Response speed optimization method for electric control valve actuators applied to ships

Abstract

The present invention discloses a response speed optimization method for an electric control valve actuator of a ship, which specifically relates to the field of response regulation of electric control valves of ships, and is used to solve the problems of response hysteresis and unstable speed regulation of the actuator under complex working conditions. The method is achieved by constructing a response control link with a dynamic feature set as the core, linking the actuator signal behavior with the deviation trend, and accurately separating the distortion band within the inertia compensation threshold domain; introducing a dual-parameter mechanism based on frequency domain energy level and time domain hysteresis, and using machine learning to generate a distortion identification coefficient, accurately deciding whether to introduce a correction operation, so that the speed regulation process has pertinence and controllability; the speed transition curve is driven by the correction factor to ensure that the control signal has flexible adaptability; the path numbering mechanism realizes the traceability of the control trajectory, the a posteriori deviation curve is iteratively combined with the feature set, and the driving strategy continues to evolve, thereby improving the response speed and process stability of the actuator as a whole.

Description

Technical Field

[0001] The present invention relates to the field of response regulation of ship electric control valves, and more particularly to a response speed optimization method for an actuator of a ship electric control valve. Background Art

[0002] A large number of electric control valves on ships convert control commands into valve stem displacement through motors and reduction mechanisms, thereby regulating the flow of media such as cooling water, fuel, and lubricating oil in real time. During navigation, loads fluctuate frequently, sea conditions change rapidly, and ambient temperature and power supply conditions also fluctuate accordingly. Valve openings must quickly adapt to these changes in operating conditions to maintain stable pressure, suppress pipeline oscillations, and avoid cavitation. Actuator action delays directly impact the continuous and reliable operation of propulsion and auxiliary engines. Therefore, reducing response time has become a key requirement for ship fluid control.

[0003] Current control strategies mostly rely on fixed proportional-integral-differential parameters set during the debugging phase, first completing open-loop experiments at the dock, and then manually fine-tuning during closed-loop sea trials. This static setting cannot adapt to environmental drift, component aging, and multi-valve coupling effects during navigation. When the motor torque characteristics and sensor delays change, the valve stem movement is prone to lag, overshoot, or even swing, which prolongs the steady-state time and increases additional energy consumption. Manual recalibration when necessary requires downtime, making it difficult to keep up with instantaneous load changes. Once the voltage drops or the set value changes suddenly, the pipeline pressure quickly deviates from the target and has a chain reaction effect on subsequent equipment. Therefore, there is an urgent need for a data-driven online optimization method that dynamically analyzes operating signals, captures errors, and corrects the speed control strategy at any time, so that the actuator can always maintain a fast and smooth response under complex conditions.

[0004] In order to solve the above problems, a technical solution is now provided. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a response speed optimization method for ship electric control valve actuators. By constructing a response control link with a dynamic feature set as the core, the actuator signal behavior and the deviation trend are linked and analyzed, and the distortion band is accurately separated within the inertia compensation threshold domain; a dual-parameter mechanism based on frequency domain energy level and time domain hysteresis is introduced, and the distortion identification coefficient is generated with the help of machine learning to accurately decide whether to introduce correction operations, so that the speed regulation process has targetedness and restraint; the speed transition curve is driven by the correction factor to ensure that the control signal has flexible adaptability; the path numbering mechanism realizes the traceability of the control trajectory, the a posteriori deviation curve is iteratively combined with the feature set, and the driving strategy continues to evolve, thereby improving the overall response speed and process stability of the actuator to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] S1: Real-time acquisition of actuator current curves and valve stem angular displacement signals, building a dynamic feature set through trend decomposition, and simultaneously generating a predicted deviation image as a benchmark for subsequent operations;

[0008] S2: Identify the inertial coupling strength based on the dynamic feature set, form the inertial compensation threshold through time-varying trade-off calculation, and associate the threshold information with the prediction deviation image;

[0009] S3: Extract the response distortion band within the inertia compensation threshold, obtain the self-adjustment correction factor set by performing distortion screening analysis and assign a time series label;

[0010] S4: Map the self-adjustment correction factor set to a speed transition curve, directly drive the control vector to complete real-time speed regulation, and at the same time send the execution path number to the monitoring end for archiving;

[0011] S5: The monitoring end generates a posterior deviation rate curve based on the serial number tracing, and then integrates it with the dynamic feature set to promote continuous evolution.

[0012] In a preferred embodiment, step S1 includes the following contents:

[0013] By collecting the actuator's current curve and valve stem angular displacement signals in real time and performing trend decomposition using wavelet transform, the trend component reflecting steady changes and the fluctuation component reflecting instantaneous disturbances are extracted. Based on the trend components, a dynamic feature set is constructed, including the current trend component, the rate of change of the current trend component, the angular displacement trend component, and the rate of change of the angular displacement trend component. At the same time, the difference between the actual angular displacement and the ideal angular displacement is calculated, and a predicted deviation image is generated, including the deviation, the cumulative offset of the deviation, and the sharpness of the deviation.

[0014] In a preferred embodiment, step S1 further includes the following:

[0015] The entire process is completed within the same time window, ensuring that the dynamic feature set and the prediction deviation image are synchronized in time.

[0016] In a preferred embodiment, step S2 includes the following:

[0017] The inertial coupling strength is generated by calculating the product of the rate of change of the current trend component and the rate of change of the angular displacement trend component, and combining it with the standardized trend amplitude. The average level and fluctuation amplitude of the inertial coupling strength are calculated based on the sliding window method, and the inertial compensation threshold is constructed.

[0018] In a preferred embodiment, step S2 further includes the following:

[0019] By calculating the normalized distance between the predicted deviation value and the average level of the inertia compensation threshold, the predicted deviation image is associated to ensure that the compensation operation is triggered only when the predicted deviation value exceeds the set range of the inertia compensation threshold.

[0020] In a preferred embodiment, step S3 includes the following contents:

[0021] By comparing the valve stem angular displacement signal with the inertia compensation threshold, the deviation is calculated and the response distortion band is extracted. A multi-scale wavelet transform is performed on the response distortion band, and the cumulative area of ​​the deviation of each sub-band energy from the steady-state reference spectrum is calculated to generate the resonant embedded energy level. At the same time, the current curve and the valve stem angular displacement curve are aligned within the response distortion band, the angular displacement difference of the control section is calculated, and the trapezoidal area is accumulated to generate the hysteresis progressive amplitude.

[0022] In a preferred embodiment, step S3 further includes the following:

[0023] The gradient boosting tree model is used to fuse the resonance embedding energy level and the hysteresis progressive amplitude to output the distortion identification coefficient. The response distortion band whose distortion identification coefficient exceeds the predetermined threshold is included in the set of self-adjusting correction factors and assigned a timing label to ensure that the self-adjusting correction factor accurately corresponds to the distortion period, providing a dynamic adjustment basis for response speed optimization.

