MPPT sampling error compensation method and system
By combining dynamic sliding window sampling and hierarchical median filtering algorithms with variable step size perturbation observation method, the problem of sampling noise suppression in photovoltaic systems is solved, achieving high-precision maximum power point tracking and fast response, which is applicable to different photovoltaic systems.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-10
- Publication Date
- 2026-03-13
AI Technical Summary
Existing technologies struggle to effectively suppress sampling noise in photovoltaic systems, resulting in low maximum power point tracking accuracy, particularly poor performance in dynamic environments. Furthermore, the high computational complexity makes it difficult to achieve high-precision real-time filtering on low-power microcontrollers.
A dynamic sliding window sampling combined with a tiered median filtering algorithm driven by the coefficient of variation is adopted. By dynamically adjusting the sampling window size and the tiered filtering strategy, and combining the variable step size perturbation observation method, adaptive suppression and accurate tracking of sampling noise are achieved.
It significantly improves the tracking accuracy and dynamic response speed of the maximum power point, adapts to the control accuracy under complex operating conditions, reduces the performance requirements of the processor, and is suitable for various photovoltaic systems.
Smart Images
Figure CN121664196A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data filtering technology, and more specifically to an MPPT sampling error compensation method and system. Background Technology
[0002] Currently, improving power generation efficiency has become a critical issue that the industry urgently needs to address during the large-scale application of photovoltaic systems. In the maximum power point tracking (MPPT) control of photovoltaic power generation systems, sensor sampling error is a key factor affecting tracking accuracy. These errors mainly originate from the inherent fluctuations in the output characteristics of photovoltaic modules, environmental interference, and quantization errors during analog-to-digital conversion. Existing technologies generally employ hardware filtering or software filtering methods with fixed parameters to suppress noise interference.
[0003] However, hardware filtering methods introduce hysteresis in the control loop, causing power point tracking delays under dynamic conditions of rapidly changing illumination, while also increasing circuit complexity and system cost. Traditional software filtering methods, using fixed parameters, struggle to adapt to varying sampling noise levels and complex, changing environmental conditions. In situations with excessive noise or sudden changes in illumination, a trade-off arises between filtering effectiveness and dynamic response speed, failing to balance steady-state accuracy and dynamic performance. Although solutions based on advanced intelligent algorithms such as deep learning have emerged in recent years, these methods typically have high computational complexity, requiring high-performance processors for real-time processing, making them difficult to apply in resource-constrained embedded systems. Current research focuses primarily on optimizing the maximum power point tracking algorithm itself, lacking a systematic solution for dynamic compensation of sampling errors. Achieving high-precision real-time filtering on low-power microcontrollers remains a significant technical challenge, necessitating the development of novel adaptive filtering methods that balance computational efficiency and filtering performance to meet the real-time control requirements of photovoltaic systems under different operating conditions.
[0004] Therefore, how to achieve adaptive suppression of sampling noise and thus improve the tracking accuracy of the maximum power point is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0005] In view of the above problems, the present invention is proposed to provide an MPPT sampling error compensation method and system that overcomes or at least partially solves the above problems, realizes adaptive suppression of sampling noise, and improves the tracking accuracy of the maximum power point.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, embodiments of the present invention provide an MPPT sampling error compensation method, comprising: Acquire voltage or current data collected by the target sensor as target data; Based on the target data, dynamic sliding window sampling is used to obtain multiple window data; Based on the mean and variance of the data obtained from the window data, outlier data is removed to obtain valid window data; The coefficient of variation is obtained based on the data mean and the data variance. Based on the coefficient of variation, a hierarchical filtering strategy is applied to the effective window data to obtain denoised data; The current step size is determined based on the coefficient of variation. The duty cycle of the control IGBT drive signal is obtained by combining the current step size and the denoised data with the variable step size perturbation observation method.
[0007] In one embodiment, the sampling window size of the dynamic sliding window is adjusted based on the real-time power gradient acquired by the target sensor, and the sampling window adjustment rule is as follows: ; in, W size This indicates the adjusted sampling window size. dP / dt Represents the power gradient. dP This represents the change in power. dt Indicates the change over time. P rated Indicates the rated power of the DC / DC converter.
