A motor operation simulation fault diagnosis system

Abstract

This invention relates to the field of industrial automation control and electrical equipment condition monitoring technology, specifically a motor operation simulation fault diagnosis system, comprising: a data acquisition interface unit configured to connect to an analog-to-digital converter connected to the stator circuit of an explosion-proof motor, used to read unfiltered digital discrete voltage and current sequences at a sampling rate higher than a preset multiple of the inverter carrier frequency; a signal calculation and separation unit used to perform digital bandpass filtering on the voltage and current sequences to extract transient response data components containing high-frequency switching oscillation characteristics from the fundamental energy transmission signal; a model parameter inversion unit used to iteratively calculate the set of high-frequency impedance parameters on the stator side of the motor at the current moment; and a safety status assessment unit used to generate a safety assessment data package for the explosion-proof motor. This invention solves the problem of non-intrusive monitoring in explosion-proof environments, improving safety and compliance.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control and electrical equipment condition monitoring technology, specifically a motor operation simulation fault diagnosis system. Background Technology

[0002] Explosion-proof motors are critical power equipment in high-risk industrial environments. The stator insulation performance and the integrity of the mechanical air gap structure directly determine the safety level of the system. Fault monitoring usually involves a comprehensive analysis of the motor's electrical parameters and physical state. Existing technologies mainly rely on installing vibration or temperature sensors and analyzing steady-state current signals. However, implanting sensors inside the explosion-proof motor will compromise the sealing integrity of the explosion-proof enclosure, leading to complex safety certification and on-site construction challenges. Furthermore, the hardware filtering path of traditional frequency converter control systems often treats high-frequency switching oscillation signals carrying microstructural information of the motor as interference and filters them out, resulting in the loss of key transient characteristic data. At the same time, existing monitoring methods are unable to capture the weak physical parameter drifts in the early stages of insulation aging and air gap eccentricity, and lack a deep inversion mechanism based on digital twin models. Therefore, it is impossible to achieve accurate perception and early warning of the high-frequency impedance characteristics inside the motor without intrusion. Thus, a solution is urgently needed to address the problems existing in the current technology. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a motor operation simulation fault diagnosis system. Specifically, the technical solution of this invention includes:

[0004] The data acquisition interface unit is configured to connect to the analog-to-digital converter of the explosion-proof motor stator circuit, and is used to read the unfiltered digital discrete voltage and current sequences at a sampling rate that is higher than a preset multiple of the inverter carrier frequency.

[0005] The signal calculation and separation unit is used to perform digital bandpass filtering operations on voltage and current sequences to extract transient response data components containing high-frequency switching oscillation characteristics from the fundamental energy transmission signal.

[0006] The model parameter inversion unit is used to call the preset digital twin equivalent circuit model of the explosion-proof motor, take the transient response data components as the excitation input of the model, and iteratively calculate the set of high-frequency impedance parameters on the stator side of the motor at the current moment by minimizing the sum of squared residuals between the model output and the measured data.

[0007] The safety status assessment unit is used to map the parasitic capacitance values ​​in the high-frequency impedance parameter set to the insulation dielectric aging index, map the imbalance of the leakage inductance parameter to the air gap eccentricity index, and generate a safety assessment data package for the explosion-proof motor based on the insulation dielectric aging index and the air gap eccentricity index.

[0008] Optionally, the data acquisition interface unit performs data reading operations based on oversampling technology to capture non-stationary time series data output by the frequency converter. The sampling rate is set to be no less than ten times the carrier frequency of the frequency converter, and the data reading path avoids any hardware analog low-pass filter in order to retain high-frequency harmonic characteristics that can characterize the slight changes in the dielectric constant inside the motor.

[0009] Optionally, the signal calculation and separation unit includes:

[0010] The digital low-pass filter module is used to perform convolution operations on voltage and current sequences with a cutoff frequency lower than the upper limit of the fundamental frequency, generating a fundamental data stream for conventional control.

[0011] The digital bandpass filter module is used to construct a filter transfer function with a center frequency locked to the inverter carrier frequency and a bandwidth covering the third harmonic range. It filters voltage and current sequences and outputs transient response data components.

[0012] Optionally, the model parameter inversion unit performs the following computer-implemented steps when solving for the high-frequency impedance parameter set:

[0013] Identify the falling edge trigger time of the PWM pulse from the transient response data components;

[0014] Extract the time window data from the trigger moment to the moment when the oscillation decays to zero, and construct the target observation waveform vector;

[0015] Initialize the state of the second-order differential equation model containing resistance, inductance, and capacitance variables;

[0016] The gradient descent algorithm or particle swarm optimization algorithm is used to adjust the parameter values ​​in the second-order differential equation model until the Euclidean distance between the predicted waveform vector generated by the model and the target observed waveform vector is less than the preset convergence tolerance.

[0017] The parameter values ​​that meet the convergence condition are locked and output as a set of high-frequency impedance parameters.