[0024] In a preferred embodiment, step S4 includes the following contents:

[0025] The self-tuning correction factor set derived from distortion analysis is mapped to the speed transition curve through a correction function. The correction function integrates the distortion identification coefficient and the time attenuation factor to adjust the reference speed curve to ensure a smooth transition. The speed transition curve is then converted into a control vector through linear mapping. The control vector is output to the actuator hardware as a high-frequency signal to achieve real-time speed adjustment. At the same time, a unique execution path number is generated for each tuning operation and stored in a structured database together with the corresponding self-tuning correction factor set and time interval.

[0026] In a preferred embodiment, step S5 includes the following contents:

[0027] The execution path number is retrieved from the database to trace and calculate the a posteriori deviation rate curve, which is used to quantify the difference between the actual value and the ideal value of the valve stem angular displacement. The a posteriori deviation rate curve is then fused with the dynamic feature set, which includes the current trend value, current change rate, angular displacement trend value, and angular displacement change rate, to form a five-dimensional vector, which is used to characterize the operating status and tuning effect of the actuator.

[0028] In a preferred embodiment, step S5 further includes the following:

[0029] The time-integrated energy of the five-dimensional vector is calculated to evaluate the overall fluctuation intensity. The speed adjustment coefficient is dynamically adjusted according to the posterior deviation rate and time-integrated energy to amplify or smooth the tuning operation, ensuring continuous optimization and stable operation of the actuator under different conditions.

[0030] The technical effects and advantages of the present invention applied to the method for optimizing the response speed of a ship electric control valve actuator are as follows:

[0031] The present invention constructs a response control link based on a dynamic feature set, synchronously analyzing the behavioral trajectories and deviation trends carried by current and displacement signals. A compensation threshold is preset before the inertial coupling characteristics are clear, preventing false triggering of the speed control path. Subsequently, the response distortion band is further separated within the constructed threshold. The two complementary parameters formed by frequency domain energy and displacement cumulative difference are combined to generate a distortion identification coefficient through machine learning. This coefficient is used as the core to determine whether a correction factor is required, ensuring that each speed control operation is based on a clear and targeted judgment. The self-adjusting correction factor then generates a speed transition curve through a mapping strategy, ensuring that the control signal has adaptive transition intent before execution. The addition of path numbering further enhances the localizability of feedback tracking. On this basis, the posterior deviation curve is iteratively interconnected with the initial feature set, enabling the speed control strategy to continuously evolve. The overall structure avoids the drawbacks of traditional redundant static parameter settings, enabling the actuator to quickly and stably respond to control commands under variable operating conditions, and improving the logical clarity and self-consistency of the control process. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 The figure is a flow chart of the method for optimizing the response speed of a ship electric control valve actuator according to the present invention.

[0033] Figure 2 3 is a flow chart of step S3 of the method for optimizing the response speed of a ship electric control valve actuator according to the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] Example 1: Figure 1 The present invention provides a method for optimizing the response speed of a ship electric control valve actuator, including:

[0036] S1: Real-time acquisition of actuator current curves and valve stem angular displacement signals, building a dynamic feature set through trend decomposition, and simultaneously generating a predicted deviation image as a benchmark for subsequent operations.

[0037] S2: Identify the inertial coupling strength based on the dynamic feature set, form the inertial compensation threshold through time-varying trade-off calculation, and associate the threshold information with the predicted deviation image.

[0038] S3: Extract the response distortion band within the inertia compensation threshold, obtain the self-adjustment correction factor set by performing distortion screening analysis, and assign a time series label.

[0039] S4: Map the self-adjusting correction factor set to a speed transition curve, directly drive the control vector to complete real-time speed regulation, and at the same time hand over the execution path number to the monitoring end for archiving.

[0040] S5: The monitoring end generates a posterior deviation rate curve based on serial number tracing, and then integrates it with the dynamic feature set to promote continuous evolution. The actuator response always remains fast and stable.

[0041] In complex operating conditions, such as on ships, the response speed of electric control valve actuators directly impacts system stability and operational efficiency. Existing technologies often employ a statically tuned proportional-integral-derivative (PID) control strategy. This strategy can meet the requirements of shore-based testing and short sea trials during the commissioning phase. However, during actual operation, factors such as motor temperature rise, power supply fluctuations, mechanical wear, and coupling interference can cause changes in the actuator's inertia and friction characteristics, leading to hysteresis, overshoot, and oscillation in valve stem movement, extending response time and increasing energy consumption. Slow response is generally considered a performance bottleneck for electric actuators in the industry. Despite improvements such as adaptive prediction and fuzzy reasoning, the problem persists. This paper proposes a dynamic adaptive optimization method that collects actuator current curves and valve stem angular displacement signals in real time to construct a dynamic feature set and a prediction deviation map. This method aims to provide a data foundation for subsequent inertia compensation and response control, thereby precisely improving response speed. Step S1, the initial step of the solution, is responsible for collecting and processing real-time data.

[0042] S1.1 Data collection;

[0043] When an actuator is operating, the current curve and valve stem angular displacement signal are the core data reflecting its dynamic behavior. The current curve records the motor's current value over time, reflecting the impact of power supply fluctuations and temperature rise on the motor's operating status. The valve stem angular displacement signal records the valve stem's angular position over time, directly reflecting the actuator's motion trajectory and subject to factors such as inertia and friction. To ensure that these signals capture instantaneous changes, the acquisition process uses a high-frequency recording method, acquiring data at continuous time points.

[0044] Specifically, the current curve uses a current sensor to measure the motor current at fixed intervals, generating a time-varying current value sequence. The valve stem angular displacement signal uses an angular displacement sensor to synchronously measure the valve stem angle, generating a time-varying angle value sequence. The intervals must be sufficiently small to ensure data continuity and accuracy, fully capturing the dynamic characteristics of the actuator during operation.

[0045] S1.2 Trend decomposition;

[0046] After acquiring the numerical sequences of the current curve and valve stem angular displacement signals, these signals need to be decomposed into long-term trends and short-term fluctuations to reflect the steady changes and transient disturbances of the actuator, respectively. Wavelet transform was chosen as the decomposition tool because it can separate the low-frequency and high-frequency components of the signal at different time scales, providing a clear basis for extracting dynamic features.

[0047] For the current curve, the entire current numerical sequence is first input into the wavelet transform processing flow. Through multi-layer decomposition, the sequence is divided into a low-frequency trend component and a high-frequency fluctuation component: the low-frequency trend component is obtained by multiple smoothing processes of the original sequence and represents the steady change trend of the current; the high-frequency fluctuation component is composed of the remaining part after deducting the low-frequency part in each decomposition and reflects the instantaneous change of the current.