[0008] In one embodiment, obtaining valid window data specifically includes: The average value is obtained by calculating the average value based on the window data; The standard deviation of the data is calculated based on the mean of the data. The filtering range is set based on the data mean and the data standard deviation; The valid window data is obtained by removing the abnormal data that exceeds the filtering range based on the window data. The filtering range is: [μ] [3σ,μ+3σ], where μ represents the data mean and σ represents the data variance.
[0009] In one embodiment, it also includes: Determine whether the amount of data in the running window is less than a first preset value; If so, the sampling window is dynamically expanded to a second preset value, and the abnormal data is removed again; Otherwise, continue with the following steps.
[0010] In one embodiment, the hierarchical filtering strategy is specifically as follows: Determine whether the coefficient of variation is less than or equal to a first threshold; If so, then perform single-point median filtering on the effective window data to obtain the denoised data; Otherwise, determine whether the coefficient of variation is less than or equal to the second threshold; If so, then perform weighted median filtering on the effective window data to obtain the denoised data; Otherwise, cascaded median filtering is performed on the effective window data to obtain the denoised data.
[0011] In one embodiment, the weighted median filtering specifically refers to: The data is sorted in descending order based on the size of the effective window data to obtain the sorted data. The denoised data is obtained by weighting the sorted data according to the Pascal's triangle value corresponding to the number of rows of the sorted data by dividing it by a set value.
[0012] In one embodiment, the cascaded median filtering specifically comprises: Based on the current valid window data, a median value is taken every preset number of times to obtain the first new array; Based on the next valid window data, a median value is taken every preset number of times to obtain a second new array; And so on, to obtain a new array of the preset number; The denoised data is obtained by reorganizing and sorting all the values in the new array and taking the median value.
[0013] In one embodiment, it also includes: Obtain the slope of the anomaly coefficients for a set number of the sampling windows; When the slope is greater than a set threshold, the weighted median filtering is triggered.
[0014] In one embodiment, the formula for determining the current step size is: ; in, U n+1 This represents the size of the (n+1)th perturbation step, i.e., the current step size. K base Indicates the basic step size. CV Represents the coefficient of variation. dV It represents the amount of voltage change. U n This represents the size of the nth perturbation step.
[0015] In a second aspect, embodiments of the present invention provide an MPPT sampling error compensation system, comprising: a data acquisition module, a sliding sampling module, an anomaly removal module, a hierarchical filtering module, and a control output module; The data acquisition module is used to acquire voltage or current data collected by the target sensor as target data; The sliding sampling module is used to obtain multiple window data by using dynamic sliding window sampling based on the target data; The anomaly removal module is used to remove abnormal data based on the data mean and data variance obtained from the window data, so as to obtain valid window data. The hierarchical filtering module is used to obtain the coefficient of variation based on the data mean and the data variance; and to apply a hierarchical filtering strategy to the effective window data based on the coefficient of variation to obtain denoised data. The control output module is used to determine the current step size based on the coefficient of variation; and to obtain the duty cycle of the control IGBT drive signal based on the current step size and the denoised data combined with the variable step size perturbation observation method.
[0016] As can be seen from the above technical solution, compared with the prior art, the present invention discloses an MPPT sampling error compensation method and system, which has the following beneficial effects: 1. In terms of signal processing, this invention adopts a dynamic sliding window-based filtering technique, combined with a graded median filtering algorithm driven by the coefficient of variation, to achieve adaptive suppression of sampling noise. In terms of control algorithm, this invention develops a dynamic step size adjustment strategy with adaptive CV value and a multi-peak fast search mechanism, which significantly improves the dynamic response speed of the system and the control accuracy under complex working conditions.
[0017] 2. Under conditions without light sensors, this invention utilizes power gradient analysis and dynamic sliding window technology to achieve precise adaptation to complex working conditions, ensuring rapid response in steady state and effectively suppressing noise interference during sudden environmental changes.
