A method and system for online parameter tuning and optimization of a PPEC controller

Abstract

The present application belongs to the technical field of adaptive control, and particularly relates to a PPEC controller online parameter setting and optimization method and system, which comprises the following steps: obtaining output power and modulation instructions in a current window, determining trajectory deviation and dynamic damping ratio according to the change of the output power and the modulation instructions; taking the difference between the trajectory deviation and the reference value of the trajectory deviation, and the dynamic damping ratio as two components of the evolutionary attractor; determining the polarity coefficient by comparing the two components of the evolutionary attractor, and then determining the correction increment, so as to adaptively update the real-time control parameters. The present application drives the servo adaptive update of the control parameters by the evolutionary attractor driving the setting step, thereby suppressing the hardware oscillation, and significantly improving the tracking accuracy and response efficiency of the PPEC under complex dynamic load.

Description

Technical Field

[0001] This invention relates to the field of adaptive control technology. More specifically, this invention relates to a method and system for online parameter tuning and optimization of a PPEC controller. Background Technology

[0002] In the digital control of high-performance power electronic converters, the Programmable Power Electronics Controller (PPEC) integrates complex soft-switching control and multi-loop energy management logic. In actual R&D and industrial application scenarios, due to the nonlinear magnetic saturation characteristics exhibited by physical components such as inductors and capacitors in the main circuit under different current stresses, the preset fixed parameters are difficult to cover the full load range. Engineers usually need to use the PPEC Workbench host computer tool to read and write the underlying registers in real time through the Modbus communication link to correct the control operators affected by load fluctuations and environmental changes.

[0003] Currently, the parameter tuning of PPEC controllers adopts a semi-automatic interactive mode based on PPEC Workbench. The implementation method is to use a static observation-manual step adjustment strategy, that is, the control register is manually adjusted stepwise according to the feedback waveform. The core adjustment mechanism of this method is to find the stable point by adding or subtracting values ​​with a fixed tuning step size, ignoring the dynamic characteristics of the system in the energy evolution process, resulting in extremely low adjustment efficiency.

[0004] However, when faced with high-power, rapidly switching loads, the energy evolution of power electronic systems exhibits a high degree of nonlinearity. Due to the use of a fixed tuning step size, it is impossible to adaptively adjust the regulation intensity according to the degree of deviation of the system from the steady state. This often leads to regulation oscillation during the tuning process, that is, the parameters repeatedly jump around the optimal value, resulting in low tuning accuracy. In some cases, the uncontrolled step size may even induce hardware-level oscillations, which cannot meet the requirements of modern high-frequency power electronic equipment for parameter self-adaptation and oscillation-free optimization. Summary of the Invention

[0005] To address the technical problems of low regulation oscillation and tuning accuracy caused by the fixed parameter tuning step size and neglect of energy evolution characteristics in the PPEC controller under complex dynamic loads, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for online parameter tuning and optimization of a PPEC controller, comprising: acquiring the output power sequence and modulation command sequence within the current window; determining the trajectory deviation based on the cumulative difference between the output power and the modulation command at each sampling moment within the current window; acquiring the variation rate based on the change in output power within the current window; mapping the variation rate to a dynamic damping ratio using an inverse proportional function; using the difference between the trajectory deviation and a baseline value of the trajectory deviation, and the dynamic damping ratio, respectively, as two components for constructing an evolution attractor; comparing the magnitude relationship between the two components of the evolution attractor to determine the polarity coefficient; determining the correction increment based on the evolution attractor and the polarity coefficient; acquiring the mean value of the real-time control parameters of the register within the current window, and summing the correction increment with the mean value to obtain the updated control parameters, thereby realizing online optimization of the PPEC controller.

[0007] This invention constructs an evolutionary attractor to integrate discrete waveform distortion and energy damping characteristics into a continuous state space displacement, comprehensively analyzing the degree to which the system deviates from steady state. It introduces an adaptive polarity switching mechanism to automatically determine the adjustment direction based on the dominant component of the evolutionary attractor: derating to suppress vibration when oscillation is dominant, and gain correction when error is dominant. This variable step size strategy based on physical energy evolution effectively suppresses hardware oscillation while significantly improving the tracking accuracy and dynamic response speed of the controller.

[0008] Preferably, determining the trajectory deviation includes: calculating the square of the difference between the output power and the modulation command at each sampling time within the current window, obtaining the mean of the squares of all the differences within the current window, and normalizing the mean to obtain the trajectory deviation.