[0048] The same process is applied to the rotation angle sequence of the valve stem angular displacement signal, generating corresponding low-frequency trend components and high-frequency fluctuation components. The number of decomposition levels chosen depends on the signal complexity and analysis requirements, and is typically set to a fixed number to balance the separation of trends and fluctuations. This decomposition method separates the overall signal behavior from local disturbances, facilitating the subsequent extraction of dynamic features by focusing on key changes.

[0049] S1.3 build dynamic feature set;

[0050] The low-frequency trend component derived from trend decomposition provides the foundational data for constructing a dynamic feature set, which comprehensively reflects the dynamic behavior of current and valve stem angular displacement. The dynamic feature set consists of four components: the current trend component, the rate of change of the current trend component, the angular displacement trend component, and the rate of change of the angular displacement trend component. The calculation process is as follows: the current trend component is directly derived from the decomposition results and is a time-varying low-frequency current numerical sequence. The rate of change of the current trend component is calculated by taking the difference between the current trend components at adjacent time points and dividing it by the corresponding time interval to obtain the velocity of change at each time point, generating a numerical sequence of rate of change values. The angular displacement trend component is also derived from the decomposition results and is a time-varying low-frequency rotation angle numerical sequence. The rate of change of the angular displacement trend component is calculated by taking the difference between the angular displacement trend components at adjacent time points and dividing it by the corresponding time interval to generate a numerical sequence of rotation angle velocity. These four sets of data together constitute the dynamic feature set, which fully describes the dynamic characteristics of the actuator during operation and provides multi-dimensional information support for subsequent analysis. This feature extraction method, by introducing the rate of change, enhances the ability to characterize instantaneous changes in actuator motion.

[0051] S1.4 generates a prediction deviation map;

[0052] The predicted deviation map quantifies the difference between the actual and ideal valve stem angular displacement signals. It consists of three components: the difference between the actual and ideal angular displacements, the cumulative offset of the deviation, and the sharpness of the deviation. The calculation process is as follows: First, the actual angular displacement is derived from the original acquired rotation angle sequence, while the ideal angular displacement is the preset actuator target angle sequence. The two are subtracted to produce a deviation sequence, representing the response error at each time point. Next, the cumulative offset of the deviation is calculated by accumulating the deviation sequence over time. Starting from the starting time point, the deviation values ​​are added point by point to obtain the cumulative sum up to each time point, reflecting the long-term effect of the deviation. Finally, the sharpness of the deviation is characterized by calculating the acceleration of the deviation sequence. First, the difference between adjacent time points in the deviation sequence is calculated to obtain a first-order velocity sequence. Then, the difference between adjacent time points in this sequence is calculated to obtain a second-order acceleration sequence. The average of the absolute values ​​is taken as the sharpness, reflecting the severity of the deviation change. These three components together constitute the predicted deviation map, providing multi-faceted deviation characteristics and comprehensive data support for subsequent analysis of actuator response characteristics.

[0053] S1.5 Synchronous generation;

[0054] To ensure the temporal consistency of the dynamic feature set and the predicted deviation image, the collection, decomposition, and feature extraction processes must be completed within the same time window. The specific implementation method is: using a fixed time period as the processing unit, the current curve and valve stem angular displacement signal are simultaneously collected at high frequency to generate a corresponding numerical sequence; then, trend decomposition is performed on these sequences within the same time period to extract the current trend component and the angular displacement trend component; then, based on the decomposition results, the four parts of the dynamic feature set and the three parts of the predicted deviation image are calculated. All calculations are based on the time point to ensure that the feature data and the deviation data at each time point correspond one to one. This synchronous processing method can ensure the alignment of the data in time series, so that the dynamic feature set and the predicted deviation image can be seamlessly connected for subsequent analysis and calculation, maintaining the consistency and accuracy of the entire processing process.

[0055] Step S1 collects the actuator current curve and valve stem angular displacement signals in real time, extracts a dynamic feature set using trend decomposition techniques, and simultaneously generates a predicted deviation image, thus establishing a complete computational foundation for reflecting the actuator's dynamic behavior and response deviation. The entire process begins with the acquisition of raw data, progresses through decomposition and feature extraction, and ultimately generates multi-dimensional feature and deviation data, ensuring that subsequent analysis is based on accurate and consistent input information.

[0056] Step S1 collects the actuator's current curve and valve stem angular displacement signals in real time and uses trend decomposition techniques to generate a dynamic feature set and predicted deviation image, laying the data foundation for subsequent analysis. This data reflects the actuator's behavior and deviation trends under complex operating conditions. However, variations in inertial coupling strength caused by factors such as motor temperature rise and power supply fluctuations directly affect the response speed of the valve stem movement. It is imperative to quantify and identify this characteristic in step S2 and establish a compensation mechanism to ensure that subsequent steps can optimize the response distortion.

[0057] The processing logic of step S2 is based on the dynamic feature set and predicted deviation image generated in step S1. It aims to identify the inertial coupling strength and build a dynamic compensation mechanism through precise calculation and analysis, providing a reliable basis for optimizing the actuator response characteristics.

[0058] S2.1 Identify the inertial coupling strength based on the dynamic feature set;

[0059] In the optimization of actuator response speed, the inertial coupling strength is a key indicator that characterizes the degree of mutual influence between current changes and angular displacement changes. Its identification relies on the dynamic feature set generated in step S1. The dynamic feature set includes the current trend component, the rate of change of the current trend component, the valve stem angular displacement trend component, and the rate of change of the angular displacement trend component. These data comprehensively reflect the operating status of the actuator. To ensure the accuracy and consistency of the calculation results, the current trend component and the angular displacement trend component are first standardized. Specifically, the standardized current trend component is obtained by dividing the original current trend component by a preset current maximum reference value; similarly, the standardized angular displacement trend component is obtained by dividing the original angular displacement trend component by a preset angular displacement maximum reference value. This processing eliminates the interference of different dimensions on subsequent calculations, making the data comparable.

[0060] Next, the process of calculating the inertial coupling strength is divided into two parts: the numerator and the denominator. The numerator is obtained by multiplying the rate of change of the current trend component with the rate of change of the angular displacement trend component and taking the absolute value, reflecting the dynamic interaction strength between the current change and the angular displacement change. The rate of change of the current trend component is calculated by the difference between the standardized current trend components at adjacent time points, and the rate of change of the angular displacement trend component is calculated by the difference between the standardized angular displacement trend components at adjacent time points. The denominator first calculates the sum of the square of the standardized current trend component and the square of the standardized angular displacement trend component, and then takes the square root of the sum to generate a normalization factor. The inertial coupling strength value is obtained by dividing the numerator by the denominator. This calculation method can dynamically capture the inertial characteristics of the actuator under different operating conditions, ensuring that the results not only reflect real-time changes but also have stable quantitative significance, providing accurate data support for subsequent compensation decisions.

