Rotating electric machine operation support system and method

Abstract

This invention relates to the field of fault prediction and health management technology for rotating electrical machines, specifically a rotating electrical machine operation support system and method, comprising: collecting operating data of the rotating electrical machine; extracting harmonic voltage and harmonic current based on the operating data; calculating electrical stress index and mechanical stress index to generate systematic latent stress; calculating the evolution rate of systematic latent stress; generating dynamic failure boundary based on electrical stress index; calculating mode switching decision function value based on systematic latent stress, evolution rate, and dynamic failure boundary; generating control mode switching decision based on function value; generating and executing a suppressive control strategy in response to the control mode switching decision; collecting operating data after executing the suppressive control strategy; and performing closed-loop correction of the suppressive control strategy based on the updated operating data. This invention transforms complex potential risks into quantifiable unified engineering indicators, significantly improving the accuracy and robustness of risk assessment.

Description

Technical Field

[0001] This invention relates to the field of fault prediction and health management technology for rotating electric machines, specifically to a rotating electric machine operation support system and method. Background Technology

[0002] Rotating electrical machines are core equipment in the industrial and energy sectors, and their reliability and safety are of paramount importance. Under complex operating environments, rotating electrical machine systems, especially cluster systems, simultaneously bear the coupled pressure of multiple physical fields, including harmonic disturbances from the power grid and vibrations of their own structures. This interplay of electrical and mechanical stresses can induce systemic instability risks such as subsynchronous oscillations, which traditional monitoring methods struggle to address effectively.

[0003] Current technologies for monitoring the operational status of rotating electric machines typically rely on analyzing data from a single physical field or using static, fixed thresholds to trigger alarms. The limitation of this approach is that it cannot comprehensively quantify the integrated, potential system collapse risks arising from the interaction of electrical and mechanical factors. Furthermore, static thresholds cannot adapt to dynamic changes in external operating conditions, leading to missed alarms under severe conditions or false alarms under normal conditions. Therefore, current technologies often remain at the level of passively responding to faults, making it difficult to provide effective early warning and proactive avoidance in the early stages of systemic risk accumulation. Consequently, they cannot effectively prevent catastrophic equipment damage and large-scale downtime accidents, posing a potential threat to the operational reliability and safety of critical rotating electric machine systems. Summary of the Invention

[0004] To solve the above-mentioned technical problems, the present invention provides a rotating electric motor operation support system and method. Specifically, the technical solution of the present invention is as follows:

[0005] A method for supporting the operation of a rotating electric machine includes the following steps:

[0006] S1. Collect the operating data of the rotating electric motor, including three-phase current, terminal voltage and mechanical vibration acceleration; and extract harmonic voltage and harmonic current based on the operating data.

[0007] S2. Calculate the electrical stress index and the mechanical stress index. The electrical stress index is determined based on harmonic voltage and harmonic current, and the mechanical stress index is determined based on mechanical vibration acceleration. Combine the electrical stress index and the mechanical stress index to generate the systemic latent stress.

[0008] S3. Calculate the evolution rate of the systemic latent stress; and generate dynamic failure boundaries based on the electrical stress index.

[0009] S4. Based on the systemic latent stress, evolution rate and dynamic failure boundary, calculate the mode switching decision function value, and generate control mode switching decision based on the function value;

[0010] S5. In response to the control mode switching decision, generate and execute an inhibitory control strategy;

[0011] S6. Collect the operational data after the implementation of the inhibitory control strategy, and based on the updated operational data, make closed-loop corrections to the inhibitory control strategy.

[0012] Preferably, S2 specifically includes:

[0013] S21. Compare the monitored subsynchronous harmonic power with the rated power of the motor, and calculate the electrical stress index by normalization.

[0014] S22. Compare the accumulated vibration energy over a period of time with the preset reference energy value, and calculate the mechanical stress index by normalization.

[0015] S23. The electrical stress index and the mechanical stress index are treated as orthogonal components in the two-dimensional risk space and fused through vector synthesis logic to generate systemic latent stress.

[0016] Preferably, S3 specifically includes:

[0017] S31. Store the real-time calculated systemic latent stress sequence into a time window buffer, and calculate the evolution rate by applying the difference method or Kalman filter algorithm to the time series data.

[0018] S32. Using the electrical stress index as a quantitative indicator of the current harmonic interference intensity, the baseline boundary is adjusted exponentially to generate a dynamic failure boundary.

[0019] Preferably, the method further includes:

[0020] Based on the calculated systemic latent stress, evolution rate, and dynamic failure boundary, risk level assessment rules are established, classifying risks into concern level and alarm level.

[0021] Preferably, S4 specifically includes:

[0022] S41. Construct a mode switching decision function. The mode switching decision function integrates the normalized risk degree and the normalized risk rate to calculate the mode switching decision function value.

[0023] S42. Compare the mode switching decision function value with the preset switching threshold and hysteresis threshold to generate a control mode switching decision. When the mode switching decision function value exceeds the switching threshold, generate a decision to switch to the survival optimal mode; when the mode switching decision function value is lower than the hysteresis threshold, generate a decision to switch back to the performance optimal mode; when the mode switching decision function value is between the switching threshold and the hysteresis threshold, maintain the current control mode.