[0009] This invention uses mean square error normalization to calculate trajectory deviation. Compared with single-point sampling, this index can effectively reduce the impact of random noise on the judgment result, reflect the overall distortion degree and phase lag of the waveform, and provide a high-confidence logical benchmark for subsequent attractor construction.

[0010] Preferably, obtaining the mutation rate includes: calculating the absolute value of the first difference of the output power sequence, normalizing all absolute values ​​and calculating the mean of all normalized values, and using the mean as the mutation rate of the current window.

[0011] Preferably, the dynamic damping ratio satisfies the expression: In the formula, The dynamic damping ratio of the current window; Sensitivity coefficient; This represents the mutation rate of the current window.

[0012] This invention uses an inverse proportional function to map the waveform variation rate to a dynamic damping ratio. This mapping mechanism can sensitively capture the precursors of micro-oscillations caused by excessive parameters, providing an accurate physical basis for the subsequent priority execution of derating and vibration suppression strategies, and effectively preventing hardware overcurrent or machine failure caused by blindly increasing the gain.

[0013] Preferably, the baseline value of the trajectory deviation is obtained by: extracting data from each window of the PPEC controller under the factory nominal operating conditions, calculating the trajectory deviation of each window, and taking the average of the trajectory deviations of all extracted windows as the baseline value of the trajectory deviation.

[0014] Preferably, the evolution attractor satisfies the expression: In the formula, The current window's evolution attractor; , The trajectory deviation and dynamic damping ratio for the current window; This serves as the baseline value for trajectory deviation. These are the weighting coefficients; This is the first constituent term of the evolution attractor of the current window; This is the second constituent term of the evolution attractor of the current window.

[0015] This invention constructs an evolution attractor using weighted Euclidean distance, achieving vector synthesis of both trajectory distortion and physical energy instability. This ensures that the tuning step size can simultaneously respond to the requirements of large deviation following and small-amplitude oscillation suppression, enabling the controller to adaptively cope with structural deviations and energy divergences.

[0016] Preferably, determining the polarity coefficient includes: if the second constituent item is greater than the first constituent item, setting the polarity coefficient. If the second component is less than or equal to the first component, calculate the mean of the output power sequence within the current window. The mean of the modulation command sequence Calculate the polarity coefficient using the sign function , .

[0017] Preferably, the correction increment satisfies the expression: In the formula, The increment for the current window; The current window's evolution attractor; The steepness factor; This represents the polarity coefficient of the current window; Used as the base step size coefficient.

[0018] This invention achieves rapid correction by nonlinearly expanding the step size when the attractor is large and far from the steady state; and achieves fine locking by automatically collapsing the step size to the micro level when the attractor is small and close to the steady state. This nonlinear mapping, which adjusts large errors and finely adjusts small errors, ensures the recovery speed under high dynamic conditions and the control smoothness under steady state.

[0019] Preferably, the method further includes: writing the updated control parameters into the core storage unit corresponding to PPEC through a communication interface to achieve online adaptive optimization of controller parameters.

[0020] Secondly, the present invention provides an online parameter tuning and optimization system for a PPEC controller, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned online parameter tuning and optimization method for a PPEC controller is implemented.

[0021] By adopting the above technical solution, a computer program is generated from the above-mentioned method for online parameter tuning and optimization of PPEC controller, and stored in the memory so that it can be loaded and executed by the processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0022] The beneficial effects of this invention are as follows:

[0023] This invention, by analyzing the dominant relationship between physical energy instability and trajectory distortion intensity within a window, endows the PPEC controller with the physical perception capability of the causes of system instability. Utilizing an adaptive polarity switching mechanism, it achieves differentiated control for derating and stabilizing under oscillating conditions and for gain correction under error conditions. This strategy effectively avoids the risk of hardware oscillation caused by excessive parameters at the physical energy level, realizes oscillation-free adaptive optimization under complex nonlinear conditions, and significantly improves the dynamic tracking quality of power electronic equipment. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating an online parameter tuning and optimization method for a PPEC controller according to the present invention;

[0025] Figure 2 This is a schematic diagram illustrating the performance comparison of PPEC controllers. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0028] This invention discloses an online parameter tuning and optimization method for a PPEC controller, referring to... Figure 1 This includes steps S1-S5:

[0029] S1. Obtain the output power sequence and modulation command sequence within the current window.