[0018] 3. The intelligent evaluation mechanism based on the coefficient of variation of this invention can predict the fluctuation trend in advance, and eliminate various sampling errors through multi-level filtering strategies, so that the system can maintain high tracking accuracy while exhibiting excellent dynamic performance; the improved disturbance observation algorithm combined with adaptive step size adjustment effectively avoids the power oscillation problem of traditional methods.
[0019] 4. The method of the present invention does not require an external hardware filtering circuit, has low requirements for processor performance, and has good compatibility, which can be flexibly adapted to photovoltaic systems of different specifications. The comprehensive performance advantages of the method of the present invention enable it to maintain stable and efficient operation in various harsh environments, providing a reliable solution for the intelligent control of photovoltaic power generation systems. Attached Figure Description
[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0021] Figure 1 The present invention provides a flowchart of an MPPT sampling error compensation method.
[0022] Figure 2 The flowchart of the hierarchical filtering strategy method provided by the present invention is shown.
[0023] Figure 3 This is a schematic diagram of an MPPT sampling error compensation system provided by the present invention. Detailed Implementation
[0024] 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 embodiments of the present invention, and not all embodiments. 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.
[0025] Example 1 Traditional MPPT algorithms suffer from power calculation errors due to noise interference from voltage / current sensors, quantization errors in AD conversion, and fluctuations in sampled signals caused by sudden environmental changes (such as cloud cover). This leads to misjudgments in the MPPT algorithm: continuous power point oscillations under steady-state conditions and tracking lag under dynamic lighting conditions. More seriously, the multi-peak characteristics caused by local shading are amplified by sampling errors, causing the system to get stuck in local extrema and resulting in power generation losses of up to 20%. In addition, high-frequency switching noise may trigger false protection actions, affecting system reliability. These problems make it difficult for existing photovoltaic systems to reach their theoretical maximum output power in actual operation, thus limiting the economic benefits of photovoltaic power generation.
[0026] Existing technologies using hardware filtering circuits (such as RC low-pass filters) introduce phase delay, affecting the dynamic response speed of MPPT; software filtering methods with fixed thresholds cannot adapt to the fluctuation characteristics under different light intensities, resulting in under-filtering under light load or strong light.
[0027] To address the aforementioned issues, this invention employs a dynamic weighted hybrid filtering algorithm and a parameter adaptive adjustment mechanism to suppress sampling noise while maintaining the MPPT response speed. The specific solution is as follows: like Figure 1 As shown, this embodiment of the invention discloses an MPPT sampling error compensation method, including the following steps. For ease of description, these steps are numbered S1 to S7, which are not used to limit the sequential relationship between the various steps of this invention: S1 acquires the voltage or current data collected by the target sensor as the target data.
[0028] Taking a 50kW photovoltaic system as an example, the voltage or current data obtained from the voltage or current sensors installed in the system are used as target data for subsequent processing steps.
[0029] S2 uses dynamic sliding window sampling based on the target data to obtain multiple window data.
[0030] Taking a 50kW photovoltaic system as an example, the processor of the control equipment usually uses a single-core MCU, which triggers the control algorithm through multiple interrupt functions. Current and voltage sampling is generally carried out in high-speed interrupt functions, and the frequency is generally much higher than the execution frequency of the MPPT algorithm (more than 100 times). Therefore, dynamic sliding window sampling can be used to sample voltage or current data to obtain multiple window data.
[0031] Furthermore, the sampling window size of the dynamic sliding window sampling is adjusted based on the real-time power gradient acquired by the target sensor. The sampling window adjustment rule is as follows: ; in, W size This indicates the adjusted sampling window size. dP / dt Represents the power gradient. dP This represents the change in power. dt Indicates the change over time. P rated This indicates the rated power of the DC / DC converter. In practical applications, a threshold calibration is required based on the rated power.
[0032] Furthermore, power gradient dP / dt for: ; in, P k express k Power sample value at time 10:00 P k-1 express k The power sample value at time -1 t Indicates the sampling period.
[0033] S3 uses the mean and variance of the data obtained from the window data to remove outliers and obtain valid window data.