[0061] S2.2 The inertia compensation threshold is formed through time-varying trade-off calculation;

[0062] Due to the complexity and variability of ship operating conditions, the inertial coupling strength fluctuates over time. A single fixed threshold cannot adapt to this dynamic change. Therefore, a time-adjustable inertial compensation threshold is required to determine whether compensation should be triggered. A sliding window method is used in the calculation process to dynamically track the inertial coupling strength. Specifically, a fixed-length time window is first determined, within which continuous inertial coupling strength values ​​are collected. Next, the average level of these values ​​is calculated by summing all inertial coupling strength values ​​within the time window and dividing the sum by the total number of data points. The fluctuation amplitude is then calculated by squaring the difference between each inertial coupling strength value and the average level within the time window. All squared values ​​are then summed and divided by the total number of data points. Finally, the square root of the result is taken to obtain the fluctuation amplitude.

[0063] Based on the average level and fluctuation amplitude, the inertia compensation threshold is determined by adding and subtracting the fluctuation amplitude multiplied by an adjustment coefficient from the average level, creating the upper and lower limits of the threshold. The adjustment coefficient is determined by the actuator's sensitivity and is typically pre-set based on actual operating experience. The resulting inertia compensation threshold is a range of upper and lower limits that updates over the sliding time window, reflecting the short-term trend of inertial coupling strength. This dynamic threshold construction balances the statistical nature of the data with real-time performance, ensuring that compensation decisions are neither too frequent nor delayed, impacting actuator performance.

[0064] S2.3 associating the threshold information with the prediction bias map;

[0065] To closely align the inertia compensation operation with the actuator's actual response deviation, the inertia compensation threshold must be correlated with the predicted deviation image generated in step S1. The predicted deviation image contains the deviation value, the cumulative offset of the deviation, and the sharpness of the deviation, providing information on the trend of the actuator's response deviation. This correlation is achieved by calculating the relative position of the deviation value and the inertia compensation threshold. Specifically, the difference between the deviation value and the average level of the inertia compensation threshold is first calculated, where the average level is the average value of the inertial coupling strength within the time window. This difference is then divided by the fluctuation amplitude of the inertia compensation threshold to obtain a normalized distance value, which represents the degree to which the deviation value deviates from the average level.

[0066] Subsequently, the absolute value of the normalized distance value is taken and compared with the pre-set adjustment coefficient. If the absolute value of the normalized distance value is greater than the adjustment coefficient, it indicates that the deviation value has exceeded the range of the inertia compensation threshold and needs to trigger subsequent compensation operations; conversely, if it is less than or equal to the adjustment coefficient, the deviation is considered to be within the acceptable range and no adjustment is required. The size of the adjustment coefficient is related to the response sensitivity and stability requirements of the actuator and is usually determined through experiments or simulations. This correlation method effectively integrates the deviation trend with the threshold information, ensuring that compensation operations are only performed when the deviation deviates significantly from the normal range, thereby improving the response speed while maintaining the stability of the system.

[0067] In step S1, real-time data collection and trend decomposition complete the construction of a dynamic feature set and a predicted deviation map, providing the necessary data input for the analysis in step S2. Relying on this data, step S2 dynamically evaluates the actuator's response characteristics by identifying the inertial coupling strength, generating an inertial compensation threshold, and correlating it with the predicted deviation map. The subsequent step S3 utilizes the results of step S2 to further analyze response distortion and implement optimized control. This seamless process ensures efficient actuator operation under complex operating conditions.

[0068] Step S2 identifies the inertial coupling strength based on the dynamic feature set, forms the inertial compensation threshold through time-varying trade-off calculation, and associates it with the predicted deviation image, completing the construction of a preliminary compensation framework for the dynamic behavior of the actuator. However, under complex working conditions, the actuator response process may cause abnormal fluctuations in the valve stem motion trajectory due to factors such as motor temperature rise, power supply fluctuations or mechanical wear, which manifests as lag, overshoot or oscillation, directly affecting the response speed and stability. Therefore, if Figure 2 The step S3 shown needs to further extract and analyze these response distortion bands within the inertia compensation threshold to generate a set of self-adjusting correction factors, thereby providing an accurate correction basis for subsequent speed optimization.

[0069] S3.1 Extracting the response distortion band within the inertia compensation threshold;

[0070] Abnormal fluctuations in the valve stem angular displacement signal often reflect actuator response distortion within a specific time period. To accurately locate these distortion bands, step S3 first uses the inertia compensation threshold generated in step S2 as the basis for analysis. The inertia compensation threshold, consisting of an upper and lower limit, defines the normal fluctuation range of the valve stem angular displacement signal. Based on this, a time series monitoring method is used to perform point-by-point analysis of the valve stem angular displacement signal. Specifically, within a continuous time window, each value of the valve stem angular displacement signal is compared with the upper and lower limits of the inertia compensation threshold to calculate the deviation. If the valve stem angular displacement signal value exceeds the upper limit, the deviation is defined as the difference between the signal value and the upper limit; if the signal value is below the lower limit, the deviation is defined as the difference between the lower limit and the signal value; if the signal value is between the upper and lower limits, the deviation is zero. By traversing the entire time series and recording the consecutive time periods where the deviation is non-zero, a set of response distortion bands is generated, with the start and end times of each band accurately recorded.

[0071] By comparing the relative position of the signal and the threshold, we ensure that only abnormal fluctuations beyond the normal range are captured, avoiding misjudging subtle changes in normal operation as distortion, thereby providing reliable distortion data for subsequent analysis.

[0072] S3.2 Perform aberration screening analysis;

[0073] After extracting the response distortion bands, further quantitative analysis of their characteristics is crucial to determine which bands require self-adjustment. To this end, we introduce two parameters: the resonance embedding energy level and the hysteresis progressive amplitude. This is combined with a gradient boosting tree model to generate distortion identification coefficients, enabling intelligent distortion screening. The following three sections describe the calculation process in detail.

[0074] S3.2.1 Calculate the resonant embedding energy level;

[0075] Abnormal energy distribution in the valve stem angular displacement signal at specific frequencies can cause actuator resonance, affecting response speed. To quantify this phenomenon, a multi-scale wavelet transform is used to decompose the valve stem angular displacement signal within the response distortion band, generating multiple energy sub-bands. The specific process involves first applying a wavelet transform to the signal to decompose sub-band signals at different frequency scales. The signal coefficients of each sub-band are then squared and integrated over time to obtain the energy value for that sub-band. To assess the degree of abnormality, a steady-state reference spectrum is pre-established, and the energy values ​​of each sub-band during normal operation are recorded. The energy value of each sub-band is subtracted from the energy value of the corresponding sub-band in the steady-state reference spectrum to calculate the deviation. The deviations are then accumulated across all sub-bands to obtain the resonant embedded energy level. This value reflects the degree of abnormality in the distortion band in the frequency domain and provides a frequency-domain signature for screening. This approach effectively captures potential resonance issues in the frequency domain, ensuring a comprehensive analysis.

