A method for dynamic calculation and state evaluation of power of a rotary drive unit

Abstract

The present application relates to mechanical transmission energy representation technical field, disclose a kind of dynamic measurement and state evaluation method of rotary drive unit power, comprising: synchronous acquisition prime mover execution end phase voltage, phase current and mechanical transmission path double-end pulse signal, according to the deviation of pulse count value determines transient torsional angle offset, and utilize offset change gradient to determine the relaxation parameter of internal physical hysteresis characteristic, establish dynamics model, by relaxation parameter correction torsional stiffness, subtract inertia and after correction component deconstruction transient mechanical transmission power, calculate power residual and time-domain integration, generate physical fatigue evaluation index, the present application is by relaxation parameter correction torsional stiffness, realize state change and mechanical transient coupling, exclude inertia interference and restore energy transmission essence, scheme will microscopic damage risk be converted into quantitative index, realize the technical leap from performance monitoring to fatigue evaluation.

Description

Technical Field

[0001] This invention relates to a method for dynamic calculation and state assessment of the power of a rotary drive unit, belonging to the field of mechanical transmission energy characterization technology. Background Technology

[0002] Power monitoring of rotating machinery drive systems is a fundamental means of ensuring the healthy operation of large industrial equipment. In high-value shaft systems such as wind turbines, large rolling mills, and precision CNC machine tools, stator current inversion or direct torque sensing methods are commonly used. These methods offer measurement reliability under steady-state operating conditions, and by acquiring drive-end current and speed signals, they can obtain basic parameters reflecting the system's energy transfer. However, as industrial drive systems evolve towards higher power density and extremely accelerated operating conditions, rotating shaft systems exhibit nonlinear mechanical characteristics when subjected to sudden load changes. Existing methods assume the shaft system is an absolutely rigid body or an ideal linear elastic body, and assume that the drive-end input and load-end response are within a certain range. While maintaining synchronization in the time domain, in actual physical processes, the microscopic gaps at the shaft connection interface and the physical dislocations within the material cause a time delay in energy conduction. This physical time delay causes amplitude distortion at the peak of the measured instantaneous power waveform. The application of electromagnetic torque, material dislocations, and mechanical interface gaps together constitute the causal chain of energy transfer. When the drive unit faces a sudden load change, energy is consumed to overcome the rotor's rotational inertia before being transferred to the load end through the shaft system. The shaft material has physical resistance when subjected to transient alternating stress, which triggers transient structural relaxation. This phenomenon causes a dynamic shift in the real-time torsional stiffness of the shaft system, manifested as a phase lag in torque transmission.

[0003] Besides limitations in hardware deployment, software analysis methods also have shortcomings in handling transient responses. For example, Chinese invention patent CN119688210A discloses a method for monitoring torsional resonance in the shaft system of a high-voltage frequency converter drive load. It relies on encoder pulse signals to perform discrete Fourier transform frequency analysis, using the torsional angle amplitude at a specific frequency as the monitoring benchmark. The scheme's logic anchors to the frequency domain characterization of stationary random signals. Due to the periodic accumulation of signals, when faced with extremely accelerated conditions or millisecond-level impact loads, the internal physical hysteresis and structural relaxation of the shaft system exhibit non-stationary transient characteristics, which are difficult to detect using discrete Fourier transform. Transform analysis suffers from window function effects and response delays, making it impossible to capture instantaneous phase shifts caused by energy conduction chains and difficult to isolate the interference of rotational inertia absorption on the actual work curve. Industry improvement approaches often focus on increasing sampling frequency or optimizing signal filtering. These methods can only reduce electromagnetic noise and cannot address the physical delays caused by energy absorption by rotational inertia and shaft flexibility. Increasing filter bandwidth will introduce high-frequency oscillation noise, causing the measurement results to deviate from the physical reality. Existing measurement architectures cannot characterize transient structural relaxation phenomena under high dynamic conditions and cannot effectively separate work from parasitic losses.

[0004] Therefore, how to accurately separate the electromagnetic power generated by the rotary drive unit under dynamic operating conditions, the rotor inertial power, and the actual mechanical power transmitted by the shaft system, and achieve high-fidelity calculation and state assessment of the micromechanical response of the transmission shaft system, has become the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A method for dynamic calculation and state evaluation of the power of a rotary drive unit, comprising the following steps: Step 101: Synchronously acquire phase voltage data, phase current data and pulse sequence signal of the dual-end encoder of the mechanical transmission path at the input end of the prime mover using the same sampling clock signal, and perform sub-microsecond time stamping on the three types of data using the sampling clock signal to align with the time axis of the physical signal. Step 102: Determine the transient torsional angle offset of the mechanical transmission path based on the count value deviation between the two-ended pulse sequence signals, calculate the first derivative of the transient torsional angle offset with time to obtain the gradient of angular displacement change, and thus determine the relaxation parameter characterizing the molecular-level hysteresis characteristics generated inside the mechanical transmission path under alternating stress. Step 103: Establish a dynamic balance model that includes electromagnetic torque, rotor inertial torque and load transmission torque. Introduce the relaxation parameter as a feedback factor into the transfer function of the dynamic balance model to dynamically correct the boundary constraint conditions of torsional stiffness. By subtracting the rotor inertial torque and the stiffness correction amount generated by the relaxation parameter from the electromagnetic torque, the actual transient mechanical transmission power is obtained by deconstruction. Step 104: Subtract the transient mechanical transmission power from the transient input electromagnetic power calculated based on phase voltage data and phase current data to obtain the transient residual power that reflects the internal friction loss and micro-deformation work of the transmission system. Perform time-domain integration based on the symplectic algorithm on the transient residual power within a preset time window to generate a physical fatigue assessment index characterizing the cumulative amount of non-productive energy dissipation.

