industrial equipment based on parameter changes according to claim 1, characterized in that: The dominant coupling mode is determined as follows: the time trajectories of the three physical domains—mechanical, temperature, and current—are extracted from the feature set; the coherence of the feature trajectories of different physical domains is quantified using the Pearson correlation coefficient, and the coupling strength of the mechanical-temperature domain, mechanical-current domain, and temperature-current domain are calculated respectively. The combination of physical domains with the largest absolute value of coupling strength is selected as the dominant coupling mode, and is divided into three categories: mechanical-temperature coupling dominant, mechanical-current coupling dominant, and temperature-current coupling dominant. When two types of physical domains have equal absolute values ​​of coupling strength, they are judged according to a preset priority. Mechanical-temperature coupling has the highest priority, followed by mechanical-current coupling, and then temperature-current coupling has the lowest priority.

9. The method for monitoring the degradation status of industrial equipment based on parameter changes according to claim 1, characterized in that: The final degradation state level is determined as follows: Differentiated arbitration weights are set for different dominant coupling modes. Under the mechanical-temperature coupling dominant mode, the weight of the co-degradation of vibration features and temperature features accounts for 60%; under the mechanical-current coupling dominant mode, the weight of the co-degradation of vibration features and current features accounts for 60%; under the temperature-current coupling dominant mode, the weight of the co-degradation of temperature features and current features accounts for 60%. The state level is re-evaluated and corrected based on the arbitration weights. If all co-degradation features under the dominant coupling mode reach the threshold corresponding to the corrected state level, the level is maintained. If only a single feature meets the standard, the level is downgraded by one level, with the lowest being the normal level. The state confidence is calculated based on the coupling strength of the dominant coupling mode, with a confidence value ranging from 0.7 to 1.

0. The final degraded state level is output by integrating the arbitrated state level and the confidence.

10. An industrial equipment degradation status monitoring system based on parameter changes, used to implement the industrial equipment degradation status monitoring method based on parameter changes as described in any one of claims 1-9, characterized in that, include: Signal acquisition module: used to acquire vibration signals, temperature signals and current signals of industrial equipment during operation; The signal preprocessing module performs synchronization alignment, noise reduction filtering, and amplitude normalization on all raw signals, outputting a preprocessed synchronization signal. The feature extraction module extracts mechanical vibration features, temperature change features, and current change features from the synchronization signal, outputting a feature set containing multiple types of raw features. The comprehensive index and prediction module constructs a time-adaptive weighted state mapping function based on the feature set, mapping the feature set to a comprehensive degradation index. Based on this index, it constructs a nonlinear dynamic model, outputting the instantaneous rate of change of the comprehensive degradation index and the predicted future trajectory sequence. The state level determination module includes an adaptive threshold decision unit, a trend persistence validator, a coupled pattern recognition unit, and a coupled pattern arbitrator, sequentially performing basic state level determination, corrected state level output, dominant coupled pattern recognition, and final state level arbitration, outputting the final degradation state level with confidence. Decision module: It receives the final degradation state level and future trajectory sequence, extracts and integrates parameters, determines validity and makes decisions based on scenario-specific logic, and outputs corresponding predictive maintenance instructions.

Images