Condition monitoring and control system for permanent magnet motors driven by oil drilling rigs

Abstract

This invention discloses a condition monitoring and control system for a permanent magnet motor driven by an oil drilling rig, relating to the field of permanent magnet motor condition monitoring technology. It includes a data acquisition and construction module, a data preprocessing and feature extraction module, a dynamic monitoring and trend analysis module, a deep learning evaluation module, and a fuzzy control and real-time compensation module. The data acquisition and construction module first acquires real-time operating data of the inverter during operation through data acquisition equipment. The acquired real-time data is stored in a time-series format, and a dataset is established to provide complete and accurate basic data for subsequent analysis and model evaluation. This invention improves the stability, reliability, and energy efficiency of the permanent magnet motor driven by the oil drilling rig through real-time data acquisition, deep learning evaluation, fuzzy control optimization, and error compensation, adaptively adjusting the dead time, reducing short-circuit risk, energy loss, and harmonic damage, optimizing system performance, and improving operating efficiency.

Description

Technical Field

[0001] This invention relates to the field of permanent magnet motor condition monitoring technology, and more specifically to a permanent magnet motor condition monitoring and control system based on oil drilling rig drive. Background Technology

[0002] Condition monitoring and control of permanent magnet motors (PMMs) driven by oil drilling rigs refers to the real-time monitoring of the operating status of the PMM and its drive system during oil drilling operations, and the effective adjustment and optimization through a control system. PMMs are widely used in oil drilling rigs due to their high efficiency, long lifespan, and high power density. Condition monitoring collects various operating data of the motor through sensors, such as current, voltage, temperature, and vibration, and uses data analysis and algorithms for fault diagnosis, performance evaluation, and trend prediction. Through feedback from the control system, precise control of the motor can be achieved, avoiding problems such as overload and overheating, improving the reliability of the motor and the working efficiency of the oil drilling rig. Simultaneously, dynamic adjustment and optimization of control strategies extend the service life of the equipment, reduce maintenance costs, and ensure the smooth progress of the drilling process.

[0003] In a permanent magnet motor drive system, the controller serves as the core command and control unit, monitoring and precisely controlling the motor's operating status in real time. Through advanced motor control algorithms, the controller dynamically adjusts the drive signal based on real-time data acquisition. Employing SVPWM (Space Vector Pulse Width Modulation) technology, the optimized voltage signal is converted into a three-phase PWM signal to drive the inverter, thereby precisely controlling the input voltage and current of the permanent magnet motor and ensuring the system's high efficiency, stability, and safety.

[0004] The existing technology has the following shortcomings:

[0005] In existing technologies, SVPWM inverter control typically uses a dead time to prevent the upper and lower IGBTs (Insulated Gate Bipolar Transistors) from conducting simultaneously (short-circuiting). However, improper dead time settings (too short or too long) can lead to serious problems. A dead time that is too short may cause the upper and lower IGBTs to conduct simultaneously, resulting in a shoot-through short circuit and subjecting the IGBTs to a huge inrush current, potentially even causing instantaneous burnout. A dead time that is too long will cause output voltage waveform distortion, reduce the operating efficiency of the permanent magnet motor, and may increase higher harmonics, leading to winding overheating, increased vibration, and even insulation breakdown, severely affecting the reliability and lifespan of the permanent magnet motor.

[0006] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide a condition monitoring and control system for a permanent magnet motor driven by an oil drilling rig. Based on real-time data acquisition, deep learning intelligent evaluation, fuzzy control dynamic optimization, and error compensation, it improves the stability, reliability, and energy efficiency of the permanent magnet motor driven by the oil drilling rig. Data preprocessing ensures accurate model input, deep learning intelligent evaluation optimizes dead-time settings, fuzzy control adaptively adjusts to avoid short-circuit risks and energy losses, and the error compensation mechanism corrects waveform distortion, reduces harmonic losses and winding overheating, and extends motor life. This invention effectively solves the dead-time mismatch problem, optimizes the performance of the motor drive system, and improves the operational stability and efficiency of oil drilling rigs under complex working conditions, thus addressing the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a permanent magnet motor condition monitoring and control system based on oil drilling rig drive, comprising a data acquisition and construction module, a data preprocessing and feature extraction module, a dynamic monitoring and trend analysis module, a deep learning evaluation module, and a fuzzy control and real-time compensation module:

[0009] The data acquisition and construction module first acquires real-time operating data of the inverter through data acquisition equipment. The acquired real-time data are stored in time series format and a dataset is established to provide complete and accurate basic data for subsequent analysis and model evaluation.

[0010] The data preprocessing and feature extraction module preprocesses the inverter operating parameter information in the dataset and extracts the core parameters that intuitively reflect the dead time mismatch from the preprocessed data, which serve as key indicators for evaluating whether the dead time is matched.

[0011] The dynamic monitoring and trend analysis module performs fine-grained analysis on the extracted key indicators under the detection window to capture the dynamic change trend caused by dead time mismatch.

[0012] The deep learning evaluation module inputs the analyzed key indicators into the pre-learned deep learning model, and uses the deep learning model to intelligently evaluate whether the dead time setting is matched from complex feature relationships;

[0013] The fuzzy control and real-time compensation module adjusts the dead time dynamically according to preset fuzzy rules based on the evaluation results when the evaluation results show that the dead time does not match the actual operation. At the same time, it uses digital signal processing technology to compensate for the output error caused by the dead time in real time, correcting the output error caused by the change in dead time.

[0014] Preferably, the specific steps for acquiring real-time operating data of the inverter during operation through data acquisition equipment and establishing a data set are as follows:

[0015] First, deploy high-precision data acquisition equipment in the inverter system to ensure real-time monitoring of the inverter's input, output, and the status of key components;

[0016] Secondly, the acquired signals are synchronized by using a high-precision clock to timestamp and align the data from all sensors to ensure data consistency.

[0017] Then, the data is stored in a time series manner through a high sampling rate data recording system to capture transient changes caused by dead time mismatch during inverter operation;

[0018] Next, we will establish a dataset for merging and classification management, and label the data according to the inverter's operating conditions for subsequent analysis.

[0019] Finally, by checking data integrity and removing outlier data points through interpolation compensation, the integrity and accuracy of the dataset are ensured, thus providing high-quality data support for subsequent feature analysis, deep learning evaluation, and dynamic regulation.

[0020] Preferably, core parameters that intuitively reflect the dead time mismatch are extracted from the preprocessed data. The extracted parameters include DC bus voltage fluctuation and IGBT actual turn-off time deviation in pulse operating mode. The DC bus voltage fluctuation and IGBT actual turn-off time deviation in pulse operating mode are analyzed under the detection window to generate DC bus feedback anomaly reference value and IGBT turn-off delay reference value, respectively. The dynamic change trend caused by dead time mismatch is quantified by the DC bus feedback anomaly reference value and IGBT turn-off delay reference value.

[0021] Preferably, the specific steps for analyzing DC bus voltage fluctuations in pulse operating mode within the detection window to generate DC bus feedback anomaly reference values ​​are as follows:

[0022] In pulse operating mode, the DC bus voltage signal exhibits periodic fluctuations. However, dead-time mismatch leads to increased irregularity in voltage fluctuations. Therefore, nonlinear feature extraction is first performed on the bus voltage signal to analyze its higher-order derivative cumulative offset, in order to quantify the fluctuation trend of the bus voltage. The feature extraction formula is as follows:

[0023]

[0024] In the formula, F1 is a nonlinear high-order characteristic exponent of DC bus voltage fluctuation, used to measure the severity of DC bus voltage fluctuation, and N is the total number of time points in the detection window. It is the third derivative of the DC bus voltage, V dc (k) represents the bus voltage value acquired at time k, and α is a nonlinear weighting coefficient used to adjust the influence of the third derivative term. It is the exponential decay term of voltage fluctuation, calculated at time k, V dc (k) and the previous time V dc The voltage variation between (k-1) is where e is the natural base and β is the exponential decay factor.