[0006] Preferably, step 102 includes the following sub-steps: step 1021, performing high-order difference operations on the pulse sequence signal to determine the transient angular acceleration at the end of the mechanical transmission path; step 1022, based on the dynamic load excitation source generated by the transient angular acceleration, combined with the response phase of the transient torsional angle offset, using the structural dynamic response curve fitting to obtain relaxation parameters, so as to quantitatively characterize the physical response delay caused by lattice dislocations and interface gaps in the mechanical transmission path.

[0007] Preferably, in step 103, the damping distribution coefficient in the dynamic balance model is adjusted in real time using the relaxation parameter to cancel the nonlinear high-frequency oscillation signal caused by the sudden change in load, thereby maintaining the waveform phase of the transient mechanical power transmission and the fundamental phase of the phase current data to achieve logical consistency after decoupling calculation.

[0008] Preferably, step 104 includes acquiring the amplitude change rate of phase current data. When the amplitude change rate exceeds a preset dynamic threshold, it is determined that the prime mover has entered a transient load condition, and the preset time window is automatically shortened from 500ms to 10ms to capture the physical fatigue assessment index under high-frequency impact load.

[0009] Preferably, the logic for generating the physical fatigue assessment index is as follows: the integral value of the transient residual power is compared online with the benchmark lossless state threshold stored in the preset database. When the drift of the integral value in a single operating cycle exceeds 15%, an early warning signal indicating that microscopic damage or mechanical performance degradation has occurred inside the mechanical transmission path is output.

[0010] Preferably, before step 102 is carried out, the method further includes acquiring real-time thermal field data of the prime mover through a temperature sensor, and using the real-time thermal field data to perform thermodynamic deviation calibration on the initial mapping relationship of the relaxation parameters, so as to compensate for the nonlinear drift of the elastic modulus of the mechanical conduction path material caused by thermal expansion.

[0011] Preferably, in step 101, a 10MHz co-source sampling pulse is used to synchronously mark the phase voltage data, phase current data, and pulse sequence signal. By comparing the time difference of the rising edges of the two pulse signals, the torsional deformation can be accurately identified, providing a high-resolution data source for the extraction of relaxation parameters.

[0012] Preferably, the physical fatigue assessment index generated in step 104 is input into a preset life prediction model. By calculating the first and second derivatives of the physical fatigue assessment index within a continuous operating sampling period, the abrupt inflection point of fatigue evolution is identified, thereby outputting the remaining effective operating time prediction data of the mechanical transmission path.

[0013] Preferably, it also includes a feedback control strategy based on the physical fatigue assessment index to adjust the upper limit of the input power of the prime mover, and by changing the drive frequency of the inverter to avoid the mechanical resonance range indicated by the relaxation parameter, the transient residual power returns to the preset stable range, thereby achieving active intervention and protection.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. In dynamic measurement and state assessment, a correction mechanism reflecting physical response hysteresis is constructed by using the time difference between the excitation origin of the stator current change rate and the response endpoint of the driven load end angular acceleration. This mechanism can identify the structural relaxation phenomenon generated by the rotary drive shaft system under transient load impact, thereby solving the torque measurement phase deviation problem caused by material nonlinearity and assembly clearance at the mechanical model level, and ensuring that the output transient power curve is consistent with the actual mechanical work behavior on the time axis.

[0015] 2. By combining the current characteristics reflecting the thermal load state with the relaxation parameters reflecting the transient mechanical response hysteresis, dynamic adjustment of the torsional stiffness of the rotating shaft system is achieved. This method couples macroscopic thermodynamic changes with microsecond-level structural mechanical transients across dimensions, enabling the measurement model to evolve in real time with the intensity of the load and operating conditions. As a result, it can maintain high accuracy in mechanical characteristic characterization even under heavy load impact or continuous temperature rise, avoiding the fundamental measurement errors caused by the insufficient environmental adaptability of traditional fixed parameter models.

[0016] 3. Establish a dynamic balance architecture that includes electromagnetic torque, rotor inertial torque, and shaft transmission torque. By accurately separating each energy component, the parasitic interference of the system's inherent inertia on the transient output power is eliminated. This processing logic extends the measurement perspective from a single end sensor to the complete energy transmission link, enabling the final generated power waveform to strip away false inertial components and restore the physical essence of the mechanical energy transmission process. Attached Figure Description

[0017] Figure 1 This is a flowchart of the dynamic power calculation and fatigue assessment of the rotary drive unit of the present invention; Figure 2 This is a logic block diagram for extracting relaxation parameters for response delay identification in this invention.