[0024] Preferably, S5 specifically includes:

[0025] S51. After switching to the survival-optimal mode, an autoregressive prediction model is constructed online that correlates the active power fine-tuning and reactive power fine-tuning with the future changes in the systemic latent stress.

[0026] S52. Based on the autoregressive prediction model, with the goal of minimizing the predicted systemic latent stress, an inhibitory control strategy including the optimal active power fine-tuning and reactive power fine-tuning is calculated.

[0027] Preferably, S6 specifically includes:

[0028] S61. Collect the new system state after implementing the inhibitory control strategy, and recalculate the real system latent stress as a feedback signal.

[0029] S62. Using feedback signals, the parameters of the autoregressive prediction model are corrected online through a recursive least squares algorithm.

[0030] A rotating electric motor operation support system includes the following modules:

[0031] The data acquisition and preprocessing module is used to acquire the operating data of the rotating motor, including three-phase current, terminal voltage and mechanical vibration acceleration, and to extract harmonic voltage and harmonic current based on the operating data.

[0032] The latent stress estimation module is used to calculate the electrical stress index and the mechanical stress index. The electrical stress index is determined based on harmonic voltage and harmonic current, and the mechanical stress index is determined based on mechanical vibration acceleration. The module combines the electrical stress index and the mechanical stress index to generate the systemic latent stress.

[0033] The risk trend assessment module is used to calculate the evolution rate of systemic latent stress and generate dynamic failure boundaries based on the electrical stress index.

[0034] The mode switching decision module is used to calculate the mode switching decision function value based on the systemic latent stress, evolution rate and dynamic failure boundary, and generate control mode switching decisions based on the function value.

[0035] The suppression strategy generation and execution module is used to generate and execute suppression control strategies in response to control mode switching decisions.

[0036] The closed-loop correction module is used to collect the operating data after the execution of the inhibitory control strategy, and to correct the inhibitory control strategy in a closed loop based on the updated operating data.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. A new paradigm for risk quantification through multi-dimensional integration: This technology integrates multi-physical field data such as three-phase current, terminal voltage, and mechanical vibration to construct electrical and mechanical stress indices and synthesize them into a single systemic latent stress index. This method transforms complex potential risks into quantifiable unified engineering indicators, accurately captures system bottleneck effects, and significantly improves the accuracy and robustness of risk assessment.

[0039] 2. Proactive early warning and dynamic failure boundary: This method not only assesses the current risk status, but also calculates its evolution rate to achieve proactive early warning. Its core is to establish a dynamic failure boundary that adapts to the intensity of power grid harmonics, replacing the fixed threshold. This mechanism can intelligently adapt to operating conditions, reduce false alarms in harsh environments and avoid false alarms in favorable environments, thus improving the sensitivity and reliability of early warning.

[0040] 3. Intelligent decision-making and predictive suppression control: This method constructs a decision function that comprehensively considers latent stress, its evolution rate, and dynamic boundaries, autonomously switching between optimal performance and survival modes. After switching, the system establishes a predictive model and actively suppresses risk growth by fine-tuning active and reactive power. This feedforward predictive control paradigm can more quickly and directly ensure the physical survival of the motor.

[0041] 4. Adaptive closed-loop correction and long-term robustness: After implementing inhibitory control, this system collects new data to recalculate latent stress and uses it as a feedback signal to correct the parameters of the prediction model online through a recursive least squares algorithm. This continuous self-learning and closed-loop calibration mechanism endows the system with strong adaptive capabilities, ensuring that the control strategy remains accurate and effective in long-term operation. Attached Figure Description

[0042] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0043] Figure 1 This is a flowchart of the method of the present invention;

[0044] Figure 2 This is a structural block diagram of the system of the present invention. Detailed Implementation

[0045] 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.

[0046] Example 1:

[0047] Please see Figure 1 A method for supporting the operation of a rotating electric motor includes the following steps:

[0048] S1. Collect the operating data of the rotating electric motor, including three-phase current, terminal voltage and mechanical vibration acceleration; and extract harmonic voltage and harmonic current based on the operating data.

[0049] S2. Calculate the electrical stress index and the mechanical stress index. The electrical stress index is determined based on harmonic voltage and harmonic current, and the mechanical stress index is determined based on mechanical vibration acceleration. Combine the electrical stress index and the mechanical stress index to generate the systemic latent stress.

[0050] S3. Calculate the evolution rate of the systemic latent stress; and generate dynamic failure boundaries based on the electrical stress index.

[0051] S4. Based on the systemic latent stress, evolution rate and dynamic failure boundary, calculate the mode switching decision function value, and generate control mode switching decision based on the function value;

[0052] S5. In response to the control mode switching decision, generate and execute an inhibitory control strategy;

[0053] S6. Collect the operational data after the implementation of the inhibitory control strategy, and based on the updated operational data, make closed-loop corrections to the inhibitory control strategy.