[0025] Abnormal fluctuations in bus voltage are not only manifested in sudden changes, but also lead to uneven energy feedback from the bus. Therefore, a nonlinear energy offset function is introduced to measure the non-uniformity of the bus voltage energy distribution, as shown in the following formula:

[0026]

[0027] In the formula, F2 is the imbalance index of DC bus energy feedback, and E rec (k) represents the stored energy at time point k, E del (k) is the energy transmitted from the bus to the load side at time k, ε is a small value to prevent the denominator from being zero and to ensure calculation stability, and γ is the exponential adjustment coefficient to control the contribution of bus feedback imbalance to the abnormal exponent.

[0028] By combining the nonlinear high-order characteristic exponent F1 of DC bus voltage fluctuation and the imbalance exponent F2 of DC bus energy feedback, and introducing a nonlinear weight distribution function, an abnormal reference value for DC bus feedback is generated. The generation formula is as follows:

[0029]

[0030] In the formula, λ1 is the weighting coefficient of the nonlinear higher-order characteristic index F1 of DC bus voltage fluctuation, λ2 is the weighting coefficient of the imbalance index F2 of DC bus energy feedback, η1 is the exponential adjustment factor of the nonlinear higher-order characteristic index F1 of DC bus voltage fluctuation, and η2 is the exponential adjustment factor of the imbalance index F2 of DC bus energy feedback. It is the exponential decay factor, where e is the natural base and ζ is a dynamic adjustment parameter used to adjust the exponential decay factor. Sensitivity.

[0031] Preferably, the specific steps for analyzing the actual IGBT turn-off time deviation within the detection window to generate an IGBT turn-off delay reference value are as follows:

[0032] Within the detection window, key signals during the IGBT turn-off process, including changes in current, voltage, and power loss, are monitored using a high-speed data acquisition system. Since the IGBT turn-off process involves rapid current decay, a turn-off transient characteristic function is defined to capture the mismatch between the current decrease rate and the voltage rise rate. The formula is as follows:

[0033]

[0034] In the formula, Φ off It is the IGBT turn-off transient characteristic, I ds It is the drain-source current when the IGBT is turned off, V ce It is the collector-emitter voltage when the IGBT is turned off, Q g It is the gate charge. It is the rate of current decrease during IGBT turn-off. It is the voltage rise rate during IGBT turn-off, and ρ is a weighting factor used to balance the influence weights of the current fall rate and the voltage rise rate.

[0035] During the turn-off process, the loss characteristics of IGBTs are affected by the dead time. When the dead time is too short, the current does not decay completely, leading to an increase in the instantaneous power peak. Therefore, an IGBT turn-off loss offset function is introduced to characterize the energy offset during the turn-off process. The calculation formula is as follows:

[0036]

[0037] In the formula, Ψ loss It is the IGBT turn-off loss offset, Ω off It is the time region when the IGBT is turned off, P sw (ζ) is the actual measured instantaneous switching power loss of the IGBT, P ideal (ζ) is the theoretically calculated IGBT turn-off power loss, and μ is the adjustment factor, which is used to adjust the proportional factor of the theoretical loss reference.

[0038] Finally, the transient characteristics Φ of IGBT turn-off are obtained. off and IGBT turn-off loss offset Ψ loss Based on comprehensive analysis, a reference value for IGBT turn-off delay is generated, and the formula is as follows:

[0039]

[0040] In the formula, ITDI is the IGBT turn-off delay reference value, max(Φ off ) represents the maximum IGBT turn-off transient characteristic within the detection window, max(Ψ) loss ) represents the offset of the maximum IGBT turn-off loss within the detection window, and κ1 represents the IGBT turn-off transient characteristic Φ.off The weighting factor, κ2 is the IGBT turn-off loss offset Ψ. loss Weighting factors It is a normalization of the IGBT turn-off transient characteristics. This is a normalization of the IGBT turn-off energy offset. β1 and β2 are both normalized exponential adjustment factors, with β2 used to control the normalized IGBT turn-off loss offset Ψ. loss In the calculation, β1 is used to control the normalized IGBT turn-off transient characteristic Φ. off Nonlinear effects in calculations.

[0041] Preferably, the analyzed DC bus feedback anomaly reference value and IGBT turn-off delay reference value are input into a pre-learned deep learning model. The deep learning model generates a dead time mismatch risk coefficient, and the dead time setting is intelligently evaluated based on the dead time mismatch risk coefficient.

[0042] Preferably, the dead time mismatch risk coefficient generated when intelligently evaluating the dead time setting using a pre-learned deep learning model is compared and analyzed with a preset dead time mismatch risk coefficient reference threshold to classify the dead time setting. The classification steps are as follows:

[0043] If the dead time mismatch risk coefficient is greater than the reference threshold for dead time mismatch risk coefficient, the current dead time setting will be classified as mismatched with actual operation; if the dead time mismatch risk coefficient is less than or equal to the reference threshold for dead time mismatch risk coefficient, the current dead time setting will be classified as matched with actual operation.

[0044] Preferably, when the evaluation results show that the dead time does not match the actual operation, the fuzzy logic controller dynamically adjusts the dead time according to the preset fuzzy rules based on the evaluation results, and simultaneously uses digital signal processing technology to compensate for the output error caused by the dead time in real time. The specific steps are as follows:

[0045] During the dead time adjustment process, the fuzzy logic controller is first used to calculate the dead time adjustment amount based on the dead time mismatch risk coefficient (DTM) and the dead time mismatch risk coefficient reference threshold. The calculation expression is as follows:

[0046] ΔDT=μ f ·(α1·(DTM-TDM ref )+α2·|I spk |+α3·|THD|)

[0047] In the formula, ΔDT is the dead time adjustment amount, μ fIt is the fuzzy adjustment factor, α1 is the dead time deviation weighting factor, TDM ref This is the reference threshold for the dead time mismatch risk factor, I. spk α1 is the current peak amplitude, α2 is the current peak weighting factor, which represents the weight of the current peak amplitude on the dead time adjustment, THD is the total harmonic distortion, and α3 is the total harmonic distortion weighting factor, which represents the weight of the total harmonic distortion on the dead time adjustment.

[0048] After determining the dead time adjustment ΔDT, the inverter's PWM modulation strategy is updated in real time to ensure that the dead time adjustment does not affect the overall stability. The calculation expression is as follows:

[0049]

[0050] DT new This is the updated dead time, DT. old It is the original dead time, e -ω·DTM This is the exponential adjustment term, an exponential scaling factor used to control the dead-time adjustment rate, where e is the natural base.

[0051] ω is the adjustment response factor, which controls the adjustment speed. It is the ripple component of the DC bus voltage. It is the rated voltage of the DC bus. It is the normalized DC bus voltage ripple. It is the DC bus voltage ripple weighting factor. It is the load current commutation dynamic weighting factor. It is the rate of change of load current with respect to commutation voltage;

[0052] Since dynamic adjustment of the dead time affects the output voltage and current waveforms of the motor, digital signal processing technology is needed to compensate for the error caused by the dead time to ensure that the output quality is not affected. The calculation expression is as follows:

[0053] In the formula, V comp This is the compensation voltage, θ1 is the dead time variation weighting factor, θ2 is the voltage error weighting factor, V ideal It is the ideal output voltage, V real This is the actual output voltage. θ3 is the second-order voltage change rate, θ4 is the voltage second derivative weighting factor, and θ5 is the current ripple weighting factor, which measures the influence of current ripple on the compensation voltage. It is the normalized current ripple, I ripple It is the current ripple amplitude, I nominal It is the rated current, used to normalize the contribution of current ripple to error compensation.

