A frame production line key station equipment predictive maintenance and health management system
By generating low-energy excitation signals through pattern arbitration and adaptive sampling strategies, and combining them with dual health parameters to assess equipment status, the problems of diagnostic accuracy and production interference in equipment maintenance are solved, and the stable and efficient operation of the chassis production line is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- XIAOYANG MASCH XIANGSHUI CO LTD
- Filing Date
- 2026-03-05
- Publication Date
- 2026-06-05
Smart Images
Figure CN122155687A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of equipment maintenance technology, and more specifically, to a predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line. Background Technology
[0002] Key equipment at critical workstations on the chassis production line (such as welding robots and servo tightening shafts) is crucial for ensuring chassis machining accuracy and production continuity. Its maintenance and management level directly impacts production line capacity and product quality. Currently, in the industrial sector, the maintenance of such equipment primarily employs two main models: periodic preventative maintenance and post-failure corrective maintenance. Periodic preventative maintenance involves performing inspections according to a pre-set fixed cycle, focusing on identifying potential problems according to plan. Post-failure corrective maintenance, on the other hand, restores equipment functionality by repairing or replacing components after a failure and is a more common passive maintenance method in production.
[0003] In actual production scenarios, existing maintenance technologies suffer from numerous unavoidable problems. Dynamic changes in production cycles directly compress the available time for equipment diagnostics, and traditional maintenance technologies lack the ability to adapt to these time constraints. This leads to insufficient data collection during periods of heavy production and shortened diagnostic windows, affecting the accuracy of fault identification and making it difficult to detect early signs of minor performance degradation. Furthermore, the excitation signals used in existing diagnostics lack rationality in energy control and timing. Some signals have excessively high energy or overlap with production processing periods, easily interfering with normal production processes and affecting frame processing quality. Signal acquisition often uses fixed sampling frequencies, unable to be flexibly adjusted according to changes in the diagnostic window, resulting in inconsistent data lengths and reduced reliability under different operating conditions, making subsequent parameter analysis difficult. In addition, health status assessments often rely on single-dimensional indicators, failing to comprehensively reflect the collaborative working status between the equipment's electrical system and mechanical transmission. Early warning information often lacks accuracy and operability, and lacks effective compensation control mechanisms. When equipment performance degradation occurs, it cannot promptly offset its impact on processing quality, easily leading to various quality defects and severely restricting the stable and efficient operation of the frame production line. Summary of the Invention
[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a predictive maintenance and health management system for critical workstation equipment in a vehicle frame production line. The system addresses the problem that existing equipment maintenance technologies, as mentioned in the background, cannot adapt to the dynamic production cycle of the production line, and that diagnostic accuracy decreases when the diagnostic window time is compressed.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a predictive maintenance and health management system for key workstation equipment in a vehicle frame production line, comprising: a mode arbitration module: acquiring the diagnostic window time for equipment diagnosis under the current production cycle time, comparing the diagnostic window time with a preset critical diagnostic window time, generating cycle pressure parameters to characterize the degree of diagnostic time limitation and dividing the working modes according to the degree of compression of the diagnostic window time relative to the critical diagnostic window time;
[0006] Excitation generation module: Under the constraint of the cycle pressure parameter, a low-energy excitation signal is generated for equipment health diagnosis. The diagnosis window time is divided into a first time period for quickly establishing the excitation response and a second time period for observing and analyzing the dynamic response of the equipment. A periodic excitation signal is generated in the first time period and a frequency-modulated excitation signal is generated in the second time period.
[0007] Signal acquisition and preprocessing module: Acquires response current signal and position signal at an adaptive sampling frequency, and resamples to a uniform standard data length through linear interpolation, and outputs a normalized signal;
[0008] Parameter calculation module: Calculates equipment health parameters related to diagnostic time constraints based on the normalized signal. The equipment health parameters include: a first health parameter characterizing the equipment's ability to maintain an effective response band under diagnostic time constraints, and a second health parameter characterizing the degree of change in the dynamic consistency between the equipment's drive response and mechanical displacement under diagnostic time constraints.
[0009] Compensation control and health decision module: Generates equipment health index based on the first health parameter and the second health parameter, outputs maintenance warning level according to the health index, and triggers compensation control action at the same time.
[0010] The technical effects and advantages of this invention are as follows:
[0011] 1. This invention accurately identifies diagnostic time constraints under different production cycles, and combines the segmented low-energy excitation design of the excitation generation module with the adaptive sampling strategy of the signal acquisition module. Even in scenarios where diagnostic time is compressed, it can still ensure effective data acquisition and analysis accuracy, breaking through the dependence of traditional diagnostic methods on fixed diagnostic time and achieving flexible adaptation to the dynamic production rhythm of the chassis production line.