[0054] This embodiment provides a method for supporting the operation of rotating electric machines, which aims to achieve a shift from passive response to proactive avoidance of systemic collapse risks through in-depth analysis and forward-looking prediction of the operating status of rotating electric machine clusters. The method constructs a complete technical closed loop from data acquisition, risk quantification, trend prediction, intelligent decision-making to closed-loop control.

[0055] In a specific application scenario, this method is applied to a rotating motor cluster in a wind farm. The process begins with S1, collecting the operating data of the rotating motors, and extracting harmonic voltage and harmonic current based on the operating data. The operating data refers to a multi-physics data stream that comprehensively reflects the electrical and mechanical characteristics of the motors, and its role is to provide raw, high-fidelity information input for subsequent risk assessment. In this embodiment, a high-speed synchronous acquisition unit is deployed in the motor control system to acquire three types of core data in real time: stator three-phase current acquired by Hall sensors or CT / PT. and terminal voltage Mechanical vibration acceleration collected by piezoelectric accelerometers attached to key locations on the casing. ; and the real-time motor speed ω and rated power obtained by the encoder and power analysis module. The raw current collected With voltage The signal is fed into a digital signal processing unit, the core purpose of which is to accurately separate specific frequency components highly correlated with system instability from the broadband signal. In this embodiment, the harmonic voltage within the 10-40Hz subsynchronous frequency band is extracted from the signal by performing a Fast Fourier Transform (FFT) algorithm. With harmonic current This provides direct input for subsequent electrical stress assessment;

[0056] Based on the collected data, step S2 is executed to calculate the electrical stress index and the mechanical stress index. The electrical stress index and the mechanical stress index are then combined to generate the systemic latent stress. The purpose of this step is to transform the unobservable, cross-domain system collapse risk into a unified and quantifiable engineering indicator. The electrical stress index is used to quantify the internal pressure of the electrical system caused by power grid harmonic disturbances. The mechanical stress index is used to quantify the internal fatigue and damage accumulation of the mechanical system caused by structural vibration. The systemic latent stress is a custom comprehensive risk indicator, the purpose of which is to integrate the risks of the electrical and mechanical dimensions to form a top-level assessment of the overall health status of the system.

[0057] S3. Calculate the evolution rate of the systemic latent stress and generate a dynamic failure boundary based on the electrical stress index. This step aims to elevate risk assessment from a static current value to a dynamic future trend level, enabling proactive early warning. The evolution rate aims to capture the speed at which risk deteriorates. A high level of latent stress accompanied by a high growth rate means that the system is accelerating towards the edge of instability. The purpose of the dynamic failure boundary is to establish a safety red line that can adaptively adjust with the intensity of external disturbances, rather than a fixed, universal threshold.

[0058] Based on the results of the preceding assessment, step S4 calculates the mode switching decision function value based on the systemic latent stress, evolution rate, and dynamic failure boundary, and generates a control mode switching decision based on the function value. This step is the core of the decision-making mechanism of this invention. Its purpose is to autonomously decide whether the system should switch between the two operating modes of optimal performance and optimal survival based on the accurate risk profile obtained from the preceding steps. The mode switching decision function value is a quantitative indicator that integrates the current risk level, deterioration rate, and current system carrying capacity, and is used to determine the urgency and necessity of performing mode switching.

[0059] Upon entering the control execution phase, S5 responds to the control mode switching decision by generating and executing a suppressive control strategy. When the decision module issues an instruction to switch to the survival-optimal mode, the purpose of this step is to immediately take proactive intervention measures. Its core objective is no longer to pursue performance indicators such as power generation efficiency, but to directly suppress the growth of systemic latent stress and ensure the physical survival of the motor system.

[0060] To ensure the robustness and adaptability of the system, S6 collects the operating data after the execution of the inhibitory control strategy, and corrects the inhibitory control strategy in a closed loop based on the updated operating data; this step constructs a complete feedback correction closed loop; its purpose is to continuously and online optimize the control model itself by monitoring the actual effect after the execution of the control command, so as to cope with the time-varying characteristics of the system and unknown disturbances in the external environment.

[0061] This invention provides a complete, closed-loop method for supporting the operation of rotating electric machines. It goes beyond simply monitoring and alarming faults; by constructing a complete process of risk quantification, trend prediction, dynamic decision-making, proactive control, and closed-loop correction, it achieves proactive management and avoidance of systemic collapse risks. Compared to existing technologies that rely on static thresholds or single physical field data, this invention can identify instability precursors earlier and more accurately, and execute risk hedging strategies at the expense of short-term performance, thereby effectively avoiding catastrophic equipment damage and large-scale downtime accidents, and significantly improving the operational reliability and safety of critical rotating electric machine systems.

[0062] The core of this method is the construction of a two-dimensional risk space with electrical and mechanical stress as the main dimensions. This is particularly suitable for monitoring and mitigating systemic instability risks caused by electromechanical interactions such as subsynchronous oscillations. For failure modes dominated by other factors such as thermal stress and slow insulation aging, it may be necessary to introduce additional monitoring dimensions to extend this model.