[0054] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0055] This invention, based on real-time data acquisition, deep learning intelligent evaluation, fuzzy control dynamic optimization, and DSP error compensation, constructs an adaptive dead-time control mechanism that effectively improves the stability, reliability, and energy efficiency of the permanent magnet motor system driven by oil drilling rigs. Through data acquisition and preprocessing, the integrity and accuracy of the input model are ensured, improving the precision and reliability of the evaluation. A deep learning model is used to uncover the complex relationship between the dead-time setting and the actual operating state, achieving high-precision intelligent evaluation. Combined with a fuzzy logic controller, the dead-time is adaptively adjusted under different operating conditions, avoiding the risk of shoot-through short circuits caused by excessively short dead times, while reducing voltage distortion, high-order harmonics, and energy loss caused by excessively long dead times, thus optimizing the motor's output performance. Furthermore, the DSP error compensation mechanism further corrects the impact of dead-time adjustment on the output waveform, ensuring stable voltage and current waveforms of the inverter, reducing harmonic losses, and minimizing winding overheating, thereby extending the motor's service life and improving the overall operating efficiency of the oil drilling rig system. This invention not only effectively solves the safety hazards caused by dead time mismatch, but also optimizes the performance of the motor drive system through intelligent means, significantly improving the operational stability and working efficiency of the permanent magnet motor driven by the oil drilling rig under complex working conditions. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0057] Figure 1 This is a schematic diagram of the module of the permanent magnet motor condition monitoring and control system based on oil drilling rig drive of the present invention. Detailed Implementation

[0058] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0059] This invention provides, for example Figure 1 The condition monitoring and control system for a permanent magnet motor driven by an oil drilling rig, as shown, includes a data acquisition and construction module, a data preprocessing and feature extraction module, a dynamic monitoring and trend analysis module, a deep learning evaluation module, and a fuzzy control and real-time compensation module.

[0060] The data acquisition and construction module first acquires real-time operating data of the inverter through data acquisition equipment. The acquired real-time data are stored in time series format and a dataset is established to provide complete and accurate basic data for subsequent analysis and model evaluation.

[0061] During inverter operation data acquisition, to ensure the accuracy and completeness of subsequent analysis, high-precision data acquisition equipment (such as sensors for voltage, current, temperature, and vibration) is needed to collect key parameters of the inverter in real time under different operating conditions. These parameters include three-phase output current, voltage waveform, DC bus voltage, IGBT switching status, switching temperature, and motor operating parameters. All acquired data is stored in a time-series format, meaning each data point is matched with a timestamp, enabling dynamic analysis based on time evolution patterns. Simultaneously, all data is organized into datasets to ensure coverage of operating information under different conditions, providing complete and high-quality foundational data for subsequent stages such as deep learning model training, feature extraction, and dead-time assessment. The key objective of this process is to obtain accurate and comprehensive real-time data, providing reliable input for intelligent analysis and control optimization.

[0062] The specific steps for acquiring real-time operating data of the inverter and establishing a dataset are as follows: First, deploy high-precision data acquisition equipment (such as sensors for current, voltage, temperature, and vibration) in the inverter system to ensure real-time monitoring of the inverter's input, output, and key component status. Second, synchronize the acquired signals by using a high-precision clock to timestamp and align the data from all sensors, ensuring data consistency. Then, store the data in a time-series format using a high-sampling-rate data recording system to capture transient changes caused by dead-time mismatches during inverter operation, such as current spikes and voltage distortions. Next, establish and merge the dataset for categorized management, labeling the data according to the inverter's operating conditions (such as different loads, speeds, and temperature conditions) for subsequent analysis. Finally, perform data integrity checks, remove outlier data points, and perform interpolation compensation to ensure the integrity and accuracy of the dataset, thus providing high-quality data support for subsequent feature analysis, deep learning evaluation, and dynamic control.

[0063] The data preprocessing and feature extraction module preprocesses the inverter operating parameter information in the dataset and extracts the core parameters that intuitively reflect the dead time mismatch from the preprocessed data, which serve as key indicators for evaluating whether the dead time is matched.

[0064] Due to the complex operating environment of inverters, the raw data collected may contain noise interference, missing values, and inconsistent sampling frequencies. In the preprocessing stage, the data needs to be denoised, filtered, interpolated, and normalized to improve the accuracy and stability of subsequent analysis. For example, moving averages or wavelet transforms can be used to smooth voltage and current waveforms and remove high-frequency interference; for missing data, linear interpolation or interpolation based on adjacent samples can be performed to improve the completeness of the dataset. Through preprocessing, useful information can be preserved to the greatest extent possible, while reducing the interference of invalid or erroneous data on judgment.

[0065] The dynamic monitoring and trend analysis module performs fine-grained analysis on the extracted key indicators under the detection window to capture the dynamic change trend caused by dead time mismatch.

[0066] The core parameters that intuitively reflect the dead time mismatch are extracted from the preprocessed data. The extracted parameters include DC bus voltage fluctuation and IGBT actual turn-off time deviation in pulse operating mode. The DC bus voltage fluctuation and IGBT actual turn-off time deviation in pulse operating mode are analyzed under the detection window to generate DC bus feedback anomaly reference value and IGBT turn-off delay reference value respectively. The dynamic change trend caused by dead time mismatch is quantified by the DC bus feedback anomaly reference value and IGBT turn-off delay reference value together.

[0067] The increased DC bus voltage fluctuations in pulse operating mode indicate a mismatch between the inverter dead time setting and actual operation. This is primarily due to the dynamic impact of the dead time on the inverter's commutation process, leading to unstable fluctuations in the DC bus. In pulse operating mode, the inverter's switching frequency is high, requiring a stable bus voltage to support normal operation. However, both excessively short and excessively long dead times can cause abnormal bus voltage fluctuations. When the dead time is too short, the upper and lower IGBT arms of the inverter may experience short-term cross-conduction, causing a momentary energy loss from the DC bus, resulting in a short-term voltage drop and potential bus current surges, leading to high-frequency oscillations in the bus voltage. When the dead time is too long, the "off" state of the inverter output voltage increases during commutation, potentially causing energy from the load side to be instantaneously fed back to the bus, resulting in an abnormal rise in bus voltage and even exacerbating bus voltage waveform distortion or oscillations in the power circuit. The greater the fluctuation range of the bus voltage, the higher the degree of dead-time mismatch. This not only affects the power supply stability of the inverter but may also increase the switching losses of power devices, accelerate the aging of bus capacitors, and reduce system reliability. Therefore, the increase in DC bus voltage fluctuation in pulse operation mode is an important indicator of the mismatch between the inverter dead-time setting and actual operation, and needs to be corrected through adaptive optimization strategies to ensure the stability and efficient operation of the system.