[0034] Furthermore, obtaining valid window data specifically includes: The average value of the data is obtained by calculating the average value based on the window data. The standard deviation of the data is calculated based on the mean of the data. Set the filtering range based on the data mean and data standard deviation; By removing outlier data that exceeds the filtering range based on the window data, valid window data is obtained. The filtering range is: [μ [3σ,μ+3σ], where μ represents the data mean and σ represents the data variance.
[0035] Furthermore, the data mean μ Specifically: ; in, m Indicates the total amount of data in the window. x i This represents the data in the i-th window.
[0036] Furthermore, data standard deviation σ Specifically: .
[0037] Furthermore, it also includes: Determine if the amount of data in the running window is less than a first preset value; If so, the sampling window is dynamically expanded to the second preset value, and abnormal data is removed again; Otherwise, continue with the following steps.
[0038] In this embodiment, the first preset value is set to 3, and the second preset value is set to 7.
[0039] S4 obtains the coefficient of variation based on the data mean and data variance.
[0040] Furthermore, the coefficient of variation CV Specifically: .
[0041] S5 uses a hierarchical filtering strategy on the effective window data based on the coefficient of variation to obtain denoised data.
[0042] Furthermore, such as Figure 2 As shown, the hierarchical filtering strategy is as follows: Determine whether the coefficient of variation is less than or equal to the first threshold; If so, then perform single-point median filtering on the valid window data to obtain denoised data; Otherwise, determine whether the coefficient of variation is less than or equal to the second threshold; If so, then perform weighted median filtering on the valid window data to obtain denoised data; Otherwise, cascaded median filtering is performed on the valid window data to obtain denoised data.
[0043] Furthermore, in this embodiment, the first threshold is set to 0.1 and the second threshold is set to 1.
[0044] Furthermore, single-point median filtering specifically involves: We directly take the median of the effective window data as the denoised data to minimize latency.
[0045] Furthermore, the weighted median filtering is specifically as follows: Sort the data in descending order based on the size of the valid window data to obtain the sorted data; The denoised data is obtained by weighting the sorted data according to the Pascal's triangle value corresponding to the number of rows of the sorted data by dividing it by a set value.
[0046] Furthermore, in this embodiment, the set value is set to 2. (m-1) Assuming the sorted data is [X1, X2, X3, X4, X5], sorted by weight [1 / 2] 4 4 / 2 4 ,6 / 2 4 4 / 2 4 ,1 / 2 4 Interpolation yielded: 1 / 2 4 *X1+4 / 2 4 *X2+6 / 2 4 *X3+4 / 2 4 *X4+1 / 2 4 *X5.
[0047] Furthermore, the cascaded median filtering specifically involves: Based on the current valid window data, take the median value every preset number of times to obtain the first new array; Based on the data of the next valid window, take the median value every preset number of times to obtain a second new array; And so on, to obtain a new array of the preset number; The denoised data is obtained by re-sorting all the values in the new array and taking the median.
[0048] Furthermore, this embodiment employs a three-level cascaded median filter, with a preset quantity of 3 and a preset number of samples set to 5. The window sampling and calculation procedures are performed in two interrupt functions. When the calculation function is executed, the window sampling is locked, and the sampling window is unlocked after the current statistics are completed. When the interrupt function of the calculation function is entered again, the window has a new set of data, and the same process is repeated.
[0049] Furthermore, it also includes: Obtain the slope of the anomaly coefficients for a set number of sampling windows; When the slope is greater than the set threshold, weighted median filtering is triggered.
[0050] In this embodiment, the quantity is set to 3 and the threshold is set to 0.1. When the slope is greater than the set threshold of 0.1, it indicates that the environmental fluctuation is aggravated, and the strong filtering mode, i.e., weighted median filtering, is triggered in advance to avoid tracking lag.
[0051] S6 determines the current step size based on the coefficient of variation.
[0052] The current step size of the perturbation-observation method is determined based on the coefficient of variation (CV). The formula for determining the current step size is as follows: ; in, U n+1 This represents the size of the (n+1)th perturbation step, i.e., the current step size. K base Indicates the basic step size. CV Represents the coefficient of variation. dV It represents the amount of voltage change. U n This represents the size of the nth perturbation step; The magnitude of the perturbation step size can be obtained from the above formula, and the direction of the perturbation step size is from... sgn ( dP / dV The value of ) determines the outcome; if it occurs 10 times consecutively... dP If the power is approximately 0 but the expected power is not achieved, the global scan algorithm is invoked to relocate the maximum power point (Maximum PowerPoint).