[0063] Example 2:

[0064] S2 specifically includes:

[0065] S21. Compare the monitored subsynchronous harmonic power with the rated power of the motor, and calculate the electrical stress index by normalization.

[0066] S22. Compare the accumulated vibration energy over a period of time with the preset reference energy value, and calculate the mechanical stress index by normalization.

[0067] S23. The electrical stress index and the mechanical stress index are treated as orthogonal components in the two-dimensional risk space and fused through vector synthesis logic to generate systemic latent stress.

[0068] This embodiment is a detailed explanation of step S2, illustrating the systematic latent stress in detail. The construction process of this system is the technological cornerstone of the entire risk assessment system.

[0069] S21. Compare the monitored subsynchronous harmonic power with the rated power of the motor, and calculate the electrical stress index using normalization; Electrical stress index This refers to the quantified electrical dimension risk, which is used to assess the impact of harmful energy caused by subsynchronous oscillations on the system. In this embodiment, its calculation formula is as follows:

[0070]

[0071] in, This represents the set of harmonic orders identified within the 10-40Hz subsynchronous frequency band;

[0072] Harmonic voltage, measured in volts (V), originates from the voltage across the terminals in step S1. Extracted by FFT analysis;

[0073] Harmonic current, measured in amperes (A), originates from the three-phase current analysis in step S1. Extracted by FFT analysis;

[0074] Rated power, measured in watts (W), is derived from the motor's inherent nameplate parameters.

[0075] The underlying logic of this formula is that it compares a physical quantity directly related to risk with a benchmark quantity characterizing the motor's load-bearing capacity, thereby achieving a dimensionless, normally defined electrical risk measure with clear physical meaning.

[0076] S22. Compare the accumulated vibration energy over a period of time with a preset reference energy value, and calculate the mechanical stress index using normalization; Mechanical stress index This refers to the quantified mechanical dimension risk, which is used to assess the cumulative fatigue and damage level of a structure under continuous vibration. In this embodiment, its calculation formula is as follows:

[0077]

[0078] in, The start time of the integration time window. Represents the cumulative time of vibrational energy;

[0079] Vibration acceleration, in meters per second² The source of this data is the real-time data collected by the piezoelectric accelerometer in step S1.

[0080] Reference acceleration, in meters per second² The acceleration values ​​are set according to the acceptable vibration limits for similar rotating equipment as specified in the international standard ISO 10816, ensuring the objectivity and standardization of the assessment.

[0081] Reference time, in seconds (s), is derived from taking 1 second as a standardized time base.

[0082] The underlying logic of this formula is that it compares the square integral of the acceleration that is proportional to the vibration energy over a period of time with a benchmark energy value that represents an acceptable level of vibration, thereby normalizing the mechanical risk.

[0083] S23. The electrical stress index and the mechanical stress index are treated as orthogonal components in a two-dimensional risk space and fused using vector synthesis logic to generate a systemic latent stress; systemic latent stress This refers to a single top-level indicator that integrates electrical and mechanical risks, serving as a core basis for all subsequent decisions; in this embodiment, its calculation formula is as follows:

[0084]

[0085] in, Electrical stress index, dimensionless, is obtained from the calculation in step S21;

[0086] Mechanical stress index, dimensionless, is obtained from the calculation in step S22;

[0087] The weighting coefficients are dimensionless, adjustable parameters. To clarify the calibration process, it is hereby stated that the variables in this calibration process are independent of the variables during model runtime. The calibration method is as follows: construct a calibration dataset containing N sets of simulation scenarios. Each of the scenes It contains complete simulation data of the motor under specific operating conditions until a preset fault occurs, and records the exact time of fault occurrence. Define an evaluation function. ,in This represents an extremely short pre-fault time window; its length should be determined based on the typical evolution duration of fault symptoms in historical data, for example, it can be taken as 1 to 5 data sampling periods before the fault occurs, and solved through an optimization algorithm. Thus, we can obtain the ability to The optimal weight combination that achieves maximum discrimination before the known fault is determined is typically optimized by imposing constraints. And can be increased as needed. Normalization constraints;

[0088] It treats electrical and mechanical stresses as orthogonal vectors in an abstract two-dimensional risk space, while This is the magnitude of the combined risk vector in this space; this vector synthesis structure ensures that a surge in risk in any single dimension will significantly increase the total risk value, effectively capturing the weakest link effect;

[0089] Compared to a general description of risk calculation, the specific calculation method provided in this embodiment has significant technical advantages; by introducing normalization and data-driven weight calibration, it accurately and robustly quantifies the originally vague and multi-dimensional risk into a single indicator with clear physical meaning. This quantitative approach not only improves the accuracy of risk assessment, but also provides high-quality and reliable input for subsequent trend prediction, dynamic boundary setting, and control decisions, which is the fundamental guarantee for the effectiveness of the entire support system.

[0090] Example 3:

[0091] S3 specifically includes:

[0092] S31. Store the real-time calculated systemic latent stress sequence into a time window buffer, and calculate the evolution rate by applying the difference method or Kalman filter algorithm to the time series data.