[0030] It should be noted that when the PPEC controller performs high-frequency power conversion tasks, there is a strong real-time correlation between its modulation command and output power. Therefore, by extracting a fixed-length synchronization sequence pair from memory, the scattered power values ​​are transformed into time-correlated feature data, providing a data basis for subsequent parameter optimization.

[0031] Specifically, the PPEC hardware high-precision timer is used to configure a periodic interrupt source with a fixed sampling frequency. At each interrupt trigger moment, the CPU responds to the interrupt and triggers the ADC to perform synchronous sampling, atomically reads the current output power, and synchronously mirrors the reading of the modulation instructions in the core control register at this time. The set of synchronous data pairs is sequentially written into the current acquisition buffer and the buffer counter is updated. When the counter value reaches the preset sampling value, the buffer flip operation is immediately triggered: the current acquisition buffer is locked and marked as ready, serving as the input data source for the current window to obtain the modulation instruction sequence and output power sequence; at the same time, the pointer is switched to another idle buffer to continue data acquisition for the next second, ensuring full time-domain coverage of the control parameters during online tuning without any monitoring blind spots.

[0032] In this embodiment, the sampling frequency is set to 1kHz and the sampling value is set to 1000. The logic behind this setting is as follows: based on the characteristic that the PPEC controlled object is usually a 50Hz / 60Hz power frequency fundamental wave, setting the sampling frequency to 1kHz can ensure that about 20 key feature points are acquired in each fundamental wave cycle, which satisfies the Nyquist sampling theorem and is sufficient to ensure a clear restoration of the sine waveform contour; setting the sampling value to 1000 ensures coverage of a dynamic evolution process of up to 1 second to capture the low-frequency oscillation trend caused by system parameter mismatch. At the same time, the storage and computational overhead of 1000 data points is extremely low, which can adapt to the hardware resource limitations of most industrial-grade embedded control chips, ensuring the real-time performance and economy of the online tuning algorithm.

[0033] At this point, the output power sequence and modulation command sequence within the current window have been obtained.

[0034] S2. Determine the trajectory deviation based on the cumulative difference between the output power and the modulation command at each sampling moment within the current window; determine the dynamic damping ratio based on the changes in output power within the current window.

[0035] It should be noted that in the PPEC control system, the most essential characteristic of control parameter mismatch is the logical inaccuracy of the response waveform in the buffer relative to the command. When the system enters a metastable state due to improper parameters, the output power sequence in the buffer will exhibit oscillations that are out of the command constraint. Therefore, it is necessary to analyze the cumulative difference between the output power and the modulation command in the current window to identify the nonlinear distortion caused by the register parameter offset.

[0036] Specifically, the trajectory deviation of the current window is determined based on the cumulative difference between the output power and the modulation command at each sampling moment within the current window; the trajectory deviation satisfies the expression:

[0037]

[0038] In the formula, The deviation of the trajectory of the current window; For the current window Output power at each sampling time; For the current window Modulation instructions for each sampling time; , This refers to the index value and total number of sampling times within the window; For the standard normalization function, by Divide by the square of the maximum rated power of the PPEC system to map it to the dimensionless interval [0,1].

[0039] in, Reflected within the current window, this value indicates the degree of deviation of the actual output power waveform from the overall trajectory of the ideal modulation command waveform. The larger the value, the more severe the accumulation of tracking error in the time domain, and the more significant the distortion, phase lag, or amplitude attenuation of the feedback waveform. This means that the control parameters in the current register cannot adapt to the current dynamic load characteristics, causing the system to be on the verge of parameter mismatch or loss of control. Conversely, a smaller value indicates that the output power waveform can accurately reproduce the geometric trajectory of the modulation command with extremely low error energy, meaning that the current control parameters are highly matched with the load characteristics, and the system is in an ideal steady-state following mode.

[0040] It should be further noted that since PPEC controllers often face two types of logic inaccuracies under dynamic loads, including tracking errors and energy instability, it is also necessary to analyze whether it belongs to the case of energy instability. In essence, the damping ratio characterizes the system's ability to suppress transient oscillations. The dynamic damping ratio can be calculated by the change of the output power waveform to characterize the current energy stability of the system and prevent hardware oscillations caused by excessive parameters.

[0041] Specifically, the mutation rate of the current window is obtained by calculating the absolute value of the first difference of the output power sequence within the current window, performing max-min normalization on all absolute values, and calculating the mean of all normalized values. The mean is then used as the mutation rate of the current window.