[0012] 2. This invention ensures that the diagnostic process will not interfere with the normal processing technology of the equipment and the quality of the chassis products by strictly controlling the energy level and output timing of the excitation signal. At the same time, by using linear interpolation resampling and data validity verification mechanisms, the collected data under different working conditions are unified to a standard dimension, which effectively solves the problems of inconsistent data length and low reliability in traditional technologies, and lays a high-quality data foundation for accurate assessment of equipment status.
[0013] 3. This invention characterizes the frequency band maintenance capability and electromechanical dynamic consistency of the equipment's electrical system through dual health parameters. Combined with the sub-mode health index calculation method, it achieves a comprehensive and accurate assessment of the equipment status. Furthermore, through the four-level maintenance warning level classification and targeted maintenance suggestion output, operators can quickly respond to signs of equipment degradation. At the same time, with the help of the coordinated control of feedforward compensation and feedback gain adjustment, it actively offsets the impact of degradation on processing quality, and continuously ensures the precision and stability of key chassis processes before maintenance.
[0014] 4. By capturing early performance degradation signals of equipment in advance, this invention transforms the traditional passive maintenance mode into proactive predictive maintenance. This not only effectively avoids production capacity losses caused by sudden equipment failures, but also optimizes maintenance resource allocation through targeted maintenance suggestions, reduces ineffective maintenance costs, and comprehensively improves the operational stability and overall production efficiency of the chassis production line. Attached Figure Description
[0015] Figure 1 This is a schematic diagram of the overall system structure of the present invention;
[0016] Figure 2 This is a schematic diagram of the workflow of the mode arbitration module of the present invention;
[0017] Figure 3 This is a schematic diagram of the excitation signal acquisition process of the present invention;
[0018] Figure 4 This is a schematic diagram of the process for obtaining health parameters according to the present invention;
[0019] Figure 5 This is a schematic diagram of the workflow of the compensation control and health decision-making module of the present invention. Detailed Implementation
[0020] 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.
[0021] As attached Figures 1 to 5 The illustrated predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line includes:
[0022] Mode Arbitration Module: Obtains the diagnostic window time for equipment diagnosis under the current production cycle time, compares the diagnostic window time with the preset critical diagnostic window time, generates cycle pressure parameters to characterize the degree of diagnostic time limitation based on the degree of compression of the diagnostic window time relative to the critical diagnostic window time, and divides the working modes.
[0023] It should be noted that the diagnostic window time for equipment diagnostics under the current production cycle time is obtained as follows:
[0024] At the entrance and exit of key workstations such as the side panel welding station and the chassis connection bolt tightening station, laser beam sensors with a response time of ≤1ms are deployed to detect the real-time position status of the chassis workpieces. When the laser beam sensor at the exit detects that the previous chassis workpiece has completely left the station and triggers a falling edge signal, the system starts the diagnostic window timing program. When the laser beam sensor at the entrance detects that the next chassis workpiece has entered the station and triggers a rising edge signal, the timing program stops. The time interval between the two signal triggers is the diagnostic window time T that can be used for equipment diagnosis under the current production cycle. w .
[0025] It should be further explained that the critical diagnostic window time T c As the core threshold for determining the strength of diagnostic time constraints, its calibration must strictly adhere to the dual criteria of frequency resolution and parameter estimation accuracy of the chassis production line equipment. Initial calibration is conducted using a combination of offline experiments and theoretical calculations. Taking a six-axis welding robot as an example, the key characteristic frequencies of the equipment, such as the resonant frequency of the welding gun arm, are first determined through spectrum analysis. Then, based on the frequency resolution requirement of Δf≤5Hz, the minimum sampling duration to meet the frequency resolution requirement is calculated. Simultaneously, based on the parameter estimation accuracy requirement, i.e., the estimated variance... ,in To estimate the variance of parameters under a long diagnostic window, the relationship between variance and sampling duration was fitted through multiple repeatable experiments. Finally, the minimum time that simultaneously satisfies two criteria was determined, which is the T value corresponding to this device. c value.
[0026] Set T c The system employs a dynamic update mechanism with a default update cycle of 3 months. When equipment modifications occur on the chassis production line, such as changing the welding torch model, adjusting the tightening shaft specifications, or altering the workpiece material (e.g., adjusting the thickness of the chassis steel), the system automatically triggers a T-wave update. c Recalibration process; operators can start the calibration program through the host computer interface, automatically collect dynamic operating data of the equipment, and recalculate T to meet the dual criteria. c The value is updated and overridden to ensure that the threshold always adapts to changes in the actual operating conditions of the production line.