[0093] S32. Using the electrical stress index as a quantitative indicator of the current harmonic interference intensity, the baseline boundary is adjusted exponentially to generate a dynamic failure boundary.

[0094] It also includes: establishing risk level assessment rules based on the calculated systemic latent stress, evolution rate and dynamic failure boundary, and classifying risks into concern level and alarm level.

[0095] This embodiment is a detailed explanation of step S3, and further introduces risk level assessment rules, aiming to build a forward-looking and adaptive risk warning system;

[0096] S31. Store the real-time calculated systemic latent stress sequence in a time window buffer. Calculate the evolution rate by applying the difference method or Kalman filter algorithm to the time series data; evolution rate Systemic latent stress The rate of change of risk is used to assess the speed of risk deterioration from a dynamic trend perspective. In this embodiment, the system maintains a fixed-length time window buffer to store continuously calculated data. Sequence; by applying the first-order backward difference method to this time series data ( Alternatively, a smoother Kalman filter algorithm can be used to estimate the current evolution rate in real time. This design expands the assessment perspective from how high the risk is to how fast the risk is growing, providing crucial dynamic information for predicting the system's state.

[0097] S32. Using the electrical stress index as a quantitative indicator of the current harmonic interference intensity, the baseline boundary is adjusted exponentially to generate a dynamic failure boundary; dynamic failure boundary This refers to the upper limit of the system's ability to withstand latent stress. Its function is to dynamically adjust the early warning threshold based on the severity of the current power grid environment. In this embodiment, its calculation formula is as follows:

[0098]

[0099] in, The baseline boundary is dimensionless and is derived from statistical analysis of a large amount of historical normal operating data of the motor system, taking the calculated value. The 99.9th percentile of the value is used as the critical value for instability under ideal working conditions;

[0100] The interference sensitivity coefficient is dimensionless and is derived from frequency sweep analysis on the system model, i.e., injecting harmonic disturbances of different frequencies and amplitudes into the observed system. The response changes were observed, and the optimal parameter values ​​were obtained by fitting these experimental data points through nonlinear regression analysis.

[0101] Electrical stress index, dimensionless, is obtained in real time from step S2;

[0102] The innovation of this formula lies in its use of the already calculated electrical stress index. As a real-time quantitative indicator of the current harmonic interference intensity; when power grid quality deteriorates and harmonic interference intensifies, Increase It will decrease rapidly in an exponential manner, which means that the system's safety margin is compressed and the warning threshold is lowered accordingly;

[0103] After calculating three core parameters—latent stress Evolution rate With dynamic failure boundary Subsequently, this embodiment establishes risk level assessment rules based on these parameters, classifying risks into concern level and alarm level;

[0104] Level 1 Warning: Triggering conditions are Threshold coefficient The determination method is based on historical data analysis, for example, through analysis of a large amount of normal operation data and near-failure data. The distribution was statistically analyzed, and the Elbow method inflection point detection algorithm was applied to determine the critical quantile where the system risk began to deteriorate significantly, and this quantile was set as... The value;

[0105] Level 2 warning: Triggering conditions are and Threshold coefficient The setting is based on the fact that the system is extremely close to the point where historical failures occurred. Statistical values; rate threshold It is a preset positive value, derived from historical fault data, from the onset of fault precursors to the instability stage. The average value represents a typical rate of dangerous growth;

[0106] Through this embodiment, the system achieves a leap from static threshold alarms to dynamic, predictive, and tiered early warnings; dynamic failure boundaries. The introduction of this technology enables the early warning system to intelligently adapt to changing power grid environments, significantly reducing the false alarm rate under adverse operating conditions and the false alarm rate under favorable operating conditions; evolution rate The addition of this feature endows the system with predictive capabilities, enabling it to issue early warnings before risks reach critical thresholds. The tiered early warning mechanism provides more refined decision support for operations and maintenance personnel, enabling differentiated management of risks with different levels of urgency and significantly improving the effectiveness and foresight of early warnings.

[0107] Example 4:

[0108] S4 specifically includes:

[0109] S41. Construct a mode switching decision function. The mode switching decision function integrates the normalized risk degree and the normalized risk rate to calculate the mode switching decision function value.

[0110] S42. Compare the mode switching decision function value with the preset switching threshold and hysteresis threshold to generate a control mode switching decision. When the mode switching decision function value exceeds the switching threshold, generate a decision to switch to the survival optimal mode; when the mode switching decision function value is lower than the hysteresis threshold, generate a decision to switch back to the performance optimal mode; when the mode switching decision function value is between the switching threshold and the hysteresis threshold, maintain the current control mode.

[0111] This embodiment is a detailed explanation of step S4, illustrating in detail how the system makes intelligent switching decisions on control modes based on the risk assessment results.

[0112] S41. Construct a mode-switching decision function. This function combines the normalized risk level and the normalized risk rate to calculate the mode-switching decision function value. The purpose is to create a single, quantifiable metric to determine the urgency and necessity of switching from the optimal performance mode to the optimal survival mode; in this embodiment, the function is designed as follows:

[0113]

[0114] in, The normalized risk level represents the relative position of the current latent stress from the dynamic failure boundary. Squaring it is done to amplify the sensitivity of the high-risk region, so that the response of the decision function is more drastic when the system approaches the boundary. It is derived from the calculation results of the previous steps.