[0068] The specific steps for analyzing DC bus voltage fluctuations under pulse operating mode within the detection window to generate DC bus feedback anomaly reference values ​​are as follows:

[0069] In pulse operating mode, the DC bus voltage signal exhibits periodic fluctuations. However, dead-time mismatch leads to increased irregularity in voltage fluctuations. Therefore, nonlinear feature extraction is first performed on the bus voltage signal to analyze its higher-order derivative cumulative offset, in order to quantify the fluctuation trend of the bus voltage. The feature extraction formula is as follows:

[0070]

[0071] In the formula, F1 is a nonlinear high-order characteristic exponent of DC bus voltage fluctuation, used to measure the severity of DC bus voltage fluctuation. The higher the value, the stronger the voltage fluctuation, which may indicate a dead time mismatch. N is the total number of time points in the detection window. It is the third derivative of the DC bus voltage, representing the degree of abrupt change in the signal over time, V dc (k) represents the bus voltage value acquired at time k, and α is a nonlinear weighting coefficient used to adjust the influence of the third derivative term. It is the exponential decay term of voltage fluctuation, calculated at time k, V dc (k) and the previous time V dc The voltage variation between (k-1) is controlled by an exponential decay function, which controls its impact on the final F1 value. e is the natural base, and β is the exponential decay factor, which is used to control the decay rate of the voltage fluctuation term and adjust the system's sensitivity to large voltage fluctuations.

[0072] This step, by calculating the cumulative shift of the third derivative of the bus voltage, can capture the severity of transient changes in the bus voltage. If the dead time is mismatched, the higher-order derivatives of the voltage fluctuation will increase significantly, thereby enhancing F1, indicating a stronger intensity of abnormal fluctuations in the bus voltage.

[0073] Abnormal fluctuations in bus voltage are not only manifested in sudden changes, but also lead to uneven energy feedback from the bus. Therefore, a nonlinear energy offset function is introduced to measure the non-uniformity of the bus voltage energy distribution, as shown in the following formula:

[0074]

[0075] In the formula, F2 is the imbalance index of DC bus energy feedback, used to measure the abnormal bus energy flow caused by dead time mismatch, and E rec (k) represents the stored energy at time k, used to quantify the amount of energy fed back to the bus by the load, E del(k) is the energy transmitted from the bus to the load side at time k, ε is a small value to prevent the denominator from being zero and to ensure calculation stability, and γ is the exponential adjustment coefficient to control the contribution of bus feedback imbalance to the abnormal exponent.

[0076] This step calculates the deviation between bus energy feedback and load energy transfer. Ideally, energy feedback and load transfer should be nearly balanced, but if the dead time is too long, it may lead to insufficient load energy supply during commutation, increasing bus feedback energy and causing E... rec Much greater than E del Conversely, an excessively short dead time may lead to cross-conduction, increasing load power and disrupting the bus energy balance. Therefore, an increase in the F2 value indicates a more severe abnormality in bus feedback and a higher likelihood of a mismatch in dead time settings.

[0077] By combining the nonlinear high-order characteristic exponent F1 of DC bus voltage fluctuation and the imbalance exponent F2 of DC bus energy feedback, and introducing a nonlinear weight distribution function, an abnormal reference value for DC bus feedback is generated. The generation formula is as follows:

[0078]

[0079] In the formula, λ1 is the weighting coefficient of the nonlinear higher-order characteristic index F1 of DC bus voltage fluctuation, λ2 is the weighting coefficient of the imbalance index F2 of DC bus energy feedback, η1 is the exponential adjustment factor of the nonlinear higher-order characteristic index F1 of DC bus voltage fluctuation, and η2 is the exponential adjustment factor of the imbalance index F2 of DC bus energy feedback.

[0080] It is the exponential decay factor, used to adjust the trade-off between F1 and F2 to prevent a single feature from excessively dominating the final anomalous exponent. e is the natural base, and ζ is a dynamically adjusted parameter used to adjust the exponential decay factor. Sensitivity.

[0081] This step uses nonlinear weighted fusion of F1 and F2 to ensure that the DC bus feedback anomaly index can comprehensively characterize the voltage fluctuation anomaly caused by dead time mismatch. When the DBFAI value is large, it indicates that the bus voltage fluctuation is severe or the bus energy feedback is highly unbalanced, meaning that the dead time may be set inappropriately and dynamic optimization adjustment is required.

[0082] The larger the DC bus feedback abnormal reference value generated after analyzing the DC bus voltage fluctuations under the pulse operating mode within the detection window, the more it indicates a mismatch between the inverter's dead time setting and actual operation. Conversely, a smaller dead time setting indicates a more reasonable dead time setting and better match with actual operation. In pulse operating mode, the amplitude and frequency of DC bus voltage fluctuations are directly affected by the dead time. When the dead time is too short, the IGBTs on the upper and lower bridge arms of the inverter may experience brief cross-conduction, causing drastic fluctuations in the bus current, which in turn leads to a momentary drop or high-frequency oscillation of the bus voltage, resulting in an increase in the DC bus feedback abnormal reference value. On the other hand, an excessively long dead time will cause a commutation time lag in the inverter's output voltage, preventing the effective transfer of energy from the load side during the dead time period, thus causing an abnormal increase in the bus voltage or instability in power feedback, also leading to an abnormal rise in this reference value. Therefore, a larger DC bus feedback abnormal reference value indicates more significant bus voltage fluctuations, suggesting a mismatch between the dead time setting and actual operating conditions, requiring optimization and adjustment. Conversely, if the reference value remains at a low level, it indicates that the dead time setting is reasonable, the bus voltage fluctuation is within the normal range, the inverter is operating stably overall, and the dead time matching degree is high.

[0083] The delayed actual turn-off time of the IGBT indicates a mismatch between the inverter's dead time setting and actual operation. This is because the dead time directly affects the IGBT's switching dynamic characteristics and the stability of the commutation process. Under normal circumstances, the IGBT needs to complete turn-off within the set dead time to ensure that the upper and lower bridge arms do not cross-conduct. However, if the dead time is set too short, the lower bridge arm IGBT may have already turned on before the IGBT turns off, resulting in a brief shoot-through short circuit. This exposes the IGBT to abnormal transient inrush current, which may exacerbate the IGBT's carrier storage effect, leading to a delayed turn-off. On the other hand, if the dead time is set too long, the load current may lack an effective freewheeling path during commutation, causing additional current oscillations after the IGBT turns off, further prolonging the turn-off time. In addition, an excessively long dead time may also cause voltage distortion during commutation, thus affecting the IGBT's dynamic behavior and preventing it from turning off at the expected time. Therefore, the delay in the actual turn-off time of the IGBT reflects a non-optimal dead time setting, which may cause problems such as cross-conduction, current spikes, and voltage distortion, thereby affecting the stability of the inverter and the operating efficiency of the motor. Thus, the turn-off time delay is an important indicator of dead time mismatch and requires real-time monitoring and optimization to ensure the reliable operation of the inverter system.

[0084] The specific steps for analyzing the actual IGBT turn-off time deviation within the detection window to generate an IGBT turn-off delay reference value are as follows:

[0085] Within the detection window, key signals during the IGBT turn-off process, including changes in current, voltage, and power loss, are monitored using a high-speed data acquisition system. Since the IGBT turn-off process involves rapid current decay, a turn-off transient characteristic function is defined to capture the mismatch between the current decrease rate and the voltage rise rate. The formula is as follows:

[0086]

[0087] In the formula, Φ off This refers to the IGBT turn-off transient characteristics, used to measure the mismatch between the current decrease rate and the voltage rise rate during IGBT turn-off, thus reflecting the impact of dead time on the IGBT's turn-off dynamic behavior. ds It is the drain-source current when the IGBT is turned off, V ce It is the collector-emitter voltage when the IGBT is turned off, Q g It refers to the gate charge, which represents the amount of charge transferred during the charging and discharging process of the IGBT gate and determines the switching speed of the IGBT. It is the rate of current decrease during IGBT turn-off, describing the drain current I during IGBT turn-off. ds With gate charge Q g The rate of change It is the voltage rise rate during IGBT turn-off, describing the collector voltage V during the IGBT turn-off process. ce With gate charge Q g The rate of change, ρ is a weighting factor used to balance the influence of the rate of current decrease and the rate of voltage increase.