[0018] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A method for dynamic calculation and state assessment of the power of a rotary drive unit includes the following steps: Step 101: Synchronously acquire phase voltage data, phase current data and pulse sequence signal of the dual-end encoder of the mechanical transmission path at the input end of the prime mover using the same sampling clock signal, and perform sub-microsecond time stamping on the three types of data using the sampling clock signal to align with the time axis of the physical signal. Step 102: Determine the transient torsional angle offset of the mechanical transmission path based on the count value deviation between the two-ended pulse sequence signals, calculate the first derivative of the transient torsional angle offset with time to obtain the gradient of angular displacement change, and thus determine the relaxation parameter characterizing the molecular-level hysteresis characteristics generated inside the mechanical transmission path under alternating stress. Step 103: Establish a dynamic balance model that includes electromagnetic torque, rotor inertial torque and load transmission torque. Introduce the relaxation parameter as a feedback factor into the transfer function of the dynamic balance model to dynamically correct the boundary constraint conditions of torsional stiffness. By subtracting the rotor inertial torque and the stiffness correction amount generated by the relaxation parameter from the electromagnetic torque, the actual transient mechanical transmission power is obtained by deconstruction. Step 104: Subtract the transient mechanical transmission power from the transient input electromagnetic power calculated based on phase voltage data and phase current data to obtain the transient residual power that reflects the internal friction loss and micro-deformation work of the transmission system. Perform time-domain integration based on the symplectic algorithm on the transient residual power within a preset time window to generate a physical fatigue assessment index characterizing the cumulative amount of non-productive energy dissipation.

[0021] Preferably, step 102 includes the following sub-steps: step 1021, performing high-order difference operations on the pulse sequence signal to determine the transient angular acceleration at the end of the mechanical transmission path; step 1022, based on the dynamic load excitation source generated by the transient angular acceleration, combined with the response phase of the transient torsional angle offset, using the structural dynamic response curve fitting to obtain relaxation parameters, so as to quantitatively characterize the physical response delay caused by lattice dislocations and interface gaps in the mechanical transmission path.

[0022] Preferably, in step 103, the damping distribution coefficient in the dynamic balance model is adjusted in real time using the relaxation parameter to cancel the nonlinear high-frequency oscillation signal caused by the sudden change in load, thereby maintaining the waveform phase of the transient mechanical power transmission and the fundamental phase of the phase current data to achieve logical consistency after decoupling calculation.

[0023] Preferably, step 104 includes acquiring the amplitude change rate of phase current data. When the amplitude change rate exceeds a preset dynamic threshold, it is determined that the prime mover has entered a transient load condition, and the preset time window is automatically shortened from 500ms to 10ms to capture the physical fatigue assessment index under high-frequency impact load.

[0024] Preferably, the logic for generating the physical fatigue assessment index is as follows: the integral value of the transient residual power is compared online with the benchmark lossless state threshold stored in the preset database. When the drift of the integral value in a single operating cycle exceeds 15%, an early warning signal indicating that microscopic damage or mechanical performance degradation has occurred inside the mechanical transmission path is output.

[0025] Preferably, before step 102 is carried out, the method further includes acquiring real-time thermal field data of the prime mover through a temperature sensor, and using the real-time thermal field data to perform thermodynamic deviation calibration on the initial mapping relationship of the relaxation parameters, so as to compensate for the nonlinear drift of the elastic modulus of the mechanical conduction path material caused by thermal expansion.

[0026] Preferably, in step 101, a 10MHz co-source sampling pulse is used to synchronously mark the phase voltage data, phase current data, and pulse sequence signal. By comparing the time difference of the rising edges of the two pulse signals, the torsional deformation can be accurately identified, providing a high-resolution data source for the extraction of relaxation parameters.

[0027] Preferably, the physical fatigue assessment index generated in step 104 is input into a preset life prediction model. By calculating the first and second derivatives of the physical fatigue assessment index within a continuous operating sampling period, the abrupt inflection point of fatigue evolution is identified, thereby outputting the remaining effective operating time prediction data of the mechanical transmission path.

[0028] Preferably, it also includes a feedback control strategy based on the physical fatigue assessment index to adjust the upper limit of the input power of the prime mover, and by changing the drive frequency of the inverter to avoid the mechanical resonance range indicated by the relaxation parameter, the transient residual power returns to the preset stable range, thereby achieving active intervention and protection.

[0029] Example 1: In a scenario where the method of this invention is applied to implement high dynamic load monitoring of the main shaft drive system of a wind turbine generator, the main shaft drive system includes a wind turbine, a main shaft, a speed-increasing gearbox, and a doubly-fed induction generator. When the system faces gust impact conditions, the mechanical energy captured at the wind turbine end generates transient fluctuations. At this time, the electromagnetic torque at the generator stator end undergoes high-frequency adjustment, causing electromechanical coupling torsional vibration in the main shaft system. Due to the physical time delay and rotational inertia absorption effect of the shaft system when the gust enters, the output power curve calculated during the gust entry phase deviates in amplitude and phase from the peak load actually borne by the wind turbine end. The system utilizes... The phase voltage data, phase current data, and pulse sequence signals of the dual-ended encoder of the mechanical conduction path are synchronously acquired using a sampling clock signal from the same source. The phase voltage data, phase current data, and pulse sequence signals are then timestamped at the sub-microsecond level using the sampling clock signal to align with the time axis of the physical signals. The transient torsional angle offset of the mechanical conduction path is determined based on the count value deviation between the dual-ended pulse sequence signals. The first derivative of the transient torsional angle offset with time is calculated to obtain the gradient of angular displacement change, thereby determining the relaxation parameters characterizing the molecular-level hysteresis characteristics generated within the mechanical conduction path under alternating stress. Specifically, this is achieved by extracting the peak timestamps of the stator active power component from the phase current data. and the angular acceleration peak timestamps calculated from the pulse sequence signals. Calculate the time delay between the two. The ratio of this delay to the inherent transmission delay of the shaft system is determined as the relaxation parameter. In this physical extraction procedure, although molecular-level slip and dislocation events within a single lattice occur at the micro-nano scale and nanosecond time dimension, under the macroscopic excitation of high-intensity alternating loads, the microscopic inelastic deformation of massive grain boundaries throughout the entire transmission shaft path inevitably produces a cross-scale statistical accumulation. This cumulative effect of massive microscopic damage manifests in the system's macroscopic dynamic response as an overall lag in the phase of the physical transmission waveform. Therefore, this scheme utilizes the macroscopic time delay extracted by latching a co-originating high-frequency clock. Through this scale-crossing mapping, the system can use macroscopically measurable response delay to accurately and non-invasively characterize the dissipation level of microstructures deep within the shaft system under a mechanical equivalent causal mechanism. When the rotary drive unit is subjected to high-dynamic alternating loads, the electromagnetic excitation of the prime mover first acts on the surface material of the shaft system and undergoes acoustic shearing. The wave propagates from the solid medium inside the drive shaft to the driven load end. In this energy conduction chain, the microscopic grain boundaries, dislocation defects, and lattice gaps within the material together constitute the microscopic inelastic dissipation core. When the macroscopic shear stress wavefront passes through these microscopic defects, it induces local lattice slip and inelastic relaxation at the micro- and nanoscale, converting some kinetic energy into micro-regional heat loss. This leads to the attenuation of the acoustic shear wavefront energy and causes nonlinear rheological changes in wave velocity at the microscopic level. These hundreds of millions of microscopic lattice dislocations and interface relaxation events are continuously propagated and accumulated at the macroscopic topological scale of the drive shaft. Ultimately, they are manifested at the macroscopic level and equivalently represented by the overall phase lag of the driven load end motion response on the time axis at the millisecond level. Thus, the macroscopic time delay is used to complete the accurate cross-scale characterization of the internal microscopic mechanical dissipation level.