[0042] The variation rate of the current window is mapped to a dynamic damping ratio using an inverse proportional function, and the dynamic damping ratio satisfies the expression:

[0043]

[0044] In the formula, The dynamic damping ratio of the current window; Sensitivity coefficient; This represents the mutation rate of the current window.

[0045] If the current window is in an ideal steady state, the waveform of the output power is smooth, and the variation rate of the current window consists only of the sensor's own noise floor. In this case, the obtained... A value approaching 1 indicates that the system within the current window is sufficiently damped and its energy converges. If the system within the current window exhibits underdamped oscillations, the output will fluctuate wildly at high frequencies, leading to a sharp increase in the variability rate. The rapid decay towards 0 indicates insufficient system damping within the current window, suggesting a risk of divergence.

[0046] It should be added that the sensitivity coefficient is a key parameter used to balance noise immunity and sensitivity. If the sensitivity coefficient is too small, the amplification effect on high-frequency variation rate will be insufficient, which may cause the system to fail to identify micro-oscillations in time. If the sensitivity coefficient is too large, normal switching ripple or sensor white noise will be misjudged as instability characteristics, resulting in an abnormally low dynamic damping ratio and causing unnecessary parameter derating. Therefore, the sensitivity coefficient is usually set to the range of [1, 10]. In this embodiment, the sensitivity coefficient is set to 5, which can ensure that the algorithm maintains a very high response speed to abnormal oscillations exceeding the safety threshold while filtering out the inherent switching ripple of PPEC, thereby accurately mapping the current energy stability. The implementer can adjust it according to the service life of the controller and the control accuracy.

[0047] At this point, the trajectory deviation and dynamic damping ratio of the current window are obtained.

[0048] S3. Construct an evolution attractor by utilizing the difference between the trajectory deviation and the baseline value of the trajectory deviation, as well as the dynamic damping ratio.

[0049] It should be noted that traditional fixed tuning step size adjustment cannot distinguish between structural parameter deviations and energy convergence defects, which can easily lead to adjustment hysteresis or overshoot at the edge of instability. Therefore, an evolution attractor index is introduced to transform discrete waveform distortion characteristics into continuous state space displacement, thereby constructing a follow-up mechanism for the tuning step size to ensure that the tuning force can be spontaneously adjusted according to the degree of deviation of the system from the steady state.

[0050] Specifically, the baseline value for trajectory deviation is obtained by extracting data from each window of the PPEC controller under the factory nominal operating conditions, calculating the trajectory deviation of each window, and calculating the average value of the trajectory deviation of all extracted windows, using the average value as the baseline value for trajectory deviation.

[0051] The difference between the trajectory deviation of the current window and the baseline value of the trajectory deviation, as well as the dynamic damping ratio, are used as two components in constructing the evolutionary attractor of the current window; the evolutionary attractor satisfies the expression:

[0052]

[0053] In the formula, The current window's evolution attractor; , The trajectory deviation and dynamic damping ratio for the current window; This serves as the baseline value for trajectory deviation. These are the weighting coefficients.

[0054] in, The first component of the evolution attractor of the current window reflects the trajectory distortion intensity of the output power sequence in the current window relative to the modulation command. The larger the value, the more disordered the output power in the current window is, which means that the core internal model parameters of the controller have failed. The second constituent term of the evolution attractor of the current window reflects the oscillatory instability potential of the system at the physical energy level within the current window. The smaller, The larger the value, the more likely it is that underdamped oscillations have occurred within the current window, accumulating a large amount of oscillating potential energy that has not been dissipated in time, and the parameters of the control loop are severely mismatched. Therefore, if the system in the current window exhibits both large waveform trajectory distortion intensity and physical energy instability, or if one of them is large, it means that the traditional fixed tuning step size has completely failed and needs to be corrected.

[0055] It should be added that the weighting coefficient is used to weigh the weight ratio of system trajectory deviation and dynamic damping ratio in the tuning decision. Its value is usually determined by the physical parameters of the converter's output inductance and capacitance. If the weighting coefficient is set too small, the system will not be able to identify energy instability caused by residual oscillations, which will cause the tuning process to ignore the risk of oscillation. If the weighting coefficient is set too large, the adjustment process will excessively sacrifice response speed in order to eliminate small oscillations. Therefore, the value range of the weighting coefficient is [1, 5]. In this embodiment, the coefficient is set to 3, which can ensure that the parameter evolution efficiency is maximized while suppressing tuning oscillations. In other embodiments, the implementer can set it in the range of [1, 5] according to the system power level.

[0056] At this point, the evolution attractor of the current window has been obtained.