[0088] This formula measures the degree of mismatch between current and voltage changes when the IGBT is turned off. If Φ off An excessively large dead time indicates that the current decays too slowly or the voltage rises too quickly during the IGBT's turn-off process, which may indicate an unreasonable dead time.

[0089] During the turn-off process, the loss characteristics of IGBTs are affected by the dead time. When the dead time is too short, the current does not decay completely, leading to an increase in the instantaneous power peak. Therefore, an IGBT turn-off loss offset function is introduced to characterize the energy offset during the turn-off process. The calculation formula is as follows:

[0090]

[0091] In the formula, Ψ loss This is the IGBT turn-off loss offset, representing the total deviation between the actual loss and the ideal loss during the IGBT turn-off process, expressed in Ω. off It is the time region when the IGBT is turned off, P sw(ζ) is the actual measured instantaneous switching power loss of the IGBT, representing the actual instantaneous switching power loss of the IGBT at time point ζ, P ideal (ζ) is the theoretically calculated IGBT turn-off power loss (assuming perfect dead time matching), and μ is the adjustment factor, which is used to adjust the scaling factor of the theoretical loss reference. The purpose is to match the theoretical calculation value with the actual loss reference under different load, current and voltage conditions.

[0092] This step measures the deviation of the actual IGBT turn-off loss from the ideal loss. If Ψ loss If the value is too high, it indicates that there is additional energy loss during the IGBT turn-off process, which is usually caused by an excessively short or long dead time.

[0093] Finally, the transient characteristics Φ of IGBT turn-off are obtained. off and IGBT turn-off loss offset Ψ loss Based on comprehensive analysis, a reference value for IGBT turn-off delay is generated, and the formula is as follows:

[0094]

[0095] In the formula, ITDI is the IGBT turn-off delay reference value, max(Φ off ) represents the maximum IGBT turn-off transient characteristic within the detection window, max(Ψ) loss ) represents the offset of the maximum IGBT turn-off loss within the detection window, and κ1 represents the IGBT turn-off transient characteristic Φ. off The weighting factor, κ2 is the IGBT turn-off loss offset Ψ. loss Weighting factors It is a normalization of the IGBT turn-off transient characteristics. This is a normalization of the IGBT turn-off energy offset. β1 and β2 are both normalized exponential adjustment factors, with β2 used to control the normalized IGBT turn-off loss offset Ψ. loss In the calculation, β1 is used to control the normalized IGBT turn-off transient characteristic Φ. off Nonlinear effects in calculations.

[0096] The purpose of this step is to construct a normalized index that comprehensively considers the transient characteristic deviation and energy loss offset during the IGBT turn-off process. A high ITDI value indicates that the IGBT turn-off process is adversely affected by the dead time, potentially leading to commutation anomalies, cross-conduction risks, or increased high-order harmonics. Conversely, a low ITDI value indicates a good match between the IGBT turn-off time and the dead time, stable commutation, and a reasonable dead time setting.

[0097] The larger the IGBT turn-off time deviation generated after analyzing it within the detection window, the larger the IGBT turn-off delay reference value indicates a mismatch between the inverter dead time setting and actual operation; conversely, a smaller value indicates a more reasonable dead time setting that matches actual operation. The IGBT turn-off delay reference value is calculated based on the actual IGBT turn-off time deviation within the monitoring window, and its value directly reflects the impact of dead time on the IGBT turn-off dynamic behavior. A high reference value indicates a significant deviation between the IGBT turn-off time and the expected dead time, which may be caused by the following two situations:

[0098] (1) If the dead time is too short, the IGBT will be forced to turn off before the carriers are completely exhausted, which will delay the turn-off process and may be accompanied by cross-conduction.

[0099] (2) An excessively long dead time results in a lack of an effective freewheeling path during commutation, causing additional delays after the IGBT turns off due to current surges or parasitic capacitance effects. Both situations indicate that the dead time fails to accurately match the actual operating conditions, potentially leading to decreased commutation efficiency, current distortion, or increased switching losses. Conversely, a low IGBT turn-off delay reference value means that the IGBT turn-off time is basically consistent with the expected dead time, and the commutation process is stable, indicating that the current dead time setting is reasonable and matches actual operation.

[0100] The deep learning evaluation module inputs the analyzed key indicators into the pre-learned deep learning model, and uses the deep learning model to intelligently evaluate whether the dead time setting is matched from complex feature relationships;

[0101] The analyzed DC bus feedback anomaly reference value and IGBT turn-off delay reference value are input into a pre-learned deep learning model. The deep learning model generates a dead time mismatch risk coefficient, and the dead time setting is intelligently evaluated based on the dead time mismatch risk coefficient.

[0102] A pre-learned deep learning model refers to a model trained on a large amount of historical data before practical application, enabling it to intelligently assess the risk of dead-time mismatch. The training process typically includes key steps such as data acquisition, data preprocessing, feature extraction, model building, optimization training, and testing and validation. Specifically, it first requires collecting inverter operating data under different operating conditions, including DC bus voltage fluctuations in pulsed operation mode, IGBT turn-off delays, and other parameters that may affect dead-time matching. The data is then denoised, normalized, and feature-engineered to improve the quality of the training data and the robustness of the model. Next, training and testing datasets are constructed using this processed data, and a suitable deep learning model is selected for training using supervised or unsupervised learning methods. For example, a convolutional neural network (CNN) can extract features from voltage waveform data to identify subtle distortions caused by dead-time mismatch; a long short-term memory network (LSTM) can analyze trends in time-series data to capture the dynamic effects caused by dead time. Furthermore, combining Transformer or self-supervised learning models can enhance the model's generalization ability under different operating conditions. During training, evaluation metrics such as cross-entropy loss function and mean squared error (MSE) are used for optimization, and hyperparameters (such as learning rate, network depth, activation function, etc.) are continuously adjusted to ensure that the model can accurately determine dead-time matching.

[0103] Once the model is trained, it possesses the ability to intelligently assess dead-time matching. During actual operation, it can receive analyzed DC bus feedback anomaly reference values ​​and IGBT turn-off delay reference values ​​as input in real time. Through the forward propagation process of the neural network, it calculates the current dead-time mismatch risk and outputs a dead-time mismatch risk coefficient. This risk coefficient quantifies the rationality of the current dead-time setting and can be used to determine whether dead-time adjustment is needed to optimize the inverter's operating state. For example, if the risk coefficient is high (e.g., close to 1), it indicates a severe dead-time mismatch, which may lead to short circuits, increased harmonics, or excessive energy loss. In this case, the system should immediately adjust the dead-time. If the risk coefficient is low (e.g., close to 0), it indicates that the dead-time setting is reasonable and no adjustment is needed. In this way, the pre-learned deep learning model can effectively improve the intelligence level of dead-time optimization, avoid the errors caused by traditional experience-based dead-time settings, and improve the safety, stability, and efficiency of inverter operation.

[0104] The deep learning model is not limited here, but it can achieve the feedback of abnormal reference values ​​from the DC bus.

[0105] Both DBFAI and IGBT turn-off delay reference value ITDI can be used to generate a deep learning model for dead time mismatch risk coefficient DTM through comprehensive analysis. In order to realize the technical solution of this invention, this invention provides a specific implementation method.

[0106] The formula for generating the Dead Time Mismatch Risk Factor (DTM) is as follows:

[0107]

[0108] In the formula, m1 and m2 are the preset proportional coefficients of the DC bus feedback abnormal reference value DBFAI and the IGBT turn-off delay reference value ITDI, respectively, and both m1 and m2 are greater than 0.