[0027] It should be further explained that the cycle pressure parameter is obtained in the following way:
[0028] First, the collected T w The timing data is filtered to remove outliers caused by sensor interference. A single measurement exceeding ±20% of the average of the most recent 10 valid measurements is considered abnormal and replaced by the average of the previous three valid measurements. After preprocessing, the data is processed using the formula... The clock pressure parameter, also known as the window compression ratio N, is calculated, where the max function ensures that N ≥ 1; when T w ≥T c When N=1, it indicates that there is no compression constraint on the current diagnosis time; when T w <T c When N>1, the numerical value is positively correlated with the degree of compression in diagnosis time; for example, T w =100ms and T c When T = 150ms, N = 1.5, indicating a 50% compression constraint on the current diagnosis time; w When the time is 200ms, N=1, indicating that there is sufficient time for diagnosis.
[0029] Based on the N value, two working modes are defined: when N=1, it is determined to be a relaxed beat mode, in which traditional frequency domain analysis methods can meet the accuracy requirements of equipment health diagnosis, and the module outputs a relaxed mode identifier and the N=1 parameter to subsequent modules; when N>1, it is determined to be a tense beat mode, in which traditional diagnostic methods cannot meet the accuracy requirements, and an adaptive diagnostic process needs to be initiated, and the module synchronously outputs a tense mode identifier and the real-time calculated N value to subsequent modules.
[0030] Excitation generation module: Under the constraint of the cycle pressure parameter, a low-energy excitation signal is generated for equipment health diagnosis. The diagnosis window time is divided into a first time period for quickly establishing the excitation response and a second time period for observing and analyzing the dynamic response of the equipment. A periodic excitation signal is generated in the first time period and a frequency-modulated excitation signal is generated in the second time period.
[0031] It should be specifically noted that the low-energy excitation signal for equipment health diagnosis is generated under the constraint of the aforementioned cycle pressure parameter, and the specific process is as follows:
[0032] Real-time diagnostic window time T transmitted by the receiver mode arbitration module w With the cycle pressure parameter N, and according to a fixed ratio, T w The time period is divided into two functionally distinct time segments; the first time segment is the rapid setup segment, with a duration of T. r =T w / 4, its main function is to quickly establish a stable excitation response for the equipment, avoiding response distortion caused by sudden changes in the excitation signal; the second time period is the observation and analysis period, with a duration set to T. s =3T w / 4 serves as the primary observation interval for the equipment's dynamic response, used to collect effective signal data that reflects the equipment's health status.
[0033] The generation logic of the periodic excitation signal in the first time period:
[0034] Within the rapid setup segment, generate the periodic excitation current command I. d (t), whose mathematical expression is I d (t) = A·[1-exp(-t / τ)]·sin(2π·f0·t), where 0≤t <T r In this expression, parameter A is the amplitude of the excitation signal, specifically calculated as A = 0.05 × I. rated (I) rated The rated current of the equipment drive motor (imported into the system from the equipment's factory parameters) is used to strictly ensure the low-energy characteristics of the excitation signal and avoid interference with the normal operation of the equipment; the parameter τ is the settling time constant, whose value is related to the cycle pressure parameter N, and the specific calculation formula is τ=T r / (3N), the larger the N value, the higher the degree of diagnostic window time compression; the smaller the τ value, the faster the excitation signal amplitude rises, and the faster it can reach a stable amplitude in a shorter time, adapting to the ultra-short T in tense beat modes. r Duration; parameter f0 is the starting frequency, which is set to 10Hz by default. The selection of this frequency value is based on the low-frequency characteristic frequency band of key equipment in the chassis production line, which can effectively stimulate the basic vibration response of the equipment transmission chain.
[0035] The logic for generating the frequency modulation excitation signal in the second time period:
[0036] During the observation and analysis segment, a frequency modulation excitation current command I is generated. d (t), whose mathematical expression is I d (t) = A·sin(2πf0(tT) r )+πR(tT r ) 2 ), where T r ≤t <T w In this expression, the parameter R is the adaptive sweep rate, and its value is linearly related to the cycle pressure parameter N. Specifically, it is calculated as R = R b ·N,R b The base sweep rate is set to 50Hz / s by default. When N>1 is in the intense beat mode, the R value increases with N, which can achieve a higher sweep rate within a limited T. sCovering a wider frequency range within the duration ensures that the frequency sweep process includes the device's key characteristic frequencies; when N=1 is in the relaxed tick mode, the R value equals R0. b Excitation is completed at a conventional frequency sweep rate, ensuring that the frequency domain resolution of the signal meets the requirements of traditional analysis.