[0057] S4. Compare the magnitudes of the constituent terms of the evolution attractor to determine the polarity coefficient; based on the evolution attractor and the polarity coefficient, determine the correction increment.

[0058] It should be noted that during parameter tuning, it is crucial to accurately distinguish between two different failure modes: steady-state error and oscillation instability. When the control system oscillates, adjusting solely based on the sign of the error in the output power and modulation command can easily lead to an erroneous increase in gain, resulting in the risk of positive feedback failure. Therefore, an adaptive polarity switching mechanism is introduced to guide the correction increment: when the evolving attractor is mainly caused by insufficient damping, i.e., oscillation, the step size is forced to reverse to weaken the gain; when the evolving attractor is mainly caused by the intensity of trajectory distortion, i.e., the error in the tracking output power and modulation command, the step size direction is determined based on the sign of the error to eliminate the deviation, thereby improving the control effect.

[0059] Specifically, by comparing the magnitudes of the two constituent terms of the evolution attractor in the current window, the dominant failure mode of the system within the current window is determined:

[0060] If the second constituent term of the evolution attractor in the current window is greater than the first constituent term, it is determined to be oscillation-dominant, indicating that the physical energy instability of the system within the current window is higher than the trajectory distortion intensity, and a polarity coefficient is set. If the second component of the evolution attractor in the current window is less than or equal to the first component of the evolution attractor in the current window, it is determined that the output power and modulation command error dominate, indicating that the system mainly faces tracking accuracy problems within the current window. The mean of the output power sequence within the current window is then calculated. The mean of the modulation command sequence Calculate the polarity coefficient using the sign function , .

[0061] Based on the evolution attractor and polarity coefficient of the current window, determine the correction increment for the current window; the correction increment satisfies the expression:

[0062]

[0063] In the formula, The increment for the current window; The current window's evolution attractor; The steepness factor; This represents the polarity coefficient of the current window; Used as the base step size coefficient.

[0064] When the system is in a state of oscillation, the correction increment is always negative, which means that once insufficient waveform damping is detected, the derating strategy will be executed first to quickly reduce the control parameters to suppress hardware oscillation. This reflects the spontaneous extension of the current window adjustment step size as the system becomes more unstable. The larger the value, the greater the risk of instability detected by the evolution attractor within the current window. The correction increment of the function output will expand nonlinearly and rapidly, and the system needs to quickly compensate for the parameter gap with an extremely high dynamic response speed. Conversely, the smaller the value, the more the system has returned to the ideal steady-state center. The correction action automatically drops to an extremely small level when approaching the optimal parameter point, thereby eliminating the reciprocating jump oscillations of the fixed tuning step size near the equilibrium point. The basic step size coefficient is a scaling factor that maps the dimensionless feature intensity of the algorithm output to the physical dimension of the controller target register. It also acts as a safety limiter for the single adjustment amplitude. In this embodiment, the basic step size coefficient is set to 1% of the average value of the register parameter values ​​within the current window to achieve dynamic safety adaptation of the adjustment step size.

[0065] It should be added that the kurtosis factor is used to control the responsiveness of the adjustment mode. If the kurtosis factor is set too small, the step size changes too smoothly with the evolution attractor, resulting in a longer tuning recovery time under high dynamics. If the kurtosis factor is set too large, the step size switches too abruptly near the threshold point, which may cause subharmonic oscillations in the control loop. Therefore, the value range of the kurtosis factor is [5, 20]. In this embodiment, the kurtosis factor is set to 10, which can achieve smooth adaptive step size over a wide power range. Implementers can make fine adjustments based on the effect of the adjustment.

[0066] At this point, the correction increment for the current window has been obtained.

[0067] S5. Obtain the average value of the real-time control parameters of the registers in the current window, sum the correction increment with the average value to obtain the updated control parameters, thereby realizing the online optimization of the PPEC controller.

[0068] It should be noted that the current window's correction increment is used as the execution operator and applied to the target register inside the PPEC via the Modbus protocol. By using incremental superposition logic, the controller parameters can be adjusted in real time, thereby solving the accuracy loss and oscillation problems caused by fixed tuning step size at the physical level.

[0069] Specifically, the average value of the real-time control parameters of the registers within the current window is obtained, and the average value is summed with the correction increment of the current window to obtain the updated control parameters of the current window. The updated control parameters are then written into the core storage unit corresponding to the PPEC through the communication interface to achieve online adaptive optimization of the controller parameters.