[0115] Evolution rate, which is derived from the calculation results of the preceding steps;

[0116] Reference rate, unit and Consistency is determined by extracting the initial stages of all typical instability faults from the historical fault database. The growth rate data is used, and the 75th percentile of these data is taken to ensure that the reference value represents a significant and dangerous growth trend; the function design has inherent robustness. As can be seen from the definition of dynamic failure boundary, The value remains positive, avoiding the risk of the denominator being zero. When the system is under extreme operating conditions, electrical stress... Extremely large, leading to Approaching zero, the risk of normalization at this point It will increase sharply, causing the decision function value to... The switching threshold is rapidly exceeded, triggering protection; this behavior aligns with physical principles and safety requirements; reference rate As a statistical value for historical data, it is set to a normal number during system initialization.

[0117] S42. Compare the mode switching decision function value with the preset switching threshold and hysteresis threshold to generate a control mode switching decision; the purpose of this step is based on The value determines the execution of a stable, jitter-free switching logic;

[0118] Switch to survival-optimal mode: when the decision function value Exceeding the preset switching threshold For example The system immediately generates a decision to switch to the optimal survival mode; The settings are linked to the secondary warning conditions and fine-tuned through simulation tests to achieve a balance between sensitivity and stability;

[0119] Switching back to optimal performance mode: To avoid frequent switching of the system near the critical point, a hysteresis threshold is introduced. ,For example ,and Once the system enters its optimal survival mode, it can only survive if the risk is effectively suppressed. The system will only generate a decision to switch back to the optimal performance mode when the value is continuously lower than the hysteresis threshold for a period of time.

[0120] Maintain the current control mode: when the decision function value Between the switching threshold Hysteresis threshold During this period, the system will maintain the current control mode.

[0121] The decision function designed in detail in this embodiment Its switching logic, compared to simple threshold judgment, has significant advantages. It not only considers both the current value and trend of the risk, but also achieves more sensitive and robust judgment of dangerous states through normalization and square amplification. The introduced hysteresis threshold mechanism follows mature design principles in control engineering, ensuring the stability and reliability of mode switching, avoiding unnecessary oscillations in the control system, and ensuring a smooth transition of the entire motor system under critical conditions. To ensure the robustness of decision-making, the following aspects should also be considered in the actual deployment of this system: establishing a sensor fault detection mechanism to verify the validity of input signals and avoid misjudgment of risk due to data errors; and adjusting the reference rate... Parameters based on historical data should be given reasonable non-zero initial values, and a procedural division-to-zero protection should be added; decision function. When the value is negative, it indicates that the systemic latent stress is decreasing, which is a favorable condition for triggering the decision to switch back to the optimal performance mode.

[0122] Example 5:

[0123] S5 specifically includes:

[0124] S51. After switching to the survival-optimal mode, an autoregressive prediction model is constructed online that correlates the active power fine-tuning and reactive power fine-tuning with the future changes in the systemic latent stress.

[0125] S52. Based on the autoregressive prediction model, with the goal of minimizing the predicted systemic latent stress, an inhibitory control strategy including the optimal active power fine-tuning and reactive power fine-tuning is calculated.

[0126] This embodiment is a detailed explanation of step S5, describing in detail how the system generates and executes a control strategy with the core objective of mitigating risk after switching to the survival-optimal mode.

[0127] S51. After switching to the survival-optimal mode, an autoregressive prediction model is constructed online that correlates the active power fine-tuning and reactive power fine-tuning with the future changes in the system's latent stress. The purpose of this step is to establish a mathematical model that can predict the relationship between control input and control target. In this embodiment, the autoregressive prediction model is designed as a linear structure that is easy to solve quickly, and its specific form is as follows:

[0128]

[0129] in, The predicted value of latent stress is the target output of the model;

[0130] and The measured value at the current moment is the known input of the model, and its source is provided in real time by the previous steps;

[0131] The system's control cycle or data sampling cycle, measured in seconds (s);

[0132] and Active power fine-tuning and reactive power fine-tuning are the control variables of the system.

[0133] and The sensitivity coefficient is a key parameter of the model, with units of 1 / W. They represent the effect of unit active and reactive power adjustments on the system's latent stress, respectively. The magnitude of the direct impact ensures the formula The dimensions at both ends are unified; the source of these two coefficients is online identification and continuous updating through subsequent closed-loop correction steps;

[0134] S52. Based on the autoregressive prediction model, with the objective of minimizing the predicted systemic latent stress, a suppressive control strategy including optimal active power fine-tuning and reactive power fine-tuning is calculated. The purpose of this step is to solve for the optimal control action within the current control cycle. The objective function of the control is explicitly defined as minimizing the predicted value of the latent stress, i.e., Minimize. Substituting the above prediction model into the objective function, the optimization problem is transformed into solving a function that enables... The smallest item Combinations; those skilled in the art can efficiently obtain the current optimal combination of inhibitory controls through gradient descent or direct solution. This combination is then sent to the underlying frequency converter or excitation controller of the motor for execution.