[0037] It should be further explained that dual constraints are used to ensure that the excitation signal does not interfere with the production process. First, there is an amplitude constraint, where the excitation signal amplitude A is strictly limited to within 5% of the equipment's rated current. Experiments have verified that this amplitude level will not affect the welding robot's welding torch attitude control or the torque output accuracy of the servo tightening shaft. Second, there is a timing constraint, where the generation and output of the excitation signal are strictly limited to the diagnostic window time T. w The interval between the departure of the previous frame workpiece from the workstation and the entry of the next frame workpiece into the workstation is completely avoided during the normal processing time of the equipment, so as to prevent the excitation process from affecting the process quality of frame welding or bolt tightening.
[0038] The excitation current command is finally converted into an analog signal and output to the drive control unit of the device, which drives the motor to generate corresponding micro-amplitude vibration excitation. At the same time, the module transmits the parameter information of the excitation signal to the signal acquisition and preprocessing module.
[0039] Signal acquisition and preprocessing module: Acquires response current signal and position signal at an adaptive sampling frequency, and resamples to a uniform standard data length through linear interpolation, and outputs a normalized signal;
[0040] It should be noted that the response current signal and position signal are acquired using an adaptive sampling frequency, and the specific process is as follows:
[0041] Determination of adaptive sampling frequency: The clock pressure parameter N transmitted by the receiver mode arbitration module and the timing parameters of the excitation signal transmitted by the excitation generation module are determined based on a preset reference sampling frequency F. b and the device hardware maximum sampling frequency F max Determine the adaptive sampling frequency F under the current operating conditions. s The specific calculation formula is F s =min(F b ·N 2 ,F max ); Reference sampling frequency F b The default setting is 20kHz, and the device's maximum hardware sampling frequency F max The default setting is 200kHz (adjustable according to actual hardware parameters). This frequency range can cover the effective detection range of the welding robot's welding torch arm resonance frequency and the torsional vibration frequency of the servo tightening shaft drive chain; when N=1 is in the relaxed beat mode, F s =F bSignal acquisition is performed at a conventional sampling frequency; when N>1 is in a tense beat mode, F s It increases with the square of N (not exceeding F). max The increase in sampling frequency can be achieved within a limited T w To acquire more sampling data points within a given timeframe, thus compensating for the insufficient data volume caused by the compression of the diagnostic window time.
[0042] Synchronous acquisition of dual-channel response signals: based on a determined adaptive sampling frequency F s Simultaneously acquire two response signals directly related to the health status of the equipment. The first signal is the actual feedback current signal I from the equipment drive unit. f (t), the second path is the actual position signal P(t) of the motor shaft; the actual feedback current signal I f (t) is obtained through a current sensor integrated in the equipment drive controller, which can directly reflect the load changes and response characteristics of the equipment's electrical system; the actual position signal P(t) of the motor shaft is obtained through a motor encoder, which can accurately characterize the motion state and displacement response of the equipment's mechanical transmission chain.
[0043] The acquisition process of the two signals is strictly synchronized with the excitation signal output process of the excitation generation module. The sampling trigger signal and the excitation signal start signal are generated from the same source, ensuring that the acquired response signal and excitation signal are perfectly aligned on the time axis, avoiding signal characteristic distortion caused by timing misalignment. The signal acquisition time range is strictly limited to the diagnostic window time T. w Within this range, from the moment the excitation signal is activated to the moment it is deactivated, no signal data is collected during the normal processing period of the equipment, thus eliminating interference from production conditions with the diagnostic signals.
[0044] It should be further explained that the normalized signal is output by resampling to a uniform standard data length through linear interpolation. The specific process is as follows:
[0045] The method for calculating the actual acquired data length is as follows: After signal acquisition is completed, first calculate the actual acquired data length L. w The specific calculation formula is L w =F s ·T w This parameter represents the total number of sampled data points acquired under the current adaptive sampling frequency and diagnostic window time conditions, and its value is related to F. s and T w Both showed a positive correlation; under the relaxed tempo mode, T w Longer and F s At a normal level, L w It can meet the basic data requirements for parameter calculation; in the fast-paced mode, T w shorten but Fs Significantly improved, through F s With N 2 The associated design ensures L w It remains at a level similar to the relaxed tempo pattern.
[0046] The standard data length is set based on: the preset standard data length L. s As a unified benchmark for data normalization processing, the specific calculation formula is L. s =F b ·T c This parameter represents the number of standard sampling data points under the conditions of the reference sampling frequency and the critical diagnostic window time. Its value remains fixed and is not affected by the current operating condition T. w The influence of N; standard data length L s This setting allows for the unified mapping of actual data collected under different beat modes and diagnostic window times to the same data length dimension, thus solving the problem of T... w The problem of inconsistent data lengths caused by changes.