[0109] The preset proportional coefficients m1 and m2 are adjustment parameters used to balance the impact of the DC bus feedback anomaly reference value DBFAI and the IGBT turn-off delay reference value ITDI on the dead time mismatch risk coefficient DTM. These two coefficients determine the relative contribution ratios of DBFAI and ITDI in calculating DTM, ensuring that DTM accurately reflects the comprehensive risk caused by dead time mismatch.

[0110] In practical applications, DBFAI reflects the impact of dead-time mismatch on DC bus energy feedback, while ITDI reflects the impact of dead-time mismatch on IGBT switching dynamic behavior. Different inverter topologies, load characteristics, and operating conditions may have varying sensitivities to these two factors, therefore weighted adjustments using m1 and m2 are necessary. For example:

[0111] If the system is more concerned about the impact of DC bus voltage fluctuations on inverter stability, then it can be set to...

[0112] If m1 > m2, assign a higher weight to DBFAI.

[0113] If the system is more concerned about the impact of IGBT turn-off delay on power loss and cross-conduction, then m2 > m1 can be set so that ITDI has a larger share when calculating DTM.

[0114] In addition, in the denominator The normalization factor is used to balance the calculated DTM values ​​under different operating conditions, ensuring a certain degree of scale consistency and preventing excessive amplification of the calculation results due to excessively large m1 and m2. Overall, the preset scaling factors m1 and m2 are weight parameters set according to system requirements, used to optimize DTM calculations and more accurately represent the comprehensive risk of dead-time mismatch.

[0115] As can be seen from the dead time mismatch risk coefficient, the larger the DC bus feedback abnormality reference value generated after analyzing the DC bus voltage fluctuation in pulse operation mode under the detection window, and the larger the IGBT turn-off delay reference value generated after analyzing the IGBT actual turn-off time deviation under the detection window, the larger the dead time mismatch risk coefficient generated when intelligently evaluating the dead time setting through the pre-learned deep learning model, indicating that the current dead time setting does not match the actual operation. Conversely, it indicates that the current dead time setting matches the actual operation.

[0116] The dead time mismatch risk coefficient generated when intelligently evaluating dead time settings using a pre-learned deep learning model is compared and analyzed with a preset dead time mismatch risk coefficient reference threshold to classify the dead time settings. The classification steps are as follows:

[0117] If the dead time mismatch risk coefficient is greater than the reference threshold for dead time mismatch risk coefficient, the current dead time setting will be classified as mismatched with actual operation; if the dead time mismatch risk coefficient is less than or equal to the reference threshold for dead time mismatch risk coefficient, the current dead time setting will be classified as matched with actual operation.

[0118] The fuzzy control and real-time compensation module dynamically adjusts the dead time based on the evaluation results when the evaluation results show that the dead time does not match the actual operation. This adjustment is made according to preset fuzzy rules (such as "if the dead time is insufficient and an abnormal peak current is detected, the dead time should be appropriately increased"; "if the dead time is too long and the harmonic distortion rate increases, the dead time should be shortened"). At the same time, digital signal processing (DSP) technology is used to compensate for the output error caused by the dead time in real time, correcting the output error caused by the change in dead time.

[0119] When the evaluation results show that the dead time does not match the actual operation, the fuzzy logic controller and digital signal processing (DSP) technology work together to dynamically optimize the dead time setting so that it matches the current operating conditions of the inverter, and to compensate for the voltage and current errors caused by the adjustment in real time, so as to ensure system stability, power device safety and permanent magnet motor operation quality.

[0120] First, based on the evaluation results, the fuzzy logic controller (FLC) determines the dead time adjustment strategy according to preset fuzzy rules. Since the dead time cannot be set too short (to avoid IGBT cross-conduction leading to short circuits) or too long (to avoid output voltage distortion and increased high-order harmonics), the fuzzy controller combines the dead time mismatch risk factor (DTM), harmonic distortion rate (THD), and current spike (IT) to determine the dead time adjustment strategy. spkThe system intelligently calculates the optimal amount of dead time based on key parameters such as load, power grid fluctuations, and temperature, and dynamically adjusts the dead time in real time. This adjustment method based on fuzzy control is more adaptable than a fixed dead time setting, and can maintain the optimal dead time range as external conditions such as load, power grid fluctuations, and temperature change, thereby improving system stability.

[0121] However, adjusting the dead time directly affects the PWM signal output, leading to instantaneous errors in the voltage and current waveforms of the permanent magnet motor. This can result in unstable commutation, increased torque fluctuations, and even affect the precise control of the permanent magnet motor. Therefore, after adjusting the dead time, real-time error compensation is required using digital signal processing (DSP) technology.

[0122] The DSP analyzes and calculates voltage and current deviations caused by dead time variations through high-frequency data acquisition, and performs waveform correction based on compensation algorithms (such as PI control and voltage feedforward compensation) to bring the output signal closer to the ideal state. This process ensures that even if the dead time is dynamically adjusted, the system output maintains high precision and stability, and the operation quality of the permanent magnet motor will not be affected by adjusting the dead time.

[0123] In summary, this step aims to intelligently and adaptively optimize the dead time and ensure the stability and efficiency of the permanent magnet motor drive system through DSP error compensation. It avoids problems such as short circuits, increased power loss, permanent magnet motor vibration, or harmonic pollution caused by dead time mismatch, thereby improving the system's safety, energy efficiency, and long-term reliability.

[0124] When the evaluation results show that the dead time does not match the actual operation, the fuzzy logic controller dynamically adjusts the dead time according to the preset fuzzy rules based on the evaluation results, and at the same time, uses digital signal processing technology to compensate for the output error caused by the dead time in real time. The specific steps are as follows:

[0125] During the dead time adjustment process, the first step is to use a fuzzy logic controller (FLC) to calculate the dead time adjustment amount based on the dead time mismatch risk coefficient (DTM) and the dead time mismatch risk coefficient reference threshold. The calculation expression is as follows:

[0126] ΔDT=μ f ·(α1·(DTM-TDM ref )+α2·|I spk |+α3·|THD|)

[0127] In the formula, ΔDT is the dead time adjustment amount, which determines the magnitude by which the dead time should be increased or decreased, and μ fIt is a fuzzy adjustment factor that is non-linearly scaled according to the dead time adjustment requirements to avoid over- or under-adjustment. α1 is the dead time deviation weighting factor, representing the dead time deviation (DTM-TDM). ref The weight of the effect of the adjustment amount ΔDT, TDM ref This is the reference threshold for the dead time mismatch risk factor, I. spk α1 is the current spike amplitude, reflecting the transient large current impact caused by the excessively short dead time. α2 is the current spike weighting factor, representing the weight of the current spike amplitude on the dead time adjustment. THD is the total harmonic distortion rate, representing the degree of harmonic pollution caused by the excessively long dead time, measuring the proportion of harmonic components in the output voltage or current. α3 is the total harmonic distortion weighting factor, representing the weight of the total harmonic distortion rate on the dead time adjustment.

[0128] The above steps first involve calculating the dead time error (DTM-TDM). ref To quantify the degree of mismatch in the current dead time;

[0129] Then, by combining peak current and harmonic distortion rate, the impact of mismatch on motor operation is evaluated.

[0130] Next, fuzzy logic rules are used to determine the direction of dead time adjustment:

[0131] If DTM > TDM ref And I spk If it is too high, the dead time should be increased appropriately;

[0132] If DTM > TDM ref If the THD is too high, the dead time should be reduced appropriately;

[0133] Finally, the dead time adjustment ΔDT is calculated and fed into the next step for application.