[0030] Based on this, a dynamic balance model is established that encompasses electromagnetic torque, rotor inertial torque, and load-transmitted torque, and the relaxation parameters are included. This feedback factor is introduced into the transfer function of the dynamic equilibrium model to dynamically correct the boundary constraints of torsional stiffness, specifically through the formula... Calculate the real-time torsional stiffness, where, The preset initial torsional stiffness, The corrected dynamic torsional stiffness is obtained by subtracting the rotor inertial torque from the electromagnetic torque and by adjusting the relaxation parameters. The resulting stiffness correction is used to deconstruct and obtain the actual transient mechanical transmission power. The transient input electromagnetic power calculated based on phase voltage and phase current data Subtract transient mechanical transmission power This yields the transient residual power that reflects the internal frictional losses and microscopic deformation work of the transmission system. For transient residual power Within a preset 10ms time window, time-domain integration based on a symplectic algorithm is performed to generate a physical fatigue assessment index characterizing the cumulative amount of unproductive energy dissipation. This index is used to quantify the fatigue loss state of the spindle under alternating stress impact, through relaxation parameters. By modifying the torsional stiffness, the system can achieve dynamic compensation of shaft transmission power and quantitative assessment of structural health status using electromagnetic and kinematic fingerprint signals without adding strain gauge sensors. The processor defines two basic physical components of transient residual power, sets the extracted transient residual torque component as the canonical momentum parameter of the system state space, and sets the synchronously measured dynamic residual torsional deformation as the conjugate canonical coordinate parameter. In this way, a microscopic Hamiltonian function that can map the inelastic dissipative manifold of the system is constructed in the local coordinate system. The algorithm control module calls the built-in first-order Euler-Maria-Marsing difference scheme and performs high-frequency numerical iteration updates of the set of canonical Hamiltonian differential equations in microsecond-level steps within a set 10ms time window. Thus, the energy dissipation integral is derived while maintaining the conservation of the geometric area of ​​the manifold phase space, thereby avoiding the risk of pseudo-dissipation caused by the accumulation of conventional Euler method calculations.

[0031] Example 2: Testing was conducted on a dynamic performance verification platform for a high-power rotary drive unit. This platform includes a 500kW permanent magnet synchronous motor drive system with active load simulation capabilities and a dynamic torque sensor with a 10kHz sampling rate. The test data originated from the sensor sequence of the physical test platform, where the encoder resolution was [missing information]. Each pulse revolution supports sub-microsecond angular displacement acquisition; the current transformer bandwidth covers 0Hz to 100kHz with a measurement accuracy better than 0.2%; the timestamp synchronization deviation of all physical signal acquisition modules is controlled within 50ns; and the sampling period... The settings are controlled by the first-order torsional natural frequency of the measured shaft system and the real-time task scheduling load of the processor. To capture the microsecond-level relaxation response within the material and suppress signal aliasing caused by electromagnetic harmonic interference, The time was set to 50 μs, a value determined in an engineering example under a decision rule balancing signal fidelity and computational overhead. Gaussian white noise with a signal-to-noise ratio of 20 dB was superimposed on the acquired raw phase current data. A control group and the sample group of this invention were compared and tested. The control group used the conventional electric power inversion method with a constant stiffness assumption, while the sample group of this invention applied complete relaxation parameters. The corrected logic states that when a step excitation with an amplitude of 80% of the rated torque is applied to the drive end, the power curve output by the control group exhibits a peak deviation and phase misalignment of 35.82% within the 20ms time window at the onset of the impact, failing to reflect the energy absorption process caused by the relaxation of the shaft structure. However, the sample group of this invention extracts the first timestamp... With the second timestamp Determine the time delay amount The relaxation time is 1.324ms, from which the relaxation parameters are obtained. The value is 1.156, using the formula Real-time reduction of initial torsional stiffness, resulting in transient mechanical power transmission. The overlap between the curve and the measured value from the torque sensor is better than 95.74%, and the relative error is reduced to 4.256%. This confirms that using relaxation parameters to correct torsional stiffness can suppress phase interference caused by shaft hysteresis and eliminate the influence of measurement noise. In determining the final relaxation parameters, to address the glitch interference caused by background environmental noise directly quotienting a single timestamp, the system introduces an exponential structural dynamic decay curve equation for high-noise-resistance data fitting. The basic equation for this fitting is set as follows: The amplitude of the excitation source The value is taken from the fundamental peak value of the current during the current servo drive switching transient. As the characteristic decay time constant of the structural dynamic response, the processor substitutes the measured multi-sampling point response time series data into the equation and uses it as the judgment benchmark point group. It performs gradient optimization iterative calculation through the built-in Gauss-Newton nonlinear least squares method. After the residual sum of squares converges and stabilizes, the hidden target characteristic time scalar is solved and separated. The comparison result between the time scalar and the benchmark time delay is then confirmed as the relaxation parameter of the final dynamic output, realizing the smooth extraction of physical time delay under high disturbance conditions.