[0018] Optionally, the security status assessment unit performs the following parameter mapping logic:

[0019] Extract the distributed capacitance to ground value from the high-frequency impedance parameter set, calculate the percentage change relative to the initial calibration value, and use it as the aging index of the insulating medium.

[0020] Extract the three-phase leakage inductance values ​​from the high-frequency impedance parameter set, and calculate the ratio of the standard deviation to the arithmetic mean of the three-phase leakage inductance values ​​as the air gap eccentricity index.

[0021] Optionally, the system also includes an active detection control unit for performing the following data quality enhancement logic:

[0022] The ratio of the signal power to the background noise power of the transient response data component is calculated to obtain the real-time signal-to-noise ratio value.

[0023] The real-time signal-to-noise ratio value is compared with the preset minimum confidence threshold;

[0024] If the real-time signal-to-noise ratio is less than the minimum confidence threshold, a high-frequency perturbation injection command is generated and sent to the inverter controller to request the superposition of a jitter signal of a specific frequency in the output voltage.

[0025] If the real-time signal-to-noise ratio is greater than or equal to the minimum confidence threshold, an instruction to maintain the current sampling mode is generated.

[0026] Optionally, the security status assessment unit performs the following closed-loop logic judgment when generating the security assessment data packet:

[0027] Retrieve the insulation failure threshold and rudder sweep risk threshold stored in the database;

[0028] If the aging index of the insulating medium is greater than the insulation failure threshold, the status bit of the safety assessment data packet will be marked as insulation critical, and the insulation prediction life data will be written.

[0029] If the air gap eccentricity index is greater than the sweep risk threshold, the status bit of the safety assessment data packet is marked as mechanical emergency, and the eccentricity orientation data is written.

[0030] If the insulation medium aging index is less than or equal to the insulation failure threshold, and the air gap eccentricity index is less than or equal to the rudder sweep risk threshold, the status bit of the safety assessment data packet is marked as operationally healthy, and written to the status log of the current calculation cycle.

[0031] Optionally, the system is configured as a non-invasive data processing terminal, which is deployed outside the safety boundary of the explosion-proof hazardous area. It analyzes the digital electrical signals transmitted by the data acquisition interface unit and reconstructs the state of the insulation medium and the geometric topology of the air gap inside the explosion-proof motor using a digital twin equivalent circuit model without implanting physical sensors inside the explosion-proof motor or damaging the sealing integrity of the explosion-proof enclosure.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. This invention solves the problem of non-invasive monitoring in explosion-proof environments, improving safety and compliance; it is equipped with a data processing terminal deployed outside the safety boundary of the hazardous area, which can remotely invert the internal state using a digital twin model without damaging the sealing integrity of the explosion-proof shell and without implanting physical sensors; this effectively solves the problems of complex safety certification and construction difficulties faced by explosion-proof motor monitoring, eliminates the safety hazards caused by installing sensors with the cover open, and reduces on-site implementation costs;

[0034] 2. This invention breaks through the limitations of hardware filtering and achieves complete capture of high-frequency micro-features; it adopts a high-frequency oversampling strategy that avoids hardware analog low-pass filters, and completely preserves the high-frequency transient oscillation components of the stator windings that are usually regarded as noise; combined with digital bandpass filtering technology, it accurately extracts feature data containing micro-structure information from the fundamental energy signal, overcomes the technical problem of loss of key fault features due to hardware filtering in traditional frequency conversion control systems, and lays a data foundation for accurate diagnosis.

[0035] 3. This invention realizes early warning of weak faults based on physical parameter inversion; it uses second-order differential equations to construct a digital twin equivalent circuit model, and minimizes waveform residuals through iterative algorithms to solve high-frequency impedance parameters; this method can map macroscopic electrical signals into microscopic physical parameters, keenly capture weak parameter drifts in the early stages of insulation dielectric aging and air gap eccentricity, and can achieve early warning in the early stages before insulation breakdown or rubbing accidents occur, with accuracy significantly better than traditional steady-state current analysis methods;

[0036] 4. This invention constructs an active detection and multi-dimensional decoupling diagnosis mechanism, enhancing system robustness; establishes parameter mapping logic based on the statistical characteristics of parasitic capacitance and leakage inductance, realizing quantitative decoupling of electrical insulation and mechanical air gap faults; combined with an active perturbation injection mechanism that is automatically triggered when the signal-to-noise ratio is low, it ensures that high-quality diagnostic data can be obtained under various operating conditions, and can provide fully closed-loop hierarchical early warning information including remaining life prediction and fault location, effectively guiding differentiated operation and maintenance. Attached Figure Description

[0037] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0038] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0040] Example 1:

[0041] Please see Figure 1A motor operation simulation fault diagnosis system, comprising:

[0042] The data acquisition interface unit is configured to connect to the analog-to-digital converter of the explosion-proof motor stator circuit, and is used to read the unfiltered digital discrete voltage and current sequences at a sampling rate that is higher than a preset multiple of the inverter carrier frequency.