[0134] After determining the dead time adjustment ΔDT, the inverter's PWM modulation strategy is updated in real time to ensure that the dead time adjustment does not affect the overall stability. The calculation expression is as follows:

[0135]

[0136] DT new This refers to the updated dead time, which is optimized through dynamic calculation to reduce the impact of dead time mismatch on system operation. DT old It is the original dead time, e -ω·DTM This is the exponential adjustment term, an exponential scaling factor used to control the dead-time adjustment rate, where e is the natural base.

[0137] ω is the adjustment response factor, which controls the adjustment speed. It is the ripple component of the DC bus voltage, which measures the impact of dead time on bus stability. This is the rated voltage of the DC bus, used for normalization calculations of ripple impact. It is the normalized DC bus voltage ripple, which measures the degree of deviation of the DC bus voltage ripple (transient change) from the rated voltage. This is the DC bus voltage ripple weighting factor, which measures the weight of the impact of DC bus voltage ripple on dead time adjustment.

[0138] It is the load current commutation dynamic weighting factor, which measures the impact of load current commutation characteristics on dead time adjustment. It is the rate of change of load current with respect to commutation voltage, and the effect of dead time adjustment on the dynamic response of current commutation is evaluated.

[0139] The above steps first involve calculating the new dead time DT based on ΔDT. new This ensures a smooth transition during dead time adjustment;

[0140] Then, DT new It is applied to the PWM generation unit to recalculate the duty cycle of the three-phase PWM signal;

[0141] Next, the impact of the new dead time is monitored in real time. If abnormal waveform distortion occurs, the error compensation mechanism is triggered for optimization.

[0142] Finally, ensure that the new dead time can adapt to changes in system load and improve motor control accuracy.

[0143] Since dynamic adjustment of dead time affects the output voltage and current waveforms of the motor, digital signal processing (DSP) technology is needed to compensate for the error caused by dead time to ensure that the output quality is not affected. The calculation expression is as follows:

[0144]

[0145] In the formula, V comp This is the compensation voltage, used to correct voltage errors caused by dead-time adjustment, making the system output voltage approach the ideal state. θ1 is the dead-time variation weighting factor, measuring the degree of influence of the dead-time adjustment on the compensation voltage. θ2 is the voltage error weighting factor, measuring the degree of influence of the output voltage error on the compensation voltage. V ideal It is the ideal output voltage, V real This is the actual output voltage. θ3 is the second-order voltage change rate, used to analyze the dynamic impact of dead-time adjustment on voltage fluctuations; θ4 is the voltage second derivative weighting factor, measuring the degree of influence of the voltage second derivative on the compensation voltage; and θ5 is the current ripple weighting factor, measuring the degree of influence of current ripple on the compensation voltage.

[0146] It is the normalized current ripple, I ripple It is the current ripple amplitude, which measures the degree of output current oscillation caused by dead time mismatch, I nominal It is the rated current, used to normalize the contribution of current ripple to error compensation.

[0147] The above steps first involve calculating the PWM modulation error caused by the change in dead time, and then calculating the compensation amount based on the voltage error.

[0148] Then, the PWM waveform is adjusted so that the actual output voltage gradually approaches the ideal value, reducing harmonic pollution and voltage deviation caused by dead time adjustment;

[0149] Finally, the compensation effect is monitored in real time, and if harmonic distortion still exists, the compensation parameters are adjusted iteratively.

[0150] This invention, based on real-time data acquisition, deep learning intelligent evaluation, fuzzy control dynamic optimization, and DSP error compensation, constructs an adaptive dead-time control mechanism that effectively improves the stability, reliability, and energy efficiency of the permanent magnet motor system driven by oil drilling rigs. Through data acquisition and preprocessing, the integrity and accuracy of the input model are ensured, improving the precision and reliability of the evaluation. A deep learning model is used to uncover the complex relationship between the dead-time setting and the actual operating state, achieving high-precision intelligent evaluation. Combined with a fuzzy logic controller, the dead-time is adaptively adjusted under different operating conditions, avoiding the risk of shoot-through short circuits caused by excessively short dead times, while reducing voltage distortion, high-order harmonics, and energy loss caused by excessively long dead times, thus optimizing the motor's output performance. Furthermore, the DSP error compensation mechanism further corrects the impact of dead-time adjustment on the output waveform, ensuring stable voltage and current waveforms of the inverter, reducing harmonic losses, and minimizing winding overheating, thereby extending the motor's service life and improving the overall operating efficiency of the oil drilling rig system. This invention not only effectively solves the safety hazards caused by dead time mismatch, but also optimizes the performance of the motor drive system through intelligent means, significantly improving the operational stability and working efficiency of the permanent magnet motor driven by the oil drilling rig under complex working conditions.