[0032] A load intensity gradient control system was introduced, and the above calculation process was repeated under operating conditions of 20%, 50%, 80%, 110%, and 130% of the rated torque. The response law of the technical indicators under different load levels was observed. When the load intensity was in the normal operating range of 20% to 80%, the relaxation parameter increased slowly in the range of 1.023 to 1.156, and the physical fatigue evaluation index remained stable below 15.62, indicating that the system has excellent mechanical linearity within this window. After the load intensity exceeded 110%, the performance inflection point was observed. At this time, the relaxation parameter rapidly increased to 1.482 and the gradient of change increased. Correspondingly, the physical fatigue evaluation index jumped sharply from 22.45 to 92.83, showing a nonlinear degradation effect caused by the surge of lattice dislocations inside the material. This is consistent with the physical law of plastic deformation of the spindle material under high load. By exceeding the range... Continuous monitoring of the control group under 130% load conditions revealed an exponential increase in the integral rate of the physical fatigue assessment index, confirming that the relaxation parameter range defined in this invention is the optimal monitoring window for accurately characterizing the evolution of the shaft system from the elastic stability region to the fatigue damage region. Based on the above experimental data, the method of this invention achieves high-fidelity calculation of the power of the rotating drive unit under industrial conditions full of random disturbances. Through the synergistic effect of the relaxation parameter and the dynamic equilibrium model, the transient structural relaxation physical quantity, which is difficult to measure directly, is transformed into a quantifiable correction operator, solving the technical problem of rapid accuracy degradation of traditional measurement methods under dynamic variable load scenarios. The experimental conclusions obtained are completely consistent with the beneficial effects claimed by this invention. By generating a physically meaningful fatigue assessment index through the time-domain integration of transient residual power, a technical leap is achieved from single operational performance monitoring to essential assessment of the structural health of the transmission shaft system.

[0033] Example 3: In a scenario where the method of this invention is applied to implement real-time mechanical power monitoring of a high-dynamic servo drive unit in a CNC machine tool, when the servo drive unit executes a high-speed reciprocating cutting command, the motor end faces a sinusoidal alternating load impact with a frequency of 50Hz and an amplitude reaching 120% of the rated torque. Due to the length-to-diameter ratio characteristics of the servo axis and the nonlinear contact stiffness of the coupling, a microsecond-level mechanical response lag occurs during the acceleration phase. At this time, conventional sensors are limited by bandwidth and cannot separate the transient deformation work caused by the molecular-level hysteresis within the shaft material, making it difficult for the system to accurately assess the microscopic fatigue damage state of the servo axis during the cutting process. To solve the above challenges, the system defines the starting point for physical signal acquisition and selects the first-order torsional natural frequency of the servo axis. As a benchmark, a processor with floating-point arithmetic capabilities is used to receive phase voltage data, phase current data, and pulse sequence signals generated by a dual-ended encoder from the input terminal of the prime mover via a synchronous serial interface. The encoder's single-turn pulse count is no less than 1,048,576 to support angular displacement acquisition, and the synchronous sampling clock frequency of the signal acquisition module is set to 2MHz. After acquiring the data, the processor uses the sampling clock signal to perform sub-microsecond timestamp marking on the phase voltage data, phase current data, and pulse sequence signals to ensure alignment of the electromagnetic and mechanical signals in the time domain. Then, the relaxation parameter determination procedure is initiated. The processor performs a fifth-order differential operation on the pulse sequence signals to suppress sampling noise and extract the transient angular acceleration at the load terminal. The characteristic peak moment of the transient angular acceleration in the time-domain waveform is extracted as the second timestamp. Simultaneously, the stator active power component is extracted by performing coordinate transformation on the phase current data, and its peak time is obtained as the first timestamp. Determine the time delay amount Then, the inherent transmission delay, determined by the shaft geometry and the speed of sound, pre-stored in memory, is utilized. The relaxation parameters are calculated. This parameter is used to quantify the degree of physical hysteresis within the transmission path under stress impact. Before outputting this relaxation parameter and defining it as a microscopic hysteresis index, the processor pre-executes a macroscopic time delay forced stripping program. From the calculated time delay Δt, it directly calls the internal lookup table to subtract the inverter's inherent electromagnetic dead zone delay time specified by the driver hardware manual, and accurately subtracts the macroscopic torsional elastic response constant and coupling mechanical clearance passage time obtained from the system's static calibration. After this cascaded time-domain subtraction stripping, the remaining net time delay term is limited to the internal friction effect caused only by material grain boundary slip and interface dislocation, completely completing the dimensional downsampling transformation from macroscopic mixed time difference to microscopic single lattice hysteresis.