[0076] S3.2.2 Calculate the incremental amplitude of the hysteresis;

[0077] A timing mismatch between the current curve and the valve stem angular displacement curve can cause hysteresis, affecting the smoothness of actuator motion. To quantify this mismatch, the current curve and the valve stem angular displacement curve are first aligned on the time axis within the response distortion band. The band is then divided into multiple control segments of equal duration. Within each control segment, the numerical change in the valve stem angular displacement curve is calculated: the angular displacement value at the end of the segment minus the angular displacement value at the beginning, resulting in the angular displacement difference. The angular displacement differences of all control segments are then accumulated using the trapezoidal area formula. This sums the difference of each control segment by the time interval and generates the hysteresis amplitude. This value reflects the cumulative effect of the hysteresis phenomenon in the time domain and provides a time-domain characteristic basis for distortion analysis. By refining the time period division and cumulative calculation, the impact of hysteresis on actuator performance can be accurately described.

[0078] S3.2.3 Generate distortion recognition coefficients using gradient boosting tree;

[0079] To integrate the characteristics of the resonant embedding energy level and the hysteresis progressive amplitude, a gradient boosting tree model is used for joint analysis to generate the distortion recognition coefficient. The specific process is to use the resonant embedding energy level and hysteresis progressive amplitude of each response distortion band as input features and input them into the pre-trained gradient boosting tree model.

[0080] First, the model sets an initial prediction for each example in the training data, which can be the average of all example labels (or, for classification problems, the log-odds of the majority class). Next, at each iteration, the model calculates the residual (i.e., gradient) between the current prediction and the actual label. This residual reflects the model's current prediction error. The model then trains a new decision tree (a weak learner) to fit these residuals and thereby correct the current prediction error. After training is complete, the model adds the new decision tree to the existing model, controlling the influence of the new tree using a learning rate to update the model's predictive power. A smaller learning rate means more cautious updates, potentially requiring more iterations to achieve optimal results. The model repeats this process—calculating residuals, training new trees, and updating the model—until the pre-set number of iterations is reached or the residual decreases to a certain level. During training, the model records the frequency of each feature in the decision tree and its contribution to reducing the residual error, calculating feature importance (i.e., feature weight). Finally, during the prediction phase, the model performs a weighted summation of the predictions from all trained decision trees to obtain the final prediction value. This value is then converted to a value between 0 and 1 using an activation function (such as the sigmoid function). This value serves as the distortion identification coefficient, representing the probability that a particular band is distorted. In this way, the model gradually optimizes feature weights, learns the complex relationship between features and labels, and ultimately outputs a value reflecting the distortion probability.

[0081] A predetermined threshold is set and the distortion identification coefficient is compared with the threshold. If the coefficient is greater than the threshold, the band is determined to require self-correction and is included in the set of self-correction correction factors, while the corresponding time period is recorded. If the coefficient is less than or equal to the threshold, the band is excluded. This method, through machine learning, fuses frequency and time domain features to intelligently identify the distortion bands that truly require correction, improving screening accuracy and adaptability.

[0082] The training process of the gradient boosting tree model is based on response distortion band data extracted from historical operating data. It uses the resonance embedding energy level and hysteresis progressive amplitude as input features and the distortion identification coefficient (a value ranging from 0 to 1, indicating the likelihood of distortion) as the output label. The goal is to learn the mapping relationship between features and labels to achieve intelligent classification and screening of distortions. First, by preprocessing the historical data, multiple response distortion bands are extracted and their resonance embedding energy levels and hysteresis progressive amplitudes are calculated. Simultaneously, through expert annotation or simulation analysis, each band is assigned a distortion identification coefficient label to construct a training dataset. Model training uses a serial construction of multiple decision trees: Initially, a base decision tree is trained to fit the feature-label mapping relationship and generate preliminary predictions. The residuals between the predicted values ​​and the actual labels are then calculated, and the next decision tree is trained to fit these residuals. Through multiple iterations, each new tree is optimized based on the prediction error of the previous tree. In each iteration, the model first calculates the residual between the current prediction and the actual label. This residual, known as the gradient, reflects the prediction error of the previous tree. Next, the model trains a new decision tree to fit these residuals, correcting previous predictions by learning the direction and magnitude of the error. After training, the new decision tree is added to the model, and its influence is controlled by a learning rate. A smaller learning rate means more cautious updates to avoid overfitting. The model then recalculates the residuals based on the updated predictions. This process—calculating residuals, training new trees, and updating the model—repeats until the preset number of iterations is reached or the residuals are sufficiently small. In this way, each new tree optimizes based on the error of the previous tree, gradually improving the model's predictive power and ultimately outputting more accurate results, thereby gradually improving the overall model's prediction accuracy. To ensure model performance, cross-validation is used during training to adjust hyperparameters, including the number of trees, depth, and learning rate, to balance fitting performance with generalization. During the iterative process, the model automatically learns the weights of the influence of the resonant embedding energy level and the hysteresis progressive amplitude on the distortion recognition coefficient, capturing the nonlinear relationships and interactions between features. This approach leverages both frequency and time domain features to improve recognition accuracy. After training, the model can be applied to new response distortion band analysis. By inputting the resonant embedding energy level and hysteresis progressive amplitude features, the model directly outputs the distortion recognition coefficient, providing a basis for subsequent self-adjustment and correction decisions.

[0083] S3.3 obtains a set of self-adjustment correction factors and assigns them a time series label;

[0084] After the response distortion band that needs to be corrected is determined through distortion screening analysis, the generation of a set of self-adjusting correction factors becomes the last step of step S3. Specifically, the response distortion bands whose distortion identification coefficient exceeds the predetermined threshold are sorted into a set, and each band is an element in the set, with its corresponding start time and end time as a timing label. Each element in the set is a self-adjusting correction factor, which indicates the basis for adjusting the actuator response speed within a specific time period. This structured set ensures that each self-adjusting correction factor corresponds one-to-one to the specific time period when the distortion occurs, providing accurate input data for the subsequent step S4 to generate the speed transition curve. Through clear time association and data organization, the targeted and timely nature of the correction operation can be guaranteed.

[0085] By identifying the inertial coupling strength and establishing the inertial compensation threshold, step S2 provides a dynamic analysis framework for extracting the response distortion band in step S3. Relying on this framework, step S3 completes the entire process from distortion band extraction to screening analysis and generation of a set of self-adjusting correction factors. The subsequent step S4 utilizes this set of self-adjusting correction factors to generate a speed transition curve and implement real-time speed regulation, ensuring effective optimization of the actuator's response speed and stability under complex operating conditions.