[0135] The inhibitory control strategy generation method disclosed in this embodiment is one of the core innovations of this invention; it abandons the traditional control logic that aims to track performance instructions, and instead creates a system that directly inhibits abstract risk indicators. This new control paradigm aims to achieve feedforward and predictive risk suppression by constructing an online linear prediction model and performing optimization based on this model. Compared to traditional feedback control, this method offers a faster response and more direct and accurate suppression effect. It should be noted that this linear prediction model is a locally linearized approximation of a complex nonlinear motor system, designed to achieve fast, online optimization. The model's accuracy depends on the closed-loop correction module's adjustment of parameters. and Rapid tracking and updates are a necessary design trade-off for computational efficiency.

[0136] Example 6:

[0137] S6 specifically includes:

[0138] S61. Collect the new system state after implementing the inhibitory control strategy, and recalculate the real system latent stress as a feedback signal.

[0139] S62. Using feedback signals, the parameters of the autoregressive prediction model are corrected online through a recursive least squares algorithm.

[0140] This embodiment is a detailed explanation of step S6, illustrating in detail the implementation of the closed-loop correction mechanism, which is key to ensuring the long-term robustness of the system.

[0141] S61. Collect the new system state after implementing the inhibitory control strategy, and recalculate the actual system latent stress as a feedback signal; in a control command In the next sampling cycle after execution, the system will acquire a new set of operational data through the data acquisition module in step S1; based on this new data, the system will obtain a true systemic latent stress after control execution through the calculation process in step S2. This feedback signal It is a direct measure of the actual effect of the control strategy in the previous step;

[0142] S62. Using feedback signals, the parameters of the autoregressive prediction model are corrected online using a recursive least squares algorithm. The purpose of this step is to continuously optimize the prediction model using the latest input-output data pairs. Specifically, the system has the data pairs from the previous period: Input: Output: This new data pair is then fed into a Recursive Least Squares (RLS) algorithm module. The RLS algorithm is a mature online parameter identification method that can efficiently update the sensitivity coefficients in the prediction model based on the latest data. and This continuous online correction process effectively resists model parameter drift caused by factors such as motor aging and changes in operating conditions, ensuring the accuracy of the prediction model.

[0143] The closed-loop correction method described in this embodiment endows the entire control system with strong adaptive capabilities. By using the actual state after control as feedback and applying the RLS algorithm to correct the parameters of the core prediction model online, the system can achieve self-learning and self-calibration. This design ensures that even if the system characteristics change slowly or there are initial model errors, the inhibitory control strategy can still maintain long-term accuracy and effectiveness, demonstrating high robustness.

[0144] Example 7:

[0145] Please see Figure 2 A rotating electric motor operation support system includes the following modules:

[0146] The data acquisition and preprocessing module is used to acquire the operating data of the rotating motor, including three-phase current, terminal voltage and mechanical vibration acceleration, and to extract harmonic voltage and harmonic current based on the operating data.

[0147] The latent stress estimation module is used to calculate the electrical stress index and the mechanical stress index. The electrical stress index is determined based on harmonic voltage and harmonic current, and the mechanical stress index is determined based on mechanical vibration acceleration. The module combines the electrical stress index and the mechanical stress index to generate the systemic latent stress.

[0148] The risk trend assessment module is used to calculate the evolution rate of systemic latent stress and generate dynamic failure boundaries based on the electrical stress index.

[0149] The mode switching decision module is used to calculate the mode switching decision function value based on the systemic latent stress, evolution rate and dynamic failure boundary, and generate control mode switching decisions based on the function value.

[0150] The suppression strategy generation and execution module is used to generate and execute suppression control strategies in response to control mode switching decisions.

[0151] The closed-loop correction module is used to collect the operating data after the execution of the inhibitory control strategy, and to correct the inhibitory control strategy in a closed loop based on the updated operating data.

[0152] This embodiment provides a rotating electric motor operation support system, which is typically deployed in an industrial computer or embedded controller and includes the following functional modules:

[0153] Data acquisition and preprocessing module: Its purpose is to provide real-time and accurate data input for the entire system; the hardware of this module includes Hall sensors, CT / PT, piezoelectric accelerometers, etc., and the software includes a high-speed synchronous data acquisition card and signal processing program; its specific function is step S1.

[0154] Latent stress estimation module: Its purpose is to transform multi-source, heterogeneous raw data into a unified, quantitative top-level risk indicator; this module is a core computing unit, and its specific function is to execute the S2 step;

[0155] Risk Trend Assessment Module: Its purpose is to achieve a forward-looking and dynamic assessment of risks; this module is responsible for executing the S3 and risk rating steps;

[0156] Mode switching decision module: As the decision-making center of the system, its purpose is to make key operating mode switching decisions; this module is responsible for executing step S4.

[0157] Suppression Strategy Generation and Execution Module: As the system's execution unit, its purpose is to execute proactive risk avoidance actions; this module is responsible for executing step S5.