[0043] The signal calculation and separation unit is used to perform digital bandpass filtering operations on voltage and current sequences to extract transient response data components containing high-frequency switching oscillation characteristics from the fundamental energy transmission signal.

[0044] The model parameter inversion unit is used to call the preset digital twin equivalent circuit model of the explosion-proof motor, take the transient response data components as the excitation input of the model, and iteratively calculate the set of high-frequency impedance parameters on the stator side of the motor at the current moment by minimizing the sum of squared residuals between the model output and the measured data.

[0045] The safety status assessment unit is used to map the parasitic capacitance values ​​in the high-frequency impedance parameter set to the insulation dielectric aging index, map the imbalance of the leakage inductance parameter to the air gap eccentricity index, and generate a safety assessment data package for the explosion-proof motor based on the insulation dielectric aging index and the air gap eccentricity index.

[0046] This embodiment provides a motor operation simulation fault diagnosis system; the core logic of the system is: instead of relying on traditional vibration or temperature physical sensors, it mines the high-frequency switching oscillation signal in the inverter power supply circuit, which is usually regarded as noise, and uses digital twin technology to invert the changes in the physical parameters inside the motor;

[0047] The data acquisition interface unit is the connection hub between the system and the physical world; in this embodiment, it is configured to be physically connected to the analog-to-digital converter (ADC) on the stator circuit side of the explosion-proof motor; its special feature is that it is configured to operate at a sampling rate that is much higher than the carrier frequency of the frequency converter and satisfies the high-frequency transient oscillation sampling theorem.

[0048] This unit directly reads the unfiltered digitally discretized voltage sequence from the ADC. and current sequence ;

[0049] Unlike conventional control systems, this path must bypass any hardware anti-aliasing filters in order to fully preserve the transient high-frequency components containing information about the distributed parameters of the motor stator windings.

[0050] The signal calculation and separation unit is used to extract microscopic feature signals from macroscopic energy transfer signals;

[0051] For the collected and Perform digital bandpass filtering operation;

[0052] From the fundamental energy transmission signal primarily used to drive the motor rotation, the transient response data component containing high-frequency switching oscillation characteristics is extracted. This component is a signal generated by the IGBT switching action of the frequency converter, reflecting the parasitic parameters of the motor stator winding, such as stray inductance, distributed capacitance, and damping oscillation characteristics.

[0053] The model parameter inversion unit is the core computing engine of this system, used to map data into physical parameters;

[0054] The unit has a pre-built digital twin equivalent circuit model of the explosion-proof motor; this model is a mathematical model based on second-order or higher-order differential equations, used to describe the high-frequency impedance characteristics of the motor in the megahertz MHz frequency band.

[0055] The unit will be stripped out As the excitation input to the model; through an optimization algorithm, the sum of squared residuals between the theoretical waveform output by the model and the actual waveform of the measured data is minimized. ;

[0056] Iteratively calculate the set of high-frequency impedance parameters on the stator side of the motor at the current moment. This parameter set It includes parameters such as resistance, inductance, and capacitance, which cannot be directly measured but are physically real.

[0057] The safety status assessment unit is used to translate physical parameters into health indicators that can be understood by operations and maintenance personnel;

[0058] Will The parasitic capacitance value is mapped to the aging index of the insulating medium; the physical basis is that the aging of insulating materials such as varnish and insulating paper will lead to a change in dielectric constant, thereby directly changing the parasitic capacitance value.

[0059] Will The imbalance of the leakage inductance parameter is mapped to the air gap eccentricity index; the physical basis is that the uneven eccentricity of the air gap will lead to uneven distribution of leakage flux in the three-phase winding, which in turn will lead to asymmetry in the leakage inductance parameter.

[0060] Based on the above two indices, a safety assessment data package for explosion-proof motors is generated for display or alarm on the host computer.

[0061] This embodiment achieves non-invasive and accurate diagnosis of the internal insulation and mechanical air gap states of explosion-proof motors by constructing a technical link of high-frequency sampling, signal separation, and model inversion. Compared with existing technologies, its significant advantages are: no need to install sensors inside the motor, avoiding the risk of damaging the sealing of the explosion-proof enclosure; and by utilizing transient oscillation signals that are extremely sensitive to high-frequency parasitic parameters, it can provide early warning in the early stages before insulation breakdown or rotor rubbing accidents occur, i.e., when physical parameters drift slightly, significantly improving the operational safety of equipment in high-risk environments.

[0062] Example 2:

[0063] The data acquisition interface unit performs data reading operations based on oversampling technology to capture non-stationary time series data output by the frequency converter. The sampling rate is set to be no less than ten times the carrier frequency of the frequency converter, and the data reading path avoids any hardware analog low-pass filter in order to retain the high-frequency harmonic characteristics that can characterize the slight changes in the dielectric constant inside the motor.