[0047] Linear interpolation resampling: A linear interpolation method is used to resample the actual acquired feedback current signal I. f (t) and position signal P(t) are resampled to a standard data length L s On the constructed virtual time axis T, the mapping relationship between the virtual time axis T and the actual time axis t is T = t · (L s / L w The specific steps of the resampling process are as follows: First, determine the number of target data points after resampling based on the length of the virtual time axis T. Then, based on the time series and signal amplitude of the actual acquired data, calculate the signal amplitude corresponding to each virtual time point through linear interpolation, and finally generate the normalized current signal I. n (T) and normalized position signal P n (T); The choice of linear interpolation method balances computational efficiency and signal fidelity, and can complete the standardization conversion of data length without introducing additional high-frequency noise.
[0048] Data validity verification mechanism: After resampling, the normalized current signal I... n (T) and normalized position signal P n (T) Perform validity verification to remove abnormal data points caused by sensor interference or momentary equipment malfunctions. The specific verification rule is to calculate the historical normal fluctuation range of the signal amplitude; when the amplitude of a data point exceeds the historical average by ±30%, it is determined to be an abnormal data point, and the average of the two adjacent valid data points is used to replace it. After validity verification, the integrity of the signal also needs to be checked to ensure that the number of data points in the normalized signal matches the standard data length L.s Complete consistency is ensured to avoid data loss due to interpolation calculation errors.
[0049] Parameter calculation module: Calculates equipment health parameters related to diagnostic time constraints based on the normalized signal. The equipment health parameters include: a first health parameter characterizing the equipment's ability to maintain an effective response band under diagnostic time constraints, and a second health parameter characterizing the degree of change in the dynamic consistency between the equipment's drive response and mechanical displacement under diagnostic time constraints.
[0050] It should be noted that the first health parameter is obtained in the following way:
[0051] Spectrum Analysis and Boundary Frequency Extraction: For Normalized Current Signal I n (T) Perform a Fast Fourier Transform (FFT) to convert it into a frequency domain signal to extract the response characteristics of the equipment's electrical system; through spectral amplitude analysis, determine the two boundary frequencies corresponding to the signal amplitude attenuating to 3dB from the peak value, i.e., the lower limit boundary frequency f. L With upper limit boundary frequency f H The frequency range defined by the two is the effective response frequency band of the equipment under the current excitation, which directly reflects the dynamic response stability of the electrical system.
[0052] Effective bandwidth calculation: Based on the extracted boundary frequencies, the effective bandwidth B is calculated using the formula B = f H -f L This parameter characterizes the frequency range in which the device can stably respond to excitation signals. The wider the range, the stronger the adaptability of the device's electrical system to different frequency excitations. Conversely, a narrower range indicates that the electrical system may have performance degradation (such as coil aging or slow drive response).
[0053] The final calculation of the first health parameter Z: The bandwidth is corrected by combining the beat pressure parameter N to obtain the final first health parameter Z, which is calculated as Z = B·N. -1 / 2 Introducing N -1 / 2 This is because under the tense clock mode (N>1), the diagnostic time compression will lead to a decrease in the accuracy of frequency domain analysis. This index correction can compensate for the impact of time constraints on frequency band evaluation and ensure that the physical meaning of Z is consistent under different N values. The larger the Z value, the stronger the ability of the equipment to maintain an effective frequency band under the current clock pressure. The smaller the Z value, the more obvious the performance degradation of the equipment's electrical system and the higher the sensitivity to clock pressure.
[0054] It should be noted that the second health parameter is obtained in the following way:
[0055] Instantaneous torque estimation: Based on the motor electromagnetic torque formula, it is obtained from the normalized current signal I. n(T) Estimate the instantaneous output torque τ(T) of the equipment drive unit, calculated as τ(T) = K t ·I rated ·I n (T); where K t The torque constant of the motor (dimension N·m / A) directly reflects the conversion relationship between current and torque, ensuring the accuracy of the estimation results; the instantaneous torque τ(T) characterizes the real-time state of the equipment's drive response, and the position signal P... n (T) represents the real-time state of mechanical displacement, and the joint analysis of the two can accurately capture the matching degree between the drive and the mechanical transmission.
[0056] Sliding window length setting: based on standard data length L s With the beat pressure parameter N, and the sliding window length W, the calculation formula is W=L s / (5·N); The sliding window design is used for local analysis of the correlation between torque and position. The larger the N value (the tighter the cycle time), the shorter the window length, which can achieve denser local correlation detection in a limited standard data sequence, making up for the problem of insufficient global analysis accuracy caused by diagnostic time compression.