[0086] Step S3 extracts the response distortion band within the inertia compensation threshold and performs distortion screening analysis to obtain a set of self-adjusting correction factors and assigns a timing label to each correction factor. The set of self-adjusting correction factors contains distortion identification coefficients and corresponding timing labels. This information captures the key characteristics of the actuator's response distortion under complex working conditions and lays a data foundation for subsequent optimization of response speed. However, to apply these abstract correction factors to actual control, they must be converted into specific, executable control strategies to achieve real-time speed regulation of the actuator and ensure the smoothness and traceability of the speed regulation process. Therefore, the task of step S4 is to map the set of self-adjusting correction factors into a speed transition curve, directly drive the control vector to complete real-time speed regulation, and archive the execution path number to provide support for subsequent monitoring and optimization.

[0087] S4.1 maps the set of self-adjustment correction factors into a speed transition curve;

[0088] Each element in the self-tuning correction factor set consists of a distortion identification coefficient and a corresponding time range, representing the degree of actuator response distortion and the time period over which it occurs, respectively. To translate this information into a practical speed adjustment scheme, a speed transition curve is constructed to ensure smooth actuator speed changes within the distortion period, avoiding potential impacts of sudden changes on system stability and mechanical components.

[0089] The specific process is as follows: First, adjustments are made based on the actuator's baseline velocity profile. The baseline velocity profile is the velocity trajectory of the actuator without any corrections applied, typically predetermined by the initial control strategy. For each element in the set of self-tuning correction factors, a velocity correction function is defined to dynamically modify the baseline velocity profile. The velocity correction function is designed to multiply the baseline velocity by an adjustment factor. This adjustment factor consists of two components: a distortion factor, which reflects the magnitude of the adjustment; and a time decay function, which controls the smoothness of the adjustment process.

[0090] The time decay function uses a sinusoidal form, characterized by starting at zero at the beginning of the time range, gradually rising to a maximum value, and then smoothly falling back to zero. The entire change process is confined to the corresponding time range. During the specific calculation, the position ratio of each time point in the time range relative to the starting point is first determined. For example, the time from a certain time point to the starting point is divided by the total length of the entire time range to obtain a value between 0 and 1. This value is then multiplied by pi and input into the sine function to calculate the decay value at that time point. Finally, this decay value is multiplied by the distortion identification coefficient to obtain an adjustment factor, which is then multiplied by the reference speed to obtain the target speed at that time point. By repeating this calculation for all time points in the time range, a continuous speed transition curve is generated. In this way, the speed transition curve can dynamically adjust the actuator's target speed according to the degree of distortion during the distortion period, while maintaining the smoothness of the change, ensuring the stability of the actuator operation and the response optimization effect.

[0091] S4.2 directly drives the control vector to achieve real-time speed regulation;

[0092] After the speed transition curve is generated, it needs to be converted into a control vector that can be directly applied to the actuator hardware to achieve real-time speed adjustment. The control vector is the input signal sent to the actuator hardware (such as a motor or valve) and directly determines its operating state. Therefore, it must accurately reflect the changing trend of the speed transition curve.

[0093] The specific process is as follows: First, a correspondence between the speed transition curve and the control vector is established using a linear mapping method. This means that the control vector value is proportional to the target speed of the speed transition curve. The proportionality factor is determined by the physical characteristics of the actuator hardware, such as the motor's sensitivity to input signals, and is typically determined through experimental calibration. During calculation, each target speed value on the speed transition curve is multiplied by the proportionality factor to obtain the corresponding control vector value. For example, if the target speed at a certain time point is 1.2 times the baseline speed, this value is multiplied by the proportionality factor to obtain the control vector value at that time point. Next, a digital signal processor or microcontroller calculates the control vector value at each time point in real time at a high-frequency sampling rate and outputs it to the motor driver module. This high sampling rate means that the control vector is updated multiple times in extremely short intervals, for example, every millisecond, to ensure that the actuator can keep up with changes in the speed transition curve. During this output process, the control vector is sent to the hardware as an electrical signal, driving the actuator to adjust its speed or opening to achieve an operating state consistent with the target speed. This real-time driving method can efficiently transmit speed adjustment instructions to the hardware layer, ensuring the actuator's rapid response capability and precise control effect to dynamic working conditions.

[0094] S4.3 submit the execution path number to the monitoring terminal for archiving;

[0095] After each speed regulation operation, the relevant data needs to be recorded to support subsequent performance analysis and optimization. To this end, an execution path number is introduced as a unique identifier for each speed regulation operation and is archived along with the adjustment parameters. The specific process is as follows: First, after the speed transition curve is generated and the control vector is driven, a unique execution path number is assigned to each speed regulation operation. This number can be a timestamp-based sequence, such as the exact start time of the operation, or a sequentially increasing numeric identifier. Next, the execution path number is combined with the corresponding self-adjustment correction factor (including the distortion identification coefficient and time range) and the execution time range of the speed regulation operation to form a data record. This data set is then transmitted to the monitoring end database for storage. During storage, the data is organized in a structured format, ensuring that the execution path number serves as an index for fast retrieval. By querying the execution path number, the monitoring end can retrieve the corresponding speed regulation parameters and time information for analyzing the actuator's operating status or evaluating the effectiveness of the adjustment strategy. This archiving mechanism provides complete data support for tracking the speed regulation process, ensuring transparency and continuous improvement of the optimization method.

[0096] Step S3 analyzes the distortion bands in the actuator response to generate a set of self-adjusting correction factors, including distortion identification coefficients and time ranges. This provides the input for dynamic speed control in step S4. Step S4 maps the self-adjusting correction factors into a speed transition curve and uses a control vector to drive the actuator for real-time speed control. Operation details are recorded using an execution path number archiving mechanism. Subsequent step S5 analyzes and optimizes this archived data to ensure effective improvement in actuator response speed and stability under complex operating conditions.

[0097] Step S4 realizes the dynamic response adjustment of the actuator under complex working conditions by mapping the self-adjusting correction factor set into a speed transition curve and driving the control vector to complete real-time speed regulation. At the same time, the archiving of the execution path number provides a traceable data basis for subsequent monitoring and optimization. However, factors such as motor temperature rise, power supply fluctuations and mechanical wear under complex working conditions may cause the optimization effect of the response speed to decay over time. In order to ensure that the actuator can continue to maintain a fast and stable response characteristic in a changing environment, the speed regulation effect must be monitored in real time, and the monitoring results must be integrated with the dynamic feature set to achieve continuous evolution of the control strategy. Therefore, the task of step S5 is to generate a posterior deviation rate curve based on the archived execution path number, and dynamically adjust the control parameters through integration with the dynamic feature set, thereby promoting continuous optimization and stable operation of the actuator response.

[0098] S5.1 The monitoring end generates a posterior deviation rate curve based on the execution path number traceback;

[0099] In the process of optimizing the actuator response speed, it is necessary to accurately evaluate the actual effect of each speed regulation operation to determine the effectiveness of the speed transition curve and provide data support for subsequent control adjustments.