[0064] This embodiment specifies the sampling strategy of the data acquisition interface unit in order to solve the problem of the integrity of high-frequency feature capture;

[0065] The data acquisition interface unit performs data reading operations by capturing non-stationary time series data output by the frequency converter based on oversampling technology.

[0066] Sampling rate setting: In order to effectively capture the parasitic oscillation characteristics of stator windings at the megahertz / MHz level, the sampling rate is set to... Set to a frequency no lower than the expected highest-order parasitic oscillation frequency 2.5 to 5 times. Specifically, considering that the resonant frequency of the LC circuit formed by the stray inductance and distributed capacitance of the explosion-proof motor stator is usually distributed in the range of 100kHz to 3MHz, the data acquisition interface unit in this embodiment uses a high-speed analog-to-digital converter, and the sampling rate is set to 2.5 to 5 times. MSPS, ten million samples per second, to avoid spectral aliasing of high-frequency transient signals;

[0067] Hardware path design: The data reading path is designed to bypass any hardware analog low-pass filter; conventional frequency converters usually have a filter connected to the output to filter out high-frequency noise, but this embodiment deliberately bypasses the hardware to retain the high-frequency harmonic characteristics that can characterize the slight changes in the dielectric constant inside the motor.

[0068] By employing a strategy of tenfold oversampling and no analog filtering, this embodiment ensures that the high-frequency decaying oscillation waveform carrying information about the motor insulation medium, excited by IGBT switching actions, is not reduced or smoothed by hardware. This significantly improves the signal-to-noise ratio and accuracy of subsequent model inversion and solves the technical problem of losing weak fault features due to hardware filtering in existing technologies.

[0069] Example 3:

[0070] The signal calculation and separation unit includes:

[0071] The digital low-pass filter module is used to perform convolution operations on voltage and current sequences with a cutoff frequency lower than the upper limit of the fundamental frequency, generating a fundamental data stream for conventional control.

[0072] The digital bandpass filter module is used to construct a filter transfer function with a center frequency locked to the inverter carrier frequency and a bandwidth covering the third harmonic range. It filters voltage and current sequences and outputs transient response data components.

[0073] This embodiment details the internal architecture of the signal calculation and separation unit, enabling the dual use of a single source for both control and diagnostic data;

[0074] Digital low-pass filter module: This module is used for conventional motor control; it filters the original voltage sequence. and current sequence Execution cutoff frequency Convolution operations below the upper limit of the fundamental frequency;

[0075] Generate a fundamental frequency data stream for conventional control; this data stream mainly contains 50Hz / 60Hz and its lower harmonic components, used for closed-loop vector control or torque control;

[0076] Digital bandpass filter module: This module is used for fault diagnosis; it constructs a specific filter transfer function. The transfer function employs a fourth-order Butterworth infinite impulse response (IIR) filter structure; the specific difference equation coefficients... The analog transfer function is derived using the bilinear transform method. Mapping to the digital domain yields the cutoff frequency. Set as the inverter carrier frequency of Double the bandwidth;

[0077] Center frequency Locked to the carrier frequency and bandwidth of the frequency converter Coverage extends to the third harmonic range of the carrier wave;

[0078] The voltage and current sequences are filtered to output only the transient response data components containing the characteristics of switching oscillations. ;

[0079] This embodiment adopts a dual-channel digital filtering architecture, which separates two signals with completely different purposes based on the same set of hardware sampling data; it not only ensures the stability of the motor control system, but also retains the high-frequency characteristics required for fault diagnosis, and realizes software decoupling and parallel processing of control and diagnostic functions.

[0080] Example 4:

[0081] When solving for the high-frequency impedance parameter set, the model parameter inversion unit performs the following computer-implemented steps:

[0082] Identify the falling edge trigger time of the PWM pulse from the transient response data components;

[0083] Extract the time window data from the trigger moment to the moment when the oscillation decays to zero, and construct the target observation waveform vector;

[0084] Initialize the state of the second-order differential equation model containing resistance, inductance, and capacitance variables;

[0085] The gradient descent algorithm or particle swarm optimization algorithm is used to adjust the parameter values ​​in the second-order differential equation model until the Euclidean distance between the predicted waveform vector generated by the model and the target observed waveform vector is less than the preset convergence tolerance.

[0086] The parameter values ​​that meet the convergence condition are locked and output as a set of high-frequency impedance parameters.