[0151] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0152] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0153] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely for distinguishing one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0154] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0155] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0156] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0157] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0158] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0159] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0160] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A condition monitoring and control system for a permanent magnet motor driven by an oil drilling rig, characterized in that, It includes modules for data acquisition and construction, data preprocessing and feature extraction, dynamic monitoring and trend analysis, deep learning evaluation, and fuzzy control and real-time compensation. The data acquisition and construction module first acquires real-time operating data of the inverter through data acquisition equipment. The acquired real-time data are stored in time series format and a dataset is established to provide complete and accurate basic data for subsequent analysis and model evaluation. The data preprocessing and feature extraction module preprocesses the inverter operating parameter information in the dataset and extracts the core parameters that intuitively reflect the dead time mismatch from the preprocessed data, which serve as key indicators for evaluating whether the dead time is matched. The dynamic monitoring and trend analysis module performs fine-grained analysis on the extracted key indicators under the detection window to capture the dynamic change trend caused by dead time mismatch. The deep learning evaluation module inputs the analyzed key indicators into the pre-learned deep learning model, and uses the deep learning model to intelligently evaluate whether the dead time setting is matched from complex feature relationships; The fuzzy control and real-time compensation module, when the evaluation results show that the dead time does not match the actual operation, the fuzzy logic controller dynamically adjusts the dead time according to the preset fuzzy rules based on the evaluation results. At the same time, it uses digital signal processing technology to compensate for the output error caused by the dead time in real time, correcting the output error caused by the change in dead time. The core parameters that intuitively reflect the dead-time mismatch were extracted from the preprocessed data. These extracted parameters included DC bus voltage fluctuations in pulse operating mode and... The actual turn-off time deviation will include DC bus voltage fluctuations in pulse operating mode and The actual turn-off time deviation is analyzed within the detection window, generating DC bus feedback anomaly reference values ​​and... The shutdown delay reference value is fed back through the DC bus abnormal reference value and The shutdown delay reference value is used to jointly quantify the dynamic change trend caused by dead time mismatch; The specific steps for analyzing DC bus voltage fluctuations under pulse operating mode within the detection window to generate DC bus feedback anomaly reference values ​​are as follows: In pulse operating mode, the DC bus voltage signal exhibits periodic fluctuations. However, dead-time mismatch leads to increased irregularity in voltage fluctuations. Therefore, nonlinear feature extraction is first performed on the bus voltage signal to analyze its higher-order derivative cumulative offset, in order to quantify the fluctuation trend of the bus voltage. The feature extraction formula is as follows: In the formula, It is a nonlinear high-order characteristic index of DC bus voltage fluctuation, used to measure the severity of DC bus voltage fluctuation. It is the total number of time points in the detection window. It is the third derivative of the DC bus voltage. It is a point in time. The bus voltage value collected at all times. These are non-linear weighting coefficients used to adjust the influence of the third derivative term. It is the voltage fluctuation exponential decay term, calculated at the following time points. time Compared to the previous moment The voltage variation between them It is the natural base. It is an exponential decay factor; Abnormal fluctuations in bus voltage are not only manifested in sudden changes, but also lead to uneven energy feedback from the bus. Therefore, a nonlinear energy offset function is introduced to measure the non-uniformity of the bus voltage energy distribution, as shown in the following formula: In the formula, It is an index of the imbalance in DC bus energy feedback. It is a point in time. Energy storage at all times It is a point in time. Energy transferred from the bus to the load side at any given moment. This is to prevent small values ​​where the denominator is zero, ensuring computational stability. It is the index adjustment coefficient, which controls the contribution of the bus feedback imbalance to the abnormal index; The nonlinear higher-order characteristic exponent of DC bus voltage fluctuation Imbalance index of DC bus energy feedback By combining and introducing a nonlinear weight distribution function, a DC bus feedback anomaly reference value is generated, and the generation formula is as follows: In the formula, It is a nonlinear higher-order characteristic exponent of DC bus voltage fluctuation. The weighting coefficients, It is the index of the imbalance of DC bus energy feedback. The weighting coefficients, It is a nonlinear higher-order characteristic exponent of DC bus voltage fluctuation. The index adjustment factor, It is the index of the imbalance of DC bus energy feedback. The index adjustment factor, It is an exponential decay factor. It is the natural base. It is a dynamically adjustable parameter used to adjust the exponential decay factor. Sensitivity; Will The actual turn-off time deviation is analyzed and generated within the detection window. The specific steps for setting the shutdown delay reference value are as follows: Within the detection window, monitoring is conducted via a high-speed data acquisition system. Key signals during the turn-off process include changes in current, voltage, and power loss. The turn-off process involves a rapid decay of current. A turn-off transient characteristic function is defined to capture the mismatch between the rate of current decrease and the rate of voltage rise, as shown in the following formula: In the formula, yes Turn-off transient characteristics, yes Drain-source current during turn-off yes Collector-emitter voltage at turn-off It is the gate charge. yes The rate of current decrease during turn-off. yes Voltage rise rate during turn-off It is a weighting factor used to balance the influence of the current decrease rate and the voltage rise rate. During the turn-off process, its loss characteristics are affected by the dead time. When the dead time is too short, the current does not completely decay, leading to an increase in the instantaneous power peak. Therefore, a method is introduced... The turn-off loss offset function is used to characterize the energy shift during turn-off, and the calculation formula is as follows: In the formula, yes Shutdown loss offset. yes The time range of shutdown It was actually measured. Instantaneous switching power loss It is a theoretical calculation. Power loss during shutdown It is an adjustment factor, used to adjust the scaling factor of the theoretical loss benchmark; Ultimately, through Turn-off transient characteristics and Turn-off loss offset A comprehensive analysis was conducted to generate... The shutdown delay reference value is generated using the following formula: In the formula, yes Shutdown delay reference value Within the detection window, the largest Turn-off transient characteristics, Within the detection window, the largest Shutdown loss offset. yes Turn-off transient characteristics Weighting factors yes Turn-off loss offset Weighting factors yes Normalization of turn-off transient characteristics yes Normalization of the off-energy offset and All are normalized exponential adjustment factors. Used to control the normalized result Turn-off loss offset Nonlinear effects in calculations, Used to control the normalized result Turn-off transient characteristics Nonlinear effects in calculations.

2. The permanent magnet motor condition monitoring and control system based on oil drilling rig drive according to claim 1, characterized in that, The specific steps for acquiring real-time operating data of the inverter through data acquisition equipment and establishing a dataset are as follows: First, deploy high-precision data acquisition equipment in the inverter system to ensure real-time monitoring of the inverter's input, output, and the status of key components; Secondly, the acquired signals are synchronized by using a high-precision clock to timestamp and align the data from all sensors to ensure data consistency. Then, the data is stored in a time series manner through a high sampling rate data recording system to capture transient changes caused by dead time mismatch during inverter operation; Next, we will establish a dataset for merging and classification management, and label the data according to the inverter's operating conditions for subsequent analysis. Finally, by checking data integrity and removing outlier data points through interpolation compensation, the integrity and accuracy of the dataset are ensured, thus providing high-quality data support for subsequent feature analysis, deep learning evaluation, and dynamic regulation.

3. The permanent magnet motor condition monitoring and control system based on oil drilling rig drive according to claim 1, characterized in that, The analyzed DC bus feedback abnormal reference value and The shutdown delay reference value is input into a pre-learned deep learning model, which generates a dead time mismatch risk coefficient. The dead time setting is then intelligently evaluated based on the dead time mismatch risk coefficient.

4. The permanent magnet motor condition monitoring and control system based on oil drilling rig drive according to claim 3, characterized in that, The dead time mismatch risk coefficient generated when intelligently evaluating dead time settings using a pre-learned deep learning model is compared and analyzed with a preset dead time mismatch risk coefficient reference threshold to classify the dead time settings. The classification steps are as follows: If the dead time mismatch risk coefficient is greater than the reference threshold for dead time mismatch risk coefficient, the current dead time setting will be classified as mismatched with actual operation; if the dead time mismatch risk coefficient is less than or equal to the reference threshold for dead time mismatch risk coefficient, the current dead time setting will be classified as matched with actual operation.

5. The permanent magnet motor condition monitoring and control system based on oil drilling rig drive according to claim 4, characterized in that, When the evaluation results show that the dead time does not match the actual operation, the fuzzy logic controller dynamically adjusts the dead time according to the preset fuzzy rules based on the evaluation results, and at the same time, uses digital signal processing technology to compensate for the output error caused by the dead time in real time. The specific steps are as follows: During the dead time adjustment process, the first step is to utilize a fuzzy logic controller based on the dead time mismatch risk coefficient. Based on the reference threshold for the risk factor of mismatch between dead time and time, the dead time adjustment amount is calculated using the following expression: In the formula, It is the dead time adjustment amount. It is a fuzzy adjustment factor. It is the dead time deviation weighting factor. This is a reference threshold for the risk factor of dead time mismatch. It is the amplitude of the current spike. This is the current spike weighting factor, which represents the weight of the impact of the current spike amplitude on the dead time adjustment. It is the total harmonic distortion rate. It is the total harmonic distortion weighting factor, which represents the weight of the total harmonic distortion rate on the dead time adjustment. After determining the dead time adjustment amount Then, update the inverter in real time. The modulation strategy ensures that adjusting the dead time does not affect overall stability; the calculation expression is as follows: , This is the updated dead zone time. It is the original dead time. It is an exponential adjustment term, an exponential scaling factor used to control the dead time adjustment rate. It is the natural base. It involves adjusting the response factor and controlling the adjustment speed. It is the ripple component of the DC bus voltage. It is the rated voltage of the DC bus. It is the normalized DC bus voltage ripple. It is the DC bus voltage ripple weighting factor. It is the load current commutation dynamic weighting factor. It is the rate of change of load current with respect to commutation voltage; Since dynamic adjustment of the dead time affects the output voltage and current waveforms of the motor, digital signal processing technology is needed to compensate for the error caused by the dead time to ensure that the output quality is not affected. The calculation expression is as follows: In the formula, It is a compensation voltage. It is the dead time variation weighting factor. It is the voltage error weighting factor. It is the ideal output voltage. This is the actual output voltage. It is the second-order voltage change rate. It is the voltage second derivative weighting factor. It is the current ripple weighting factor, which measures the degree of influence of current ripple on the compensation voltage. It is the normalized current ripple. It is the current ripple amplitude. It is the rated current, used to normalize the contribution of current ripple to error compensation.

Images