[0057] Cross-correlation coefficient calculation: With a set sliding window length W, slide the normalized signal sequence window by window to calculate the instantaneous torque τ(T) and the normalized position signal P. n The cross-correlation coefficient of (T) is used to extract the global maximum cross-correlation coefficient ρ. max The cross-correlation coefficient ranges from [-1, 1], ρ max The closer to 1, the better the dynamic consistency between the drive response and the mechanical displacement; the closer to 0 or a negative value, the worse the coordination between the two, and the mechanical transmission chain may have degradation problems such as increased clearance and wear.
[0058] The final calculation of the second health parameter S: The cross-correlation coefficient results are corrected by combining the beat pressure parameter N to obtain the second health parameter S, calculated as follows: Tension mode (N>1): S=(1-ρ max )·(N-1), loose mode (N=1): S=1-ρ max Among them, (1-ρ max The quantitative dynamic consistency deviation is represented by (N-1), which characterizes the amplification effect of the stress cycle pressure on the deviation. The sub-mode design can accurately reflect the degradation sensitivity of dynamic consistency under different cycle constraints. The larger the S value, the worse the dynamic consistency of the equipment drive and mechanical transmission is, and the more significant the influence of cycle pressure is, which corresponds to a higher risk of frame processing quality (such as insufficient weld penetration and excessive bolt clamping force).
[0059] It should be further explained that after calculating the two types of health parameters, validity verification is carried out to eliminate abnormal parameter values caused by signal noise and calculation errors, ensuring the reliability of the output results; the verification rules are set based on historical data of the device's health status: the preset normal range of Z [Z min Z max The normal range of values for S is [0, S]. th ], where Z min Z max S represents the boundary value of the bandgap maintenance capability under healthy conditions of the device. th The critical thresholds for dynamic consistency deviation are calibrated through offline experiments and support periodic updates. When the calculated Z or S exceeds the corresponding normal range, the module determines it as an abnormal parameter and automatically calls the normalized signal to perform a secondary calculation. If the secondary calculation result is still abnormal, the parameter abnormal state is marked and synchronously fed back to the signal acquisition and preprocessing module to trigger the signal re-acquisition process and ensure the validity of the output parameters.
[0060] Compensation control and health decision module: Generates equipment health index based on the first health parameter and the second health parameter, outputs maintenance warning level according to the health index, and triggers compensation control action at the same time.
[0061] It should be noted that the device health index is generated based on the first and second health parameters, and the specific process is as follows:
[0062] A sub-mode algorithm is used to generate the health index H, ensuring that the physical meaning of H is consistent and the values are comparable under both relaxed and tense beat modes. The specific calculation method is as follows:
[0063] Tense Beat Mode (N>1): Based on a weighted calculation of the first health parameter Z and the second health parameter S, the formula is as follows: Z0 and S0 are the baseline values of Z and S under the equipment's healthy state, respectively, and can be updated with the equipment maintenance cycle; w1 and w2 are weighting coefficients, satisfying w1+w2=1, with default settings of w1=0.4 and w2=0.6, which can be adjusted according to the importance of the workstation—welding workstations emphasize mechanical transmission consistency, so w2 can be appropriately increased; tightening workstations emphasize electrical response stability, so w1 can be appropriately increased; ε=10 -6 Z is a dimensionless minimum value used to avoid the risk of division by zero when S=0; in the formula, Z / Z0 represents the health of the electrical system, and S0 / max(S,ε) represents the health of the mechanical transmission. The weighted sum of the two yields the comprehensive health status. The value of H ranges from [0,1]. The closer it is to 1, the better the health status of the equipment. The closer it is to 0, the more serious the performance degradation.
[0064] Relaxed beat mode (N=1): H is calculated using the traditional frequency domain parameter mapping method; firstly, the main resonant frequency F of the device is extracted through traditional frequency domain analysis. r And the damping ratio D, and then based on the preset mapping function H=a·(F r / F r0 ) + b·(D0 / D) is calculated, where F r0 D0 is the reference value of the main resonant frequency and damping ratio under the health state of the equipment, and a and b are weighting coefficients (satisfying a+b=1) to ensure that the value range and judgment criteria of H under the relaxed mode are completely consistent with those under the tense mode, so as to achieve cross-mode health state comparability.
[0065] It should be further explained that, based on whether the values of Z and S exceed preset thresholds, corresponding compensation control actions are triggered. By dynamically adjusting the equipment control parameters, the impact of performance degradation on processing quality is offset, and the frame production accuracy is maintained. The specific actions are as follows:
[0066] Feedforward compensation (for scenarios where S exceeds the threshold): When S > S th (S) th When the dynamic consistency deviation threshold is reached (and is consistent with the verification threshold of the parameter calculation module), dynamic feedforward compensation torque control is activated. The compensation torque calculation formula is T. f =K s ·S·V p ·(N-1) / T c K s The feedforward gain coefficient (N·m·s) 2 )
[0067] Calibration was performed through offline experiments (default value 0.02 N·m·s). 2 ), V p T represents the current commanded speed of the device (unit: rad / s). c The critical diagnostic window time reference value (unit: seconds); compensation torque T f The real-time injection equipment drive controller dynamically adjusts according to the movement speed of the welding torch or tightening shaft to offset dynamic consistency deviations caused by mechanical transmission clearance and wear, ensuring that the weld penetration and bolt clamping force meet the standards.