[0070] The real-time control parameters of the registers within the current window are read in real time from the underlying registers via the Modbus communication link.

[0071] For example, Figure 2 The graph shows a performance comparison of the PPEC controller, with the horizontal axis representing time and the vertical axis representing power. The region exhibiting multiple large step jumps in the real-time modulation command reflects the typical nonlinear operating condition of the power electronic system under rapid dynamic load switching. Traditional control methods, when faced with large-amplitude modulation command switching, suffer from low adjustment efficiency due to their fixed tuning step size mechanism, failing to keep pace with output power fluctuations in real time, resulting in significant steady-state deviations from the modulation command. In contrast, the method of this invention, employing incremental superposition logic, can capture the precursors of modulation command jumps, significantly reducing the deviation between output power and the modulation command. The performance improvement region demonstrates that this invention significantly suppresses tracking deviation compared to traditional methods.

[0072] This invention also discloses an online parameter tuning and optimization system for a PPEC controller, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement an online parameter tuning and optimization method for a PPEC controller according to the present invention.

[0073] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

Claims

1. A PPEC controller online parameter tuning and optimization method, characterized in that, include: Obtain the output power sequence and modulation command sequence within the current window; The trajectory deviation is determined based on the cumulative difference between the output power and the modulation command at each sampling moment within the current window; the variation rate is obtained based on the change in output power within the current window; and the variation rate is mapped to the dynamic damping ratio using an inverse proportional function. The difference between the trajectory deviation and the baseline value of the trajectory deviation, and the dynamic damping ratio, are respectively used as two components for constructing the evolution attractor; The evolutionary attractor satisfies an expression: wherein, is an evolutionary attractor of a current window; , is a trajectory deviation and a dynamic damping ratio of the current window; is a reference value of the trajectory deviation; is a weight coefficient; is a first component of the evolutionary attractor of the current window; is a second component of the evolutionary attractor of the current window; By comparing the magnitudes of the two constituent terms of the evolution attractor, the polarity coefficient is determined; based on the evolution attractor and the polarity coefficient, the correction increment is determined. The average value of the real-time control parameters of the registers within the current window is obtained, and the correction increment is summed with the average value to obtain the updated control parameters, thereby realizing the online optimization of the PPEC controller.

2. The PPEC controller online parameter tuning and optimization method of claim 1, wherein, The determination of trajectory deviation includes: Calculate the square of the difference between the output power and the modulation command at each sampling moment within the current window, obtain the mean of the squares of all the differences within the current window, and normalize the mean to obtain the trajectory deviation.

3. The method of online parameter tuning and optimization of a PPEC controller according to claim 1, wherein, The acquisition of the mutation rate includes: Calculate the absolute value of the first difference of the output power sequence, normalize all absolute values ​​and calculate the mean of all normalized values, and use the mean as the variation rate of the current window.

4. The method of online parameter tuning and optimization of a PPEC controller according to claim 1, wherein, The dynamic damping ratio satisfies the expression: ； wherein is the dynamic damping ratio of the current window; is the sensitivity coefficient; is the variability rate of the current window.

5. The PPEC controller online parameter tuning and optimization method of claim 1, wherein, The baseline value of the trajectory deviation is obtained as follows: Extract data from each window of the PPEC controller under the factory nominal operating conditions, calculate the trajectory deviation of each window, and use the average of the trajectory deviations of all extracted windows as the baseline value of the trajectory deviation.

6. The PPEC controller online parameter tuning and optimization method of claim 1, wherein, The determination of the polarity coefficient includes: If the second constituent item is greater than the first constituent item, set the polarity coefficient. If the second component is less than or equal to the first component, calculate the mean of the output power sequence within the current window. The mean of the modulation command sequence Calculate the polarity coefficient using the sign function , .

7. The method for online parameter tuning and optimization of a PPEC controller according to claim 1, characterized in that, The correction increment satisfies the expression: ； In the formula, The increment for the current window; The current window's evolution attractor; The steepness factor; This represents the polarity coefficient of the current window; Used as the base step size coefficient.

8. The method for online parameter tuning and optimization of a PPEC controller according to claim 1, characterized in that, The method further includes: The updated control parameters are written into the core storage unit corresponding to PPEC through the communication interface, thereby realizing online adaptive optimization of the controller parameters.

9. A PPEC controller online parameter tuning and optimization system, characterized in that, include: The processor and memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a method for online parameter tuning and optimization of a PPEC controller according to any one of claims 1-8.

Images