[0087] This embodiment describes in detail the algorithm flow of the model parameter inversion unit, which uses a computer to perform the following steps to solve the high-frequency impedance parameter set;

[0088] Step S401: Trigger recognition

[0089] From transient response data components Alternatively, the falling edge trigger time of the PWM pulse can be identified by synchronously referencing the zero-crossing or edge-jumping characteristics of the original voltage sequence. Because the oscillation generated at the moment of IGBT turn-off best reflects the impedance characteristics of the stator circuit;

[0090] Step S402: Window extraction and vector construction

[0091] Excerpted from From the moment the oscillation decays to zero Using time window data to construct the target observation waveform vector The calculation formula is as follows:

[0092] ;

[0093] in, The instantaneous values ​​of current or voltage collected. The sampling time interval of the data acquisition interface unit, i.e. ;

[0094] Step S403: Model Initialization

[0095] Initialization includes resistance variables Inductance variable and capacitance variable The second-order differential equation model is derived by differentiating both sides of the voltage balance equation KVL describing the transient response of an RLC series circuit in time. This equation is used to mathematically transform the PWM voltage step excitation function into a pulse excitation function, simplifying the solution. This second-order differential equation describes the equivalent circuit model of the stator winding of an explosion-proof motor, one phase to ground or the other phase. This represents the high-frequency loss resistance of the winding. Represents stator stray leakage. Representing the distributed capacitance of the winding or the parasitic capacitance to ground, this equivalent circuit model is a distributed RLC series circuit of the stator winding; this model describes the stator winding under PWM voltage step excitation. The response of the series circuit, and the corrected nonhomogeneous differential equation are:

[0096] ;

[0097] in, This is the transient current response function of the stator circuit; The input transient voltage excitation function is given at the trigger time. Place, Modeled as having an amplitude of DC bus voltage The step signal, i.e. exist The time is the Dirac impulse function It should be noted that, considering the finite switching rise time of IGBT devices in actual physical systems... Typically ranging from 100 ns to 500 ns, the aforementioned Dirac pulse is merely an idealized mathematical description; in actual model parameter inversion calculations, this excitation function... It was modified to a ramp step function, that is, in to The voltage rises linearly to [a certain value] within a certain time period. This correction ensures that the model is effective against inductors at high frequencies. and capacitor The accuracy of parameter identification compensates for the transient response delay caused by the non-ideal characteristics of switching devices; These are the unknown parameters to be inverted;

[0098] Step S404: Optimization of normalized parameters

[0099] To eliminate the impact of differences in physical dimensions and orders of magnitude on optimization convergence, this step introduces dynamic normalization processing before performing waveform comparison to calculate the target observed waveform vector. The maximum absolute value is used as the normalization factor :

[0100] ;

[0101] Adjust the model parameter set using gradient descent or particle swarm optimization (PSO) algorithms. Minimize the model parameter set objective function In each iteration, the system generates a predicted waveform vector based on the current parameter set. The matching error is calculated based on the dimensionless objective function, and the corrected optimization objective function is as follows:

[0102] ;

[0103] in, Denotes the Euclidean norm of a vector; by dividing by Mapping the voltage or current waveform in physical space to A dimensionless sequence of intervals, thus ensuring convergence tolerance. It is a purely numerical scalar that is independent of specific physical units, for example, setting... ;

[0104] When dimensionless distance Less than the preset convergence tolerance When the iteration converges, it is determined that the iteration has converged.

[0105] Step S405: Parameter Output

[0106] It will satisfy the convergence condition. The parameter values ​​are locked at that time, serving as a set of high-frequency impedance parameters. Output;

[0107] This embodiment transforms the complex internal state monitoring of a motor into a mathematical optimization problem. By performing refined modeling and inversion of the high-frequency oscillations excited by the falling edge of the PWM, it is possible to accurately obtain the microscopic physical parameters inside the motor that cannot be directly measured. Its accuracy is far higher than that of traditional methods based on steady-state current analysis.

[0108] Example 5:

[0109] The safety status assessment unit executes the following parameter mapping logic:

[0110] Extract the distributed capacitance to ground value from the high-frequency impedance parameter set, calculate the percentage change relative to the initial calibration value, and use it as the aging index of the insulating medium.

[0111] Extract the three-phase leakage inductance values ​​from the high-frequency impedance parameter set, and calculate the ratio of the standard deviation to the arithmetic mean of the three-phase leakage inductance values ​​as the air gap eccentricity index.

[0112] This embodiment defines specific parameter mapping logic, which transforms abstract circuit parameters into intuitive fault indices;

[0113] The safety status assessment unit extracts the ground distributed capacitance value from the high-frequency impedance parameter set. ;

[0114] To quantify the degree of aging, its value relative to the initial calibration value is calculated. The percentage change from the baseline value of the motor at the factory or in its healthy condition is used as the insulation dielectric aging index. :

[0115] ;

[0116] Insulation aging is often accompanied by moisture or chemical decomposition, which directly changes the dielectric constant of the medium, thus leading to... Significant drift occurred;

[0117] The safety status assessment unit extracts the three-phase leakage inductance value from the high-frequency impedance parameter set. ;

[0118] Calculate the standard deviation of the three-phase leakage inductance value. with arithmetic mean The ratio of the two values ​​is used as the air gap eccentricity index. Also known as the coefficient of variation, its calculation formula is as follows:

[0119] Calculate the arithmetic mean of the three-phase leakage inductance values ;

[0120] Calculate the air gap eccentricity index:

[0121] ;

[0122] Under ideal concentric circle conditions, the three-phase leakage inductance is symmetrical; air gap eccentricity will disrupt the magnetic circuit symmetry, causing the leakage inductance of a certain phase to increase or decrease, and this index can keenly detect this imbalance.