[0068] Feedback gain adjustment (for scenarios where Z is below a threshold): When Z <Z min (Z) min When the minimum threshold of the first health parameter is consistent with the threshold verified by the parameter calculation module, the proportional gain K of the device position loop is dynamically adjusted. p Adjust the logic to K p =K p0 · (Z / Z) min )·∣V c | / V c0K p0 V is the baseline proportional gain under healthy conditions. c For real-time instruction speed, V c0 The rated command speed is adjusted to enhance the electrical system's response sensitivity to excitation signals, compensate for bandwidth contraction caused by coil aging and slow drive response, and maintain the stability of the equipment's electrical dynamic response.
[0069] Compensation action coordination logic: When Z and S exceed the threshold at the same time, feedforward compensation and feedback gain adjustment are executed synchronously, and their priority is higher than the original control parameters of the equipment. The compensation parameters are updated in real time with the changes of Z, S and N. When Z and S return to the normal range, the module automatically stops the compensation action and restores the equipment's baseline control parameters to avoid operating condition fluctuations caused by over-compensation.
[0070] It should be further explained that, based on the value range of the unified health index H, four levels of maintenance warning are divided. Targeted maintenance suggestions are generated by combining the specific degradation types of Z and S to ensure that the warning information is accurate and actionable. The specific classification criteria are as follows:
[0071] Green alert (H≥0.8): The equipment is in good health with no signs of performance degradation; a green alert icon is output, indicating "the equipment is in normal condition, maintain routine inspection", no compensation control action is triggered, only health parameters and H value are recorded for trend analysis.
[0072] Blue alert (0.6≤H<0.8): The equipment has experienced slight performance degradation, which has not affected the processing quality; the module outputs a blue alert icon, indicating "the equipment has slight degradation, strengthen parameter monitoring", and simultaneously records the changing trends of Z and S, without triggering compensation actions.
[0073] Yellow warning (0.4≤H<0.6): Equipment performance has deteriorated significantly and is approaching the quality impact threshold; the module outputs a yellow warning sign, triggering the corresponding compensation control action, and simultaneously providing targeted maintenance suggestions—when S-dominant degradation occurs, the message "Check the reducer and harmonic reducer, and investigate transmission clearance" is displayed; when Z-dominant degradation occurs, the message "Inspect the motor coil and drive unit, and investigate abnormal electrical response" is displayed, and it is recommended to complete the maintenance in the next scheduled maintenance window.
[0074] Red Alert (H<0.4): Equipment performance has severely degraded, posing a quality risk; the module outputs a red alert indicator, triggering compensation control actions and simultaneously sending an audible and visual alarm to the operator via the host computer, prompting "Equipment requires emergency maintenance to prevent the quality defect from expanding," and recommending immediate shutdown for repair. At the same time, the processing quality data of the corresponding workstation is locked for subsequent traceability.
[0075] 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.
[0076] 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 predictive maintenance and health management system for critical workstation equipment in a vehicle frame production line, characterized in that, include: Mode Arbitration Module: Obtains the diagnostic window time for equipment diagnosis under the current production cycle time, compares the diagnostic window time with the preset critical diagnostic window time, generates cycle pressure parameters to characterize the degree of diagnostic time limitation based on the degree of compression of the diagnostic window time relative to the critical diagnostic window time, and divides the working modes. Excitation generation module: Under the constraint of the cycle pressure parameter, a low-energy excitation signal is generated for equipment health diagnosis. The diagnosis window time is divided into a first time period for quickly establishing the excitation response and a second time period for observing and analyzing the dynamic response of the equipment. A periodic excitation signal is generated in the first time period and a frequency-modulated excitation signal is generated in the second time period. Signal acquisition and preprocessing module: Acquires response current signal and position signal at an adaptive sampling frequency, and resamples to a uniform standard data length through linear interpolation, and outputs a normalized signal; Parameter calculation module: Calculates equipment health parameters related to diagnostic time constraints based on the normalized signal. The equipment health parameters include: a first health parameter characterizing the equipment's ability to maintain an effective response band under diagnostic time constraints, and a second health parameter characterizing the degree of change in the dynamic consistency between the equipment's drive response and mechanical displacement under diagnostic time constraints. Compensation control and health decision module: Generates equipment health index based on the first health parameter and the second health parameter, outputs maintenance warning level according to the health index, and triggers compensation control action at the same time.