[0034] To address the challenges of high-frequency feature submersion and sub-microsecond signal jitter induced by electromagnetic interference in the operating environment, a preprocessor extracting various feature timestamps employs a hardware low-pass filter with a cutoff frequency set to 1MHz to perform broadband noise filtering on the original synchronous sampling data. Based on the Kelvin-Voigt viscoelastic material dynamics principle, the dynamic equivalent torsional stiffness of metal and composite transmission shaft systems under transient alternating stress loading exhibits an inverse proportional relationship with the time of structural relaxation within the material. Further exploring the closed-loop mapping of this principle at the engineering calculation level, the Kelvin-Voigt model decouples the mechanical transmission behavior of solid transmission shafts into an ideal Hooke spring assembly and a first-order Newtonian viscous fluid damping assembly. In the macroscopic parallel superposition architecture, when the equipment is under high-frequency step excitation conditions, the viscous damping effect rapidly expands and dominates the response, causing irreversible amplitude attenuation of the real part of the stiffness vector of the macroscopic complex plane (i.e., the effective torsional stiffness of the transmission system). The ratio of this stiffness attenuation is mathematically equivalent to the loss tangent function value characterizing the rheological hysteresis of the material. This technology system directly uses the relaxation parameter, which represents the macroscopic time delay ratio and is measured in real time, as the numerical compensation term of the above-mentioned loss tangent factor. It cleverly avoids the complex decoupling of the high-order partial differential mechanical equations of the distributed parameters and directly realizes the extremely simplified dimensionality reduction transformation from complex stiffness vector rheology to a simple algebraic reduction division formula on the underlying computation side.

[0035] Based on this, the system constructs a dynamic balance model encompassing electromagnetic torque, rotor inertial torque, and load-transmitted torque, and incorporates relaxation parameters. As a feedback factor introduced into the dynamic correction function of torsional stiffness, the processor, according to the formula... Calculate the real-time torsional stiffness, where, The initial torsional stiffness is obtained from static calibration. To correct the dynamic torsional stiffness and ensure the accuracy of power decomposition, the system extracts generator losses from the total input power using electromagnetic balance equations. The processor determines the system impedance polynomial based on pre-executed driver no-load frequency conversion operation experiments. The total apparent power obtained by multiplying phase voltage and phase current data is then subtracted in real-time from the input power by the current stator operating frequency and the stator core eddy current losses and winding heating components directly reflected by the transient phase current amplitude, thus yielding the transient input electromagnetic power. Simultaneously, the rotor inertial power is calculated using the angular acceleration obtained from the angular displacement signal and the rotor moment of inertia. According to dynamic torsional stiffness The product of the angular displacement deviations at both ends yields the true transient mechanical transmission power. And calculate the transient residual power. For transient residual power The processing avoids the phenomenon of averaging and masking of high-frequency transient impact peak values ​​caused by excessively large integration step size, based on the Nyquist sampling criterion. When the processor detects that the first-order change rate of phase current amplitude with time exceeds 10% of the rated operating parameters, the control module triggers the algorithm engine to reduce the time integration window span from the steady-state default of 500ms to the transient-specific 10ms.