[0100] The posterior deviation rate curve is generated by tracing the execution path number. The specific processing process is as follows: First, the monitoring end retrieves the execution path number from the archive database. The execution path number is a unique identifier archived after each speed regulation operation in step S4 is completed. It records the start and end time of the speed regulation period, as well as the specific application of the self-adjusting correction factor. The self-adjusting correction factor includes a distortion recognition coefficient and a time range, which is used to characterize the correction of the response distortion during the speed regulation process. Then, within the speed regulation period, the monitoring end collects the actual valve stem angular displacement signal of the actuator in real time. The actual valve stem angular displacement signal is the actual motion trajectory of the valve stem of the actuator after the speed regulation operation, which is obtained by sensor measurement. At the same time, the ideal angular displacement is extracted from the predicted deviation image generated in step S1.

[0101] The calculation of ideal angular displacement is based on a dynamic feature set and a prediction model to generate a theoretical motion trajectory, which serves as a benchmark for speed regulation performance. The specific process is as follows: First, the dynamic feature set is collected from the actuator's real-time operating status, including current trend values, current rate of change, angular displacement trend values, and angular displacement rate of change. This information fully describes the actuator's behavior under current operating conditions. These real-time dynamic feature sets are then input into a pre-trained prediction model. This prediction model uses historical data and machine learning techniques to learn the actuator's response patterns under different operating conditions. Based on the input dynamic feature set, it predicts the ideal angular displacement value that the actuator should achieve under ideal conditions. Next, during the speed regulation period, the prediction model continuously outputs multiple ideal angular displacement values. These values ​​are concatenated to form a continuous theoretical motion trajectory. This trajectory represents the actuator's optimal motion performance under the current operating conditions and serves as a theoretical benchmark. The effectiveness of the speed regulation operation can then be evaluated by comparing the actual valve stem angular displacement signal with this theoretical motion trajectory, specifically by calculating the deviation between the two. This approach combines real-time data and intelligent prediction to ensure that the theoretical motion trajectory is both accurate and adaptable to actual working conditions, providing a reliable reference for optimizing the actuator response speed.

[0102] Next, the posterior deviation is calculated by subtracting the actual valve stem angular displacement from the ideal angular displacement at each time point. This deviation value reflects the difference between the actual and ideal responses. To further quantify the relative degree of deviation, the posterior deviation rate is calculated by dividing the posterior deviation by the ideal angular displacement to generate a dimensionless ratio. By continuously calculating the posterior deviation rate during the speed regulation period, a time-varying curve is formed, called the posterior deviation rate curve.

[0103] The generation process of the posterior deviation rate curve utilizes the comparative analysis of archived data and real-time acquired signals, which can accurately reflect the actual effect of the speed regulation operation and provide an objective basis for the dynamic adjustment of the control strategy.

[0104] S5.2 is integrated with dynamic feature sets to enable continuous evolution;

[0105] On the basis of evaluating the speed regulation effect, it is necessary to combine the evaluation results with the real-time operating status of the actuator to comprehensively analyze the relationship between the deviation characteristics and the current working conditions, so as to dynamically adjust the control parameters to ensure that the response speed and stability of the actuator under complex working conditions are continuously optimized.

[0106] This goal is achieved by fusing the posterior deviation rate curve with the dynamic feature set. The specific process is as follows: First, the posterior deviation rate curve and the dynamic feature set are time-aligned. The dynamic feature set is the real-time state description constructed in step S1 and includes the current trend value, current rate of change, angular displacement trend value, and angular displacement rate of change, which respectively characterize the dynamic behavior of the actuator motor current and valve stem movement. Time alignment ensures that the posterior deviation rate and the dynamic feature set correspond to each other in the time dimension. Next, a fused feature vector is constructed by combining the four elements of the dynamic feature set—current trend value, current rate of change, angular displacement trend value, and angular displacement rate of change—with the posterior deviation rate to form a five-dimensional vector. This five-dimensional vector comprehensively characterizes the actuator's operating state and the speed regulation effect. Next, the time-integrated energy of the fused feature vector is calculated by squaring each element in the five-dimensional vector over the speed regulation period. The squared values ​​are then summed to form a single value. This sum is then integrated over the speed regulation period to obtain a single value representing the overall fluctuation intensity.

[0107] The current trend value represents the long-term variation of the actuator current curve and reflects the smoothly changing characteristics of the motor's operating state. This value is obtained by extracting the trend component from the real-time current data. This five-dimensional vector provides information about the actuator's dynamic characteristics and reveals the overall trend and stability of the motor's operation. Acquisition process: First, the actuator's current data is collected in real time via sensors to form a time-varying current curve. Next, trend decomposition techniques (such as wavelet transform or sliding average) are used to decompose the current curve into trend and fluctuation components. Finally, the trend component is extracted, which is the current trend value.

[0108] The current rate of change indicates how quickly the current trend value changes over time, reflecting the transient characteristics of the motor's operating state. This value is calculated by numerically differentiating the current trend value, providing information about the actuator's dynamic response as a five-dimensional vector and describing the motor's current fluctuations over short periods of time. Acquisition process: First, based on the previous steps, obtain the time series data of the current trend value; then, calculate the difference between the current trend values ​​at adjacent time points; finally, divide this difference by the corresponding time interval to obtain the current rate of change, which serves as a quantitative indicator of dynamic changes.

[0109] The angular displacement trend value represents the long-term trend of the valve stem angular displacement signal and reflects the smooth variation of the valve stem motion. This value is obtained by extracting the trend component from the real-time angular displacement data. This value provides information about the actuator's motion characteristics in a five-dimensional vector, revealing the overall trend of the valve stem motion. Acquisition process: First, the valve stem angular displacement data is collected in real time by a sensor, forming a time-varying angular displacement curve. Next, trend decomposition techniques are used to decompose the angular displacement curve, separating the trend component and the fluctuation component. Finally, the trend component is extracted, forming the angular displacement trend value, which is used to characterize the motion characteristics.

[0110] The angular displacement rate of change represents the rate of change of the angular displacement trend value over time, reflecting the transient characteristics of the valve stem motion. This value is calculated by numerically differentiating the angular displacement trend value, providing information about the actuator's motion response as a five-dimensional vector, capable of describing the valve stem's motion dynamics over a short period of time. Acquisition process: First, based on the aforementioned steps, the time series data of the angular displacement trend value is acquired; then, the difference in the angular displacement trend value between adjacent time points is calculated; finally, this difference is divided by the corresponding time interval to obtain the angular displacement rate of change, which serves as a quantitative indicator of the motion response.

[0111] The posterior deviation rate represents the ratio of the difference between the actual and ideal angular displacements after the speed regulation operation to the ideal angular displacement, reflecting the relative error in the speed regulation effect. This value is obtained by calculating and normalizing the deviation between the actual and ideal angular displacements. It provides information about the effectiveness of the control strategy for the five-dimensional vector and measures control accuracy. Acquisition process: First, during the speed regulation operation, the actual angular displacement data is acquired through the sensor, and the corresponding ideal angular displacement value is determined. Then, the difference between the actual and ideal angular displacements is calculated to obtain the posterior deviation. Finally, the posterior deviation is divided by the ideal angular displacement to obtain the posterior deviation rate, which serves as an evaluation indicator of control effectiveness.