[0123] The two index algorithms proposed in this embodiment realize the quantitative characterization of insulation aging due to electrical faults and air gap eccentricity due to mechanical faults, respectively; in particular, they utilize the statistical characteristics of three-phase leakage inductance to diagnose eccentricity, eliminating the influence of load fluctuations on absolute values, and exhibiting extremely high robustness and sensitivity.

[0124] Example 6:

[0125] The system also includes an active detection control unit, which performs the following data quality enhancement logic:

[0126] The ratio of the signal power to the background noise power of the transient response data component is calculated to obtain the real-time signal-to-noise ratio value.

[0127] The real-time signal-to-noise ratio value is compared with the preset minimum confidence threshold;

[0128] If the real-time signal-to-noise ratio is less than the minimum confidence threshold, a high-frequency perturbation injection command is generated and sent to the inverter controller to request the superposition of a jitter signal of a specific frequency in the output voltage.

[0129] If the real-time signal-to-noise ratio is greater than or equal to the minimum confidence threshold, an instruction to maintain the current sampling mode is generated.

[0130] This embodiment introduces an active detection mechanism to enhance data quality in noisy environments;

[0131] The system also includes an active detection control unit, whose execution logic is as follows:

[0132] Triggering time determined according to Example 4 Time-domain partitioning of data: Definition Before The time window for each sampling point is defined as the noise window. to The time window is the signal window; the square of the root mean square (RMS) of the data within the noise window is calculated as the background noise power. The square of the root mean square value of the data within the signal window is used as the signal power. Calculate the ratio of the two to obtain the real-time signal-to-noise ratio value. :

[0133] ;

[0134] Will Compared with the preset minimum confidence threshold Compare;

[0135] like This indicates that the oscillation signal generated under current operating conditions, such as light load or steady-state operation, is too weak to support high-precision parameter inversion. At this point, the unit generates a high-frequency perturbation injection command and sends it to the inverter controller, requesting the superposition of a specific frequency in the output voltage. Specifically, the active detection control unit is connected to the inverter controller via the communication interface of the industrial fieldbus. The high-frequency perturbation injection command contains two parts of data: the injection frequency... and injection amplitude The inverter controller responds to this command by adjusting the d-axis voltage setpoint in SVPWM modulation. The superposition expression is The high-frequency excitation signal; since the d-axis current mainly affects the excitation and does not generate average torque, the perturbation signal can induce a high-frequency response of the stator without causing perceptible torque pulsation to the mechanical load operation of the motor shaft, such as the jitter signal of a pseudo-random sequence near the carrier frequency.

[0136] like This indicates that the signal quality is good, and an instruction is generated to maintain the current sampling mode.

[0137] This embodiment creatively introduces an active excitation mechanism; when the signal-to-noise ratio is low due to passive monitoring failure, the control strategy of the frequency converter is actively changed to inject a perturbation signal, thereby stimulating the characteristic response required by the system; this on-demand injection strategy not only ensures the all-weather availability of diagnostic data, but also minimizes interference with the normal operation of the motor.

[0138] Example 7:

[0139] When generating a security assessment data packet, the security status assessment unit performs the following closed-loop logic judgment:

[0140] Retrieve the insulation failure threshold and rudder sweep risk threshold stored in the database;

[0141] If the aging index of the insulating medium is greater than the insulation failure threshold, the status bit of the safety assessment data packet will be marked as insulation critical, and the insulation prediction life data will be written.

[0142] If the air gap eccentricity index is greater than the sweep risk threshold, the status bit of the safety assessment data packet is marked as mechanical emergency, and the eccentricity orientation data is written.

[0143] If the insulation medium aging index is less than or equal to the insulation failure threshold, and the air gap eccentricity index is less than or equal to the rudder sweep risk threshold, the status bit of the safety assessment data packet is marked as operationally healthy, and written to the status log of the current calculation cycle.

[0144] This embodiment constructs a fully closed-loop fault diagnosis and early warning logic;

[0145] When generating a security assessment data packet, the security status assessment unit executes the following logic:

[0146] Retrieve the preset insulation failure threshold from the database. and the risk threshold of rumping These thresholds are derived from statistical analysis of a large amount of historical fault data; specifically, insulation failure thresholds. It is derived by referring to the specifications for dielectric loss or capacitance change rate in the International Electrotechnical Commission standards or industry practices, and by calibrating in conjunction with actual operating data; the rotor sweep risk threshold. The safety limit is derived by finite element analysis (FEA) of the magnetic field to simulate the leakage inductance imbalance under different eccentricities, based on the motor design parameters and the minimum air gap tolerance requirements of the moving / stationary components.