[0036] Based on the fundamental energy conservation principle of Hamiltonian systems, the integral of the work done by the true deformation of the dynamic shaft system needs to eliminate the pseudo-numerical dissipation introduced by the conventional Euler difference. The system selects a sliding time window of 20ms, the length of which is determined by the sum of the servo control cycle and the torsional vibration period of the shaft system. The processor uses a symplectic algorithm to calculate the transient residual power. By performing structurally preserved numerical integration, a physical fatigue assessment index is accumulated. According to Miner's linear fatigue cumulative damage law, the remaining mechanical life of the metal shaft system depends on the total inelastic plastic work done under repeated alternating load impacts. In the prerequisite step of applying this algorithm to the scalar characteristic of transient residual power, in order to overcome the parameter dimension barrier between one-dimensional algebraic scalars and high-dimensional symplectic geometric algorithms, the algorithm engine performs a conjugate decomposition operation beforehand, directly extracting the angular velocity time series component contained in the scalar power value, and transforming its remaining term into a generalized force (i.e., residual excitation torque), which, together with the corresponding generalized coordinate deviation (residual displacement), is mapped to a complete two-dimensional Hamiltonian loss. By deconstructing and transforming the state parameters of the discrete space system, the original simple one-dimensional power time integral summation operation is isomorphically replaced in physical and mathematical logic by solving the local geometric phase volume generated by the envelope evolution of the set of conjugate vectors in the two-dimensional phase space plane. This establishes the absolute logical closed loop of the flow of physical energy to the structure-preserving integral system. In this isomorphic replacement process, in order to satisfy the dissipation-free premise of symplectic geometric volume preservation, the algorithm engine establishes an augmented Hamiltonian state space containing a virtual potential energy term. The irreversible thermodynamic dissipation power caused by the stripped residual displacement is forcibly defined as a virtual potential energy conversion process of microscopic deformation by an auxiliary conservative force field in the augmented phase space. Through this determined mapping path, the non-productive dissipation of the original system is strictly confined within the internal conservative exchange network of the augmented system, thus fully satisfying the core mathematical prerequisite of the symplectic algorithm requiring the conservation of the overall Hamiltonian, ensuring that the integral derivation does not diverge. Based on the original 2D state variables consisting of the canonical momentum fully supported by the residual torque and the conjugate canonical coordinates acting as the dynamic residual torsional deformation, the algorithm engine additionally allocates two independent augmented data dimensions through software registers, defined as virtual momentum components and virtual displacement components, respectively. Within this 4D state space, the processor extracts the irreversible thermodynamics... Dissipated power is forcibly converted into the rate of change of virtual momentum over time, while the auxiliary conservative force field is constructed by multiplying the current residual torque by a preset dimensionless decay factor. This auxiliary conservative force field performs a first-order differential update on the virtual displacement, thereby converting the dissipated frictional loss work into virtual conservative potential energy within the augmented data dimension in an algorithmic logic isomorphic manner. Through this 4-dimensional augmented matrix construction, the external unproductive dissipation is completely enclosed in the conservation exchange network of kinetic energy and virtual potential energy within the system, satisfying the premise of overall conservation of Hamiltonian and providing a stable and non-divergent computational closed loop for the second-order stepwise numerical integration of the symplectic algorithm.The processor generates a physical fatigue assessment index by integrating the time domain and substitutes it into a preset SN fatigue decay function expression specific to the drive shaft material to calculate and output the micro-damage degree for a single stress cycle. It then performs a weighted summation of various micro-damage degrees within the continuous cutting interval. In each continuous sampling cycle, the system performs time-domain differencing on the generated physical fatigue assessment index to calculate the first derivative of the current cycle relative to the previous cycle, and then performs further differencing on the first derivative to obtain the second derivative. When the algorithm engine detects that the first derivative is consistently greater than 0 and the second derivative abruptly changes from negative to positive, it determines that the system has passed the abrupt inflection point of fatigue evolution and entered the accelerated damage stage. At this point, the life prediction module... The system extracts the current cumulative value of the physical fatigue assessment index and calculates the difference between it and the maximum allowable dissipation power threshold, which is stored in non-volatile memory and determined by the fatigue limit of the shaft material. This difference yields the remaining dissipation capacity, which is then divided by the rate of change of the first derivative in the current cycle to quantitatively estimate the remaining effective operating time before the drive shaft completely fails. When the physical fatigue assessment index is detected to be monotonically increasing over three consecutive cutting cycles and the increase exceeds the preset fatigue warning threshold, the system outputs a quantitative warning of the risk of damage to the internal structure of the servo shaft material. This preset fatigue warning threshold is calibrated as the benchmark steady-state integral as a key criterion for determining the triggering of hardware protection logic. The 15% gain threshold for drift is not a deferred estimate based on conventional engineering experience, but rather rigorously derived from a series of measured datasets of cumulative fatigue overload and extreme destructive events conducted during the system's early development. Regression analysis of extensive shaft failure data revealed that when microcrystalline crack clusters within the transmission material initiate malignant propagation, causing micro-frictional heat generation to drastically increase inelastic dissipation energy beyond the 15% boundary condition of the reference surface, the overall torsional strength of the structure will instantly drop beyond the critical point of plastic instability. Therefore, setting the system hardware cutoff and early warning control at the 15% gain node most efficiently locks in the ultimate defense window before macroscopic mechanical collapse, especially under high-frequency alternating loads. In the data analysis after the working condition was completed, the method of this invention resolved the technical contradiction between shaft nonlinear hysteresis and high-precision measurement. By introducing a relaxation parameter to correct the torsional stiffness, the instantaneous error of the calculated mechanical transmission power curve in the dynamic impact section was reduced from 22.6% in the conventional method to less than 1.15%. This result confirms the reliability of the physical logic of characterizing microscopic hysteresis characteristics by the time delay between electromagnetic excitation and mechanical response. It not only eliminates spurious power fluctuations caused by shaft flexibility, but also achieves online quantitative characterization of the fatigue state of the servo shaft through the cumulative integration of transient residual power, completing the technological evolution from single real-time performance monitoring to essential assessment of the health status of mechanical structures.

[0037] Example 4: When the system faces the initial deployment phase of the transmission shaft parameter calibration, the mechanical transmission path is in a static locked state. The mechanical brake fixes the load end to a fully locked position. The excitation current is adjusted through the input of the prime mover to generate a stepped increasing static electromagnetic torque. The processor reads the pulse sequence signal generated by the dual-end encoder and analyzes the corresponding count value deviation to determine the static torsional angular displacement of the mechanical transmission path under different torque levels. By linearly fitting the applied electromagnetic torque value with the static torsional angular displacement, the initial torsional stiffness of the transmission shaft is determined based on the slope of the fitted curve. The residual angular displacement when the torque is zero is recorded and stored in non-volatile memory as the system zero-point offset to eliminate the zero-position error caused by sensor installation bias, and to ensure that the transient torsional angular offset calculated in the subsequent dynamic monitoring stage maps the shaft torsional deformation.

[0038] In the time delay measurement procedure when the transmission system is in a no-load uniform speed operation state, the prime mover is adjusted to the rated speed operation state and the load end is in a free disengagement state. At this time, there is no mechanical load torque in the mechanical transmission path. The processor collects the stator active component ripple peak time in the phase current data and the peak time in the double-ended encoder pulse sequence signal at a synchronous sampling frequency of 2MHz. The statistical average value of the collected sampling points is calculated within a steady-state observation period of up to 10s. The time deviation value between the electrical response and mechanical feedback of the system under no alternating stress interference is extracted and determined as the reference transfer delay for subsequent relaxation parameter calculation. In the pre-processing unit, a moving average filter is used to filter out the high-frequency torsional vibration components caused by bearing friction, ensuring... The value of the time delay is mapped to the geometry and topology of the axis system itself and the sound velocity characteristics of the material, providing a basis for utilizing the time delay during the dynamic monitoring phase. Identifying physical hysteresis characteristics provides a physical reference benchmark.