[0112] The speed adjustment coefficient is dynamically adjusted based on the posterior deviation rate curve and the energy changes of the fused eigenvector. If the absolute value of the posterior deviation rate remains above the preset threshold during the speed regulation period and the energy of the fused eigenvector shows an increasing trend, the speed adjustment coefficient is increased by a specific amount determined by the product of the deviation rate and the energy value to improve the amplitude of the speed regulation operation. If the posterior deviation rate shows high-frequency changes during the speed regulation period and the energy of the fused eigenvector fluctuates violently, the speed adjustment coefficient is reduced by a specific amount determined by the product of the frequency of change and the amplitude of the energy fluctuation to smooth the response process.

[0113] This adjustment mechanism enables the control strategy to adaptively evolve based on real-time feedback and status data, maintaining rapid actuator response and stable operation. This fusion process combines the evaluation of speed regulation effectiveness with the dynamic characteristics of the operating state, forming a closed-loop optimization path and improving the control strategy's adaptability in changing environments.

[0114] Step S4 maps the set of self-tuning correction factors into a speed transition curve and drives the control vector, completing the actuator's real-time speed regulation and archiving the execution path number. Step S5, leveraging this archived data, generates a posterior deviation rate curve by tracing the execution path number. This curve is then integrated with the dynamic feature set to dynamically adjust the speed adjustment coefficient, achieving continuous optimization and adaptive evolution of the control strategy. This closed-loop feedback mechanism ensures efficient actuator operation under complex ship operating conditions.

[0115] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0116] It should be noted that the system of the present invention can be deployed on the device itself to realize embedded applications, and can also be run on a PC or other terminal with a user interface, thereby meeting a variety of hardware environments and usage requirements.

[0117] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

[0118] It should be noted that, in this document, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article or device. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device that includes the element.

[0119] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A method for optimizing the response speed of an electric control valve actuator applied to a ship, characterized in that: Including steps: S1: Real-time acquisition of actuator current curves and valve stem angular displacement signals, building a dynamic feature set through trend decomposition, and simultaneously generating a predicted deviation image as a benchmark for subsequent operations; S2: Identify the inertial coupling strength based on the dynamic feature set, form the inertial compensation threshold through time-varying trade-off calculation, and associate the threshold information with the prediction deviation image; Step S2 includes the following contents: The inertial coupling strength is generated by calculating the product of the rate of change of the current trend component and the rate of change of the angular displacement trend component, and combining it with the normalized trend amplitude. The average level and fluctuation amplitude of the inertial coupling strength are calculated based on the sliding window method to construct the inertial compensation threshold. By calculating the normalized distance between the predicted deviation value and the average level of the inertia compensation threshold, the predicted deviation image is associated to ensure that the compensation operation is triggered only when the predicted deviation value exceeds the set range of the inertia compensation threshold; S3: Extract the response distortion band within the inertia compensation threshold, obtain the self-adjustment correction factor set by performing distortion screening analysis and assign a time series label; Step S3 includes the following contents: By comparing the valve stem angular displacement signal with the inertia compensation threshold, the deviation is calculated and the response distortion band is extracted. A multi-scale wavelet transform is performed on the response distortion band, and the cumulative area of ​​deviation between the energy of each sub-band and the steady-state reference spectrum is calculated to generate the resonant embedded energy level. Simultaneously, the current curve and the valve stem angular displacement curve are aligned within the response distortion band, the angular displacement difference of the control section is calculated, and the trapezoidal area is accumulated to generate the hysteresis progressive amplitude. The gradient boosting tree model is used to fuse the resonance embedding energy level and the hysteresis progressive amplitude to output the distortion identification coefficient. The response distortion bands whose distortion identification coefficients exceed the predetermined threshold are included in the set of self-adjusting correction factors and assigned time series labels to ensure that the self-adjusting correction factors accurately correspond to the distortion period, providing a dynamic adjustment basis for response speed optimization. S4: Map the self-adjustment correction factor set to a speed transition curve, directly drive the control vector to complete real-time speed regulation, and at the same time send the execution path number to the monitoring end for archiving; S5: The monitoring end generates a posterior deviation rate curve based on the serial number tracing, and then integrates it with the dynamic feature set to promote continuous evolution.

2. The method for optimizing the response speed of a ship electric control valve actuator according to claim 1 is characterized in that: Step S1 includes the following contents: By collecting the actuator's current curve and valve stem angular displacement signal in real time and performing trend decomposition using wavelet transform, the trend component reflecting steady changes and the fluctuation component reflecting instantaneous disturbances are extracted. Based on the trend component, a dynamic feature set including the current trend component, the rate of change of the current trend component, the angular displacement trend component and the rate of change of the angular displacement trend component is constructed; At the same time, the difference between the actual angular displacement and the ideal angular displacement is calculated, and a predicted deviation map is generated, which includes the deviation, the cumulative offset of the deviation, and the sharpness of the deviation.

3. The method for optimizing the response speed of a ship electric control valve actuator according to claim 2 is characterized in that: Step S1 also includes the following: The entire process is completed within the same time window, ensuring that the dynamic feature set and the prediction deviation image are synchronized in time.

4. The method for optimizing the response speed of a ship electric control valve actuator according to claim 1 is characterized in that: Step S4 includes the following contents: The self-tuning correction factor set derived from distortion analysis is mapped to the speed transition curve through a correction function. The correction function integrates the distortion identification coefficient and the time attenuation factor to adjust the reference speed curve to ensure a smooth transition. The speed transition curve is then converted into a control vector through linear mapping. The control vector is output to the actuator hardware as a high-frequency signal to achieve real-time speed adjustment. At the same time, a unique execution path number is generated for each tuning operation and stored in a structured database together with the corresponding self-tuning correction factor set and time interval.

5. The method for optimizing the response speed of a ship electric control valve actuator according to claim 1 is characterized in that: Step S5 Includes the following: The execution path number is retrieved from the database to trace and calculate the a posteriori deviation rate curve, which is used to quantify the difference between the actual value and the ideal value of the valve stem angular displacement. The a posteriori deviation rate curve is then fused with the dynamic feature set, which includes the current trend value, current change rate, angular displacement trend value, and angular displacement change rate, to form a five-dimensional vector, which is used to characterize the operating status and tuning effect of the actuator.

6. The method for optimizing the response speed of a ship electric control valve actuator according to claim 5 is characterized in that: Step S5 also includes the following: The time-integrated energy of the five-dimensional vector is calculated to evaluate the overall fluctuation intensity. The speed adjustment coefficient is dynamically adjusted according to the posterior deviation rate and time-integrated energy to amplify or smooth the tuning operation, ensuring continuous optimization and stable operation of the actuator under different conditions.

Images