[0147] Insulation Critical Judgment: If The status bit is marked as insulation critical; the system calls the preset motor thermal model and uses the fundamental current data stream output by the digital low-pass filter module in Example 3 to calculate the Joule thermal integral and estimate the real-time temperature of the current stator winding. Simultaneously, using the remaining lifetime prediction model, where the insulation dielectric aging index... The input variable is used as a weighting factor to correct the reaction rate constant or activation energy of the insulating material. and Write insulation prediction life data to prompt maintenance personnel to replace the equipment before the expected failure time;

[0148] Mechanical emergency assessment: If Mark the status as mechanical emergency; at the same time, determine the minimum air gap position based on the unbalanced distribution of the three-phase leakage inductance, such as which phase has the smallest leakage, and write the eccentric orientation data.

[0149] Health status determination: If none of the above indices exceed the limits, it is marked as running healthily and written to the status log of the current calculation cycle for the purpose of establishing a full life cycle health record;

[0150] This embodiment uses multi-level threshold judgment logic to not only provide a binary conclusion of fault presence / fault absence, but also to provide in-depth diagnostic information such as remaining lifespan and fault location. This hierarchical early warning mechanism can effectively guide maintenance personnel to formulate differentiated maintenance strategies and avoid over-maintenance or under-maintenance.

[0151] Example 8:

[0152] The system is configured as a non-invasive data processing terminal, which is deployed outside the safety boundary of the explosion-proof hazardous area. It analyzes the digital electrical signals transmitted by the data acquisition interface unit and reconstructs the state of the insulation medium and the geometric topology of the air gap inside the explosion-proof motor by using a digital twin equivalent circuit model without implanting physical sensors inside the explosion-proof motor or damaging the sealing integrity of the explosion-proof enclosure.

[0153] This embodiment clarifies the physical form and application scenario of the system; the system is configured as a non-intrusive data processing terminal;

[0154] Physical deployment: Terminals are deployed outside the safety boundaries of explosion-proof hazardous areas, such as Zone 0 or Zone 1 of a chemical plant, such as in a power distribution room or MCC control center;

[0155] Data flow: It only operates by analyzing the digital electrical signals transmitted through the data acquisition interface unit;

[0156] Operating prerequisites: Operation is permitted without the need to implant physical sensors inside the explosion-proof motor or compromise the sealing integrity of the explosion-proof enclosure;

[0157] Core function: Using a digital twin equivalent circuit model, remotely invert and reconstruct the state of the insulation medium and the geometric topology of the air gap inside the explosion-proof motor;

[0158] This embodiment emphasizes the non-invasive and remote diagnostic features of the system. For explosion-proof motors, installing sensors by opening the cover is not only costly, but also involves complex explosion-proof certification re-examination issues. This system achieves insight into the internal physical state through pure electrical signal analysis, solving the compliance and construction problems of equipment condition monitoring in explosion-proof environments, and greatly reducing the implementation cost in industrial sites.

[0159] 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 motor operation simulation fault diagnosis system, characterized in that, include: The data acquisition interface unit is configured to connect to the analog-to-digital converter of the explosion-proof motor stator circuit, and is used to read the unfiltered digital discrete voltage and current sequences at a sampling rate that is higher than a preset multiple of the inverter carrier frequency. The signal calculation and separation unit is used to perform digital bandpass filtering operations on voltage and current sequences to extract transient response data components containing high-frequency switching oscillation characteristics from the fundamental energy transmission signal. The model parameter inversion unit is used to call the preset digital twin equivalent circuit model of the explosion-proof motor, take the transient response data components as the excitation input of the model, and iteratively calculate the set of high-frequency impedance parameters on the stator side of the motor at the current moment by minimizing the sum of squared residuals between the model output and the measured data. The safety status assessment unit is used to map the parasitic capacitance values ​​in the high-frequency impedance parameter set to the insulation dielectric aging index, map the imbalance of the leakage inductance parameter to the air gap eccentricity index, and generate a safety assessment data package for the explosion-proof motor based on the insulation dielectric aging index and the air gap eccentricity index. The safety status assessment unit executes the following parameter mapping logic: Extract the distributed capacitance to ground value from the high-frequency impedance parameter set, calculate the percentage change relative to the initial calibration value, and use it as the aging index of the insulating medium. Extracting the three-phase leakage inductance value from the high-frequency impedance parameter set Calculate the standard deviation of the three-phase leakage inductance value. with arithmetic mean The ratio of the two values ​​is used as the air gap eccentricity index. ; Calculate the arithmetic mean of the three-phase leakage inductance values ; Calculate the air gap eccentricity index: ； When solving for the high-frequency impedance parameter set, the model parameter inversion unit performs the following computer-implemented steps: Identify the falling edge trigger time of the PWM pulse from the transient response data components; Extract the time window data from the trigger moment to the moment the oscillation decays to zero, and construct the target observation waveform vector. ; Excerpted from From the moment the oscillation decays to zero Using time window data to construct the target observation waveform vector The calculation formula is as follows: ； in, The instantaneous values ​​of current or voltage collected. The sampling time interval of the data acquisition interface unit, i.e. ; Initialization includes resistance variables Inductance variable and capacitance variable The second-order differential equation model state; this second-order differential equation describes the equivalent circuit model of the stator winding of the explosion-proof motor relative to ground or relative to another phase, where This represents the high-frequency loss resistance of the winding. Represents stator stray leakage. Representing the distributed capacitance of the winding or the parasitic capacitance to ground, this equivalent circuit model is a distributed RLC series circuit of the stator winding; this model describes the stator winding under PWM voltage step excitation. The response of the series circuit, and the corrected nonhomogeneous differential equation are: ； in, This is the transient current response function of the stator circuit; The input transient voltage excitation function is given at the trigger time. Place, Modeled as having an amplitude of DC bus voltage The step signal, i.e. exist The time is the Dirac impulse function ; The gradient descent algorithm or particle swarm optimization algorithm is used to adjust the parameter values ​​in the second-order differential equation model until the Euclidean distance between the predicted waveform vector generated by the model and the target observed waveform vector is less than the preset convergence tolerance. The parameter values ​​that meet the convergence condition are locked and output as a set of high-frequency impedance parameters. To eliminate the impact of differences in physical dimensions and orders of magnitude on optimization convergence, this step introduces dynamic normalization processing before performing waveform comparison to calculate the target observed waveform vector. The maximum absolute value is used as the normalization factor : ； Adjust the model parameter set using gradient descent or particle swarm optimization (PSO) algorithms. Minimize the model parameter set objective function In each iteration, the system generates a predicted waveform vector based on the current parameter set. The matching error is calculated based on the dimensionless objective function, and the corrected optimization objective function is as follows: ； in, Denotes the Euclidean norm of a vector; by dividing by Mapping the voltage or current waveform in physical space to A dimensionless sequence of intervals, thus ensuring convergence tolerance. It is a purely numerical scalar that is independent of specific physical units, for example, setting... ; When dimensionless distance Less than the preset convergence tolerance When the iteration converges, it is determined that the iteration has converged.