[0039] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for dynamic calculation and state assessment of the power of a rotary drive unit, characterized in that, Includes the following steps: Step 101: Synchronously acquire phase voltage data, phase current data and pulse sequence signal of the dual-end encoder of the mechanical transmission path at the input end of the prime mover using the same sampling clock signal, and perform sub-microsecond time stamping on the three types of data using the sampling clock signal to align with the time axis of the physical signal. Step 102: Determine the transient torsional angle offset of the mechanical transmission path based on the count value deviation between the two-ended pulse sequence signals, calculate the first derivative of the transient torsional angle offset with time to obtain the gradient of angular displacement change, and determine the relaxation parameter characterizing the molecular-level hysteresis characteristics generated inside the mechanical transmission path under alternating stress. Step 102 includes the following sub-steps: Step 1021, performing high-order difference operations on the pulse sequence signal to determine the transient angular acceleration at the end of the mechanical transmission path; Step 1022, based on the dynamic load excitation source generated by the transient angular acceleration, combined with the response phase of the transient torsional angle offset, using the structural dynamic response curve fitting to obtain relaxation parameters, so as to quantitatively characterize the physical response delay caused by lattice dislocations and interface gaps in the mechanical transmission path; Step 103: Establish a dynamic equilibrium model encompassing electromagnetic torque, rotor inertial torque, and load-transmitted torque. Introduce relaxation parameters as feedback factors into the transfer function of the dynamic equilibrium model to dynamically correct the boundary constraints of torsional stiffness. Specifically, this is achieved through the formula... Calculate the real-time torsional stiffness, where, The preset initial torsional stiffness, The corrected dynamic torsional stiffness. These are relaxation parameters, based on dynamic torsional stiffness. The product of the angular displacement deviations at both ends yields the true transient mechanical transmission power. ; Step 104: Subtract the transient mechanical transmission power from the transient input electromagnetic power calculated based on phase voltage data and phase current data to obtain the transient residual power that reflects the internal friction loss and micro-deformation work of the transmission system. Perform time-domain integration based on the symplectic algorithm on the transient residual power within a preset time window to generate a physical fatigue assessment index characterizing the cumulative amount of non-productive energy dissipation.

2. The method for dynamic calculation and state evaluation of the power of a rotary drive unit according to claim 1, characterized in that, In step 103, the damping distribution coefficient in the dynamic equilibrium model is adjusted in real time using the relaxation parameter to cancel the nonlinear high-frequency oscillation signal caused by sudden load changes, thereby maintaining the waveform phase of transient mechanical power transmission and the fundamental phase of phase current data to achieve logical consistency after decoupling calculation.

3. The method for dynamic calculation and state evaluation of the power of a rotary drive unit according to claim 1, characterized in that, Step 104 includes acquiring the amplitude change rate of phase current data. When the amplitude change rate exceeds the preset dynamic threshold, it is determined that the prime mover has entered a transient load condition, and the preset time window is automatically shortened from 500ms to 10ms to capture the physical fatigue assessment index under high-frequency impact load.

4. The method for dynamic calculation and state evaluation of the power of a rotary drive unit according to claim 1, characterized in that, The logic for generating the physical fatigue assessment index is as follows: the integral value of the transient residual power is compared online with the benchmark lossless state threshold stored in the preset database. When the drift of the integral value in a single operating cycle exceeds 15%, an early warning signal indicating that microscopic damage or mechanical performance degradation has occurred inside the mechanical transmission path is output.

5. The method for dynamic calculation and state evaluation of the power of a rotary drive unit according to claim 1, characterized in that, Before step 102 is carried out, the process also includes acquiring real-time thermal field data of the prime mover through a temperature sensor, and using the real-time thermal field data to perform thermodynamic deviation calibration on the initial mapping relationship of the relaxation parameters in order to compensate for the nonlinear drift of the elastic modulus of the mechanical conduction path material caused by thermal expansion.

6. The method for dynamic calculation and state evaluation of the power of a rotary drive unit according to claim 1, characterized in that, In step 101, a 10MHz homogeneous sampling pulse is used to synchronously mark the phase voltage data, phase current data, and pulse sequence signal. By comparing the time difference of the rising edges of the dual-path pulse signals, the torsional deformation is accurately identified, providing a high-resolution data source for the extraction of relaxation parameters.

7. The method for dynamic calculation and state evaluation of the power of a rotary drive unit according to claim 1, characterized in that, The physical fatigue assessment index generated in step 104 is input into the preset life prediction model. By calculating the first and second derivatives of the physical fatigue assessment index within the continuous operation sampling period, the abrupt inflection point of fatigue evolution is identified.

8. The method for dynamic calculation and state evaluation of the power of a rotary drive unit according to claim 1, characterized in that, It also includes a feedback control strategy based on the physical fatigue assessment index to adjust the upper limit of the input power of the prime mover, and to avoid the mechanical resonance range indicated by the relaxation parameter by changing the drive frequency of the inverter, so that the transient residual power returns to the preset stable range, thereby achieving active intervention and protection.

Images