2. The predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line according to claim 1, characterized in that: The diagnostic window time is obtained as follows: Laser beam sensors with a response time of no more than one millisecond are deployed at the entrance and exit of the key workstations of the chassis production line. The timing starts when the sensor at the exit detects that the previous chassis workpiece has completely left and triggers a falling edge signal. The timing stops when the sensor at the entrance detects that the next chassis workpiece has entered and triggers a rising edge signal. The time interval between the two signal triggers is the diagnostic window time.
3. The predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line according to claim 1, characterized in that: The working modes are divided as follows: First, the timing data of the diagnostic window time is filtered, and then the beat pressure parameter is determined by the ratio of the critical diagnostic window time to the diagnostic window time, and the ratio is not less than one; the working modes are divided based on the beat pressure parameter, and when the ratio is equal to one, it is determined to be a relaxed beat mode, and when the ratio is greater than one, it is determined to be a tense beat mode.
4. The predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line according to claim 1, characterized in that: The periodic excitation signal is acquired as follows: the duration of the first time period is determined to be one-quarter of the diagnostic window time; the signal amplitude is set to five percent of the rated current of the equipment drive motor; the signal establishment time constant is determined based on the duration of the first time period and the cycle pressure parameter, the larger the cycle pressure parameter, the smaller the establishment time constant; the signal start frequency is set to a fixed value, which is determined based on the low-frequency characteristic frequency band of the key equipment of the chassis production line; and the periodic excitation signal is generated based on the above parameters.
5. The predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line according to claim 1, characterized in that: The frequency modulation excitation signal is obtained as follows: the duration of the second time period is determined to be three-quarters of the diagnostic window time; the reference sweep rate is set to a fixed value; the actual sweep rate is determined based on the reference sweep rate and the clock pressure parameter. The larger the clock pressure parameter, the higher the actual sweep rate. In the relaxed clock mode, the actual sweep rate is equal to the reference sweep rate; the frequency modulation excitation signal is generated based on the determined actual sweep rate.
6. The predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line according to claim 1, characterized in that: The first health parameter is obtained as follows: the normalized current signal is subjected to fast Fourier transform to obtain the frequency domain signal, the lower boundary frequency and the upper boundary frequency corresponding to the signal amplitude attenuation to the peak value of 3 dB are extracted, the difference between the two boundary frequencies is calculated to obtain the effective bandwidth, and then the effective bandwidth is corrected by combining the beat pressure parameter. The corrected value is the first health parameter.
7. The predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line according to claim 1, characterized in that: The second health parameter is obtained as follows: the instantaneous output torque of the equipment drive unit is estimated based on the motor torque constant, the rated current of the equipment drive motor, and the normalized current signal; the sliding window length is set based on the standard data length and the cycle pressure parameter, and the cross-correlation coefficient between the instantaneous output torque and the normalized position signal is calculated window by window, and the global maximum cross-correlation coefficient is extracted; in the tense cycle mode, the second health parameter is obtained by multiplying the maximum cross-correlation coefficient by the cycle pressure parameter correction value; in the relaxed cycle mode, the second health parameter is obtained by the difference between the maximum cross-correlation coefficient and the numerical value.
8. The predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line according to claim 1, characterized in that: The equipment health index is obtained as follows: In the tense beat mode, it is obtained by weighted summation of the ratio of the first health parameter to the corresponding benchmark value and the ratio of the second health parameter benchmark value to the second health parameter; in the relaxed beat mode, it is obtained by weighted summation of the ratio of the equipment's main resonant frequency to the corresponding benchmark value and the ratio of the equipment's damping ratio benchmark value to the actual damping ratio; the health index ranges from zero to one, and the closer it is to one, the better the equipment's health status.
9. The predictive maintenance and health management system for critical workstation equipment on a vehicle frame production line according to claim 1, characterized in that: The compensation control actions are as follows: When the second health parameter exceeds the preset threshold, dynamic feedforward compensation torque control is activated. The compensation torque is determined based on the feedforward gain coefficient, the second health parameter, the current command movement speed of the device, and the critical diagnostic window time reference value. When the first health parameter is lower than the preset threshold, the proportional gain of the device position loop is dynamically adjusted. The adjusted proportional gain is related to the reference proportional gain in the healthy state, the ratio of the first health parameter to the corresponding lowest threshold, and the ratio of the real-time command speed to the rated command speed. When the first health parameter and the second health parameter both exceed their corresponding thresholds, feedforward compensation and feedback gain adjustment are executed synchronously, with a higher priority than the original control parameters of the device. Once the parameters return to the normal range, the compensation action is automatically stopped and the baseline control parameters are restored.