2. The motor operation simulation fault diagnosis system according to claim 1, characterized in that, The data acquisition interface unit performs data reading operations based on oversampling technology to capture non-stationary time series data output by the frequency converter. The sampling rate is set to be no less than ten times the carrier frequency of the frequency converter, and the data reading path avoids any hardware analog low-pass filter in order to retain the high-frequency harmonic characteristics that can characterize the slight changes in the dielectric constant inside the motor.

3. The motor operation simulation fault diagnosis system according to claim 1, characterized in that, The signal calculation and separation unit includes: The digital low-pass filter module is used to perform convolution operations on voltage and current sequences with a cutoff frequency lower than the upper limit of the fundamental frequency, generating a fundamental data stream for conventional control. The digital bandpass filter module is used to construct a filter transfer function with a center frequency locked to the inverter carrier frequency and a bandwidth covering the third harmonic range. It filters voltage and current sequences and outputs transient response data components.

4. The motor operation simulation fault diagnosis system according to claim 1, characterized in that, The system also includes an active detection control unit, which performs the following data quality enhancement logic: The ratio of the signal power to the background noise power of the transient response data component is calculated to obtain the real-time signal-to-noise ratio value. The real-time signal-to-noise ratio value is compared with the preset minimum confidence threshold; If the real-time signal-to-noise ratio is less than the minimum confidence threshold, a high-frequency perturbation injection command is generated and sent to the inverter controller to request the superposition of a jitter signal of a specific frequency in the output voltage. If the real-time signal-to-noise ratio is greater than or equal to the minimum confidence threshold, an instruction to maintain the current sampling mode is generated.

5. The motor operation simulation fault diagnosis system according to claim 4, characterized in that, When generating a security assessment data packet, the security status assessment unit performs the following closed-loop logic judgment: Retrieve the insulation failure threshold and rudder sweep risk threshold stored in the database; If the aging index of the insulating medium is greater than the insulation failure threshold, the status bit of the safety assessment data packet will be marked as insulation critical, and the insulation prediction life data will be written. If the air gap eccentricity index is greater than the sweep risk threshold, the status bit of the safety assessment data packet is marked as mechanical emergency, and the eccentricity orientation data is written. If the insulation medium aging index is less than or equal to the insulation failure threshold, and the air gap eccentricity index is less than or equal to the rudder sweep risk threshold, the status bit of the safety assessment data packet is marked as operationally healthy, and written to the status log of the current calculation cycle.

6. The motor operation simulation fault diagnosis system according to claim 1, characterized in that, The system is configured as a non-invasive data processing terminal, which is deployed outside the safety boundary of the explosion-proof hazardous area. It analyzes the digital electrical signals transmitted by the data acquisition interface unit and reconstructs the state of the insulation medium and the geometric topology of the air gap inside the explosion-proof motor by using a digital twin equivalent circuit model without implanting physical sensors inside the explosion-proof motor or damaging the sealing integrity of the explosion-proof enclosure.

Images