[0158] Closed-loop correction module: Its purpose is to endow the system with adaptive and self-learning capabilities, ensuring robustness in long-term operation; this module is responsible for executing the S6 steps;

[0159] Through the organic combination and collaborative work of the above modules, this system constitutes a highly integrated and intelligent operation support system. It can autonomously complete the entire process from state perception, risk quantification, trend prediction, intelligent decision-making to closed-loop control, thereby proactively and effectively ensuring the physical survivability of rotating motor clusters under complex working conditions, representing an important technological advancement in the field of equipment health management.

[0160] 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 rotating electric machine operation assistance method characterized by comprising: Includes the following steps: S1. Collect the operating data of the rotating electric motor, including three-phase current, terminal voltage and mechanical vibration acceleration; and extract harmonic voltage and harmonic current based on the operating data. S2. Calculate the electrical stress index and the mechanical stress index. The electrical stress index is determined based on harmonic voltage and harmonic current, and the mechanical stress index is determined based on mechanical vibration acceleration. By combining the electrical stress index and the mechanical stress index, a systematic latent stress is generated; S3. Calculate the evolution rate of the systemic latent stress; and generate dynamic failure boundaries based on the electrical stress index. S4. Based on the systemic latent stress, evolution rate and dynamic failure boundary, calculate the mode switching decision function value, and generate control mode switching decision based on the function value; S5. In response to the control mode switching decision, generate and execute an inhibitory control strategy; S6. Collect the operational data after the execution of the inhibitory control strategy, and based on the updated operational data, make closed-loop corrections to the inhibitory control strategy. S2 specifically includes: S21. Compare the monitored subsynchronous harmonic power with the rated power of the motor, and calculate the electrical stress index by normalization. S22. Compare the cumulative vibration energy over a period of time with the preset reference energy value, and calculate the mechanical stress index by normalization. S23. The electrical stress index and the mechanical stress index are treated as orthogonal components in the two-dimensional risk space and fused through vector synthesis logic to generate systematic latent stress. S3 specifically includes: S31. Store the real-time calculated systemic latent stress sequence into a time window buffer, and calculate the evolution rate by applying the difference method or Kalman filter algorithm to the time series data. S32. Using the electrical stress index as a quantitative indicator of the current harmonic interference intensity, the baseline boundary is adjusted exponentially to generate a dynamic failure boundary. S4 specifically includes: S41. Construct a mode switching decision function. The mode switching decision function integrates the normalized risk degree and the normalized risk rate to calculate the mode switching decision function value. S42. Compare the mode switching decision function value with the preset switching threshold and hysteresis threshold to generate a control mode switching decision. When the mode switching decision function value exceeds the switching threshold, generate a decision to switch to the survival optimal mode; when the mode switching decision function value is lower than the hysteresis threshold, generate a decision to switch back to the performance optimal mode; when the mode switching decision function value is between the switching threshold and the hysteresis threshold, maintain the current control mode.

2. The rotating electric machine operation assist method according to claim 1, characterized by Also includes: Based on the calculated systemic latent stress, evolution rate, and dynamic failure boundary, risk level assessment rules are established, classifying risks into concern level and alarm level.

3. The rotating electric machine operation assist method according to claim 1, characterized by S5 specifically includes: S51. After switching to the survival-optimal mode, construct an autoregressive prediction model online that correlates the active power fine-tuning and reactive power fine-tuning with the future changes in the systemic latent stress. S52. Based on the autoregressive prediction model, with the goal of minimizing the predicted systemic latent stress, an inhibitory control strategy including the optimal active power fine-tuning and reactive power fine-tuning is calculated.

4. The rotating electric machine operation assist method according to claim 1, characterized by S6 specifically includes: S61. Collect the new system state after implementing the inhibitory control strategy, and recalculate the real system latent stress as a feedback signal. S62. Using feedback signals, the parameters of the autoregressive prediction model are corrected online through a recursive least squares algorithm.

5. A rotating electric machine operation support system based on the rotating electric machine operation support method according to any one of claims 1 to 4, characterized by Includes the following modules: The data acquisition and preprocessing module is used to acquire the operating data of the rotating motor, including three-phase current, terminal voltage and mechanical vibration acceleration, and to extract harmonic voltage and harmonic current based on the operating data. The latent stress estimation module is used to calculate the electrical stress index and the mechanical stress index. The electrical stress index is determined based on harmonic voltage and harmonic current, and the mechanical stress index is determined based on mechanical vibration acceleration. By combining the electrical stress index and the mechanical stress index, a systematic latent stress is generated; The risk trend assessment module is used to calculate the evolution rate of systemic latent stress and generate dynamic failure boundaries based on the electrical stress index. The mode switching decision module is used to calculate the mode switching decision function value based on the systemic latent stress, evolution rate and dynamic failure boundary, and generate control mode switching decisions based on the function value. The suppression strategy generation and execution module is used to generate and execute suppression control strategies in response to control mode switching decisions. The closed-loop correction module is used to collect the operating data after the execution of the inhibitory control strategy, and to correct the inhibitory control strategy in a closed loop based on the updated operating data.

Images