[0053] S7 obtains the duty cycle of the control IGBT drive signal based on the current step size and denoised data combined with the variable step size perturbation observation method.
[0054] The denoised data is input into the variable step size perturbation observation method. The output result is added to the current step size to obtain the current MPPT algorithm output result, which is the target voltage value on the input side (photovoltaic side) of the converter. The target voltage value is sent into the voltage loop and enters the dual-loop control. The duty cycle of the control IGBT drive signal is obtained through voltage-current dual-loop calculation to ensure stable output.
[0055] Example 2 like Figure 3 As shown, based on the same inventive concept, the present invention also provides an MPPT sampling error compensation system, including: a data acquisition module, a sliding sampling module, an anomaly removal module, a hierarchical filtering module, and a control output module; The data acquisition module is used to acquire voltage or current data collected by the target sensor as target data; The sliding sampling module is used to obtain multiple window data by dynamically sliding window sampling based on the target data; The anomaly removal module is used to remove outlier data based on the mean and variance of the data obtained from the window data, thus obtaining valid window data. The hierarchical filtering module is used to obtain the coefficient of variation based on the data mean and variance; and to apply a hierarchical filtering strategy to the effective window data based on the coefficient of variation to obtain denoised data. The control output module is used to determine the current step size based on the coefficient of variation; based on the current step size and denoised data combined with the variable step size perturbation observation method, the duty cycle of the control IGBT drive signal is obtained.
[0056] Furthermore, in this embodiment, the functional implementation methods of each functional module correspond one-to-one with the methods described above, and will not be repeated here.
[0057] Example 3 Based on the same inventive concept, the present invention also provides an electronic device, which includes a processor and a memory, wherein the memory stores instructions, characterized in that the instructions are loaded and executed by the processor to implement an MPPT sampling error compensation method as in Embodiment 1.
[0058] Based on the same inventive concept, the present invention also provides a computer device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When the processor executes a program stored in the memory, it can implement an MPPT sampling error compensation method as shown in Example 1.
[0059] The electronic device may include a processor, a communications interface, a memory, and a communication bus, wherein the processor, communications interface, and memory communicate with each other via the communication bus. The processor can invoke logical instructions in the memory to execute an MPPT sampling error compensation method as described in Embodiment 1.
[0060] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0061] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.
[0062] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. An MPPT sampling error compensation method, characterized in that, include: Acquire voltage or current data collected by the target sensor as target data; Based on the target data, dynamic sliding window sampling is used to obtain multiple window data; Based on the mean and variance of the data obtained from the window data, outlier data is removed to obtain valid window data; The coefficient of variation is obtained based on the data mean and the data variance. Based on the coefficient of variation, a hierarchical filtering strategy is applied to the effective window data to obtain denoised data; The current step size is determined based on the coefficient of variation. The duty cycle of the control IGBT drive signal is obtained by combining the current step size and the denoised data with the variable step size perturbation observation method.
2. The MPPT sampling error compensation method according to claim 1, characterized in that, The sampling window size of the dynamic sliding window is adjusted based on the real-time power gradient acquired by the target sensor. The sampling window adjustment rule is as follows: ； in, W size This indicates the adjusted sampling window size. dP / dt Represents the power gradient. dP This represents the change in power. dt Indicates the change over time. P rated Indicates the rated power of the DC / DC converter.
3. The MPPT sampling error compensation method according to claim 1, characterized in that, Obtaining valid window data specifically includes: The average value is obtained by calculating the average value based on the window data; The standard deviation of the data is calculated based on the mean of the data. The filtering range is set based on the data mean and the data standard deviation; The valid window data is obtained by removing the abnormal data that exceeds the filtering range based on the window data. The filtering range is: [μ] [3σ,μ+3σ], where μ represents the data mean and σ represents the data variance.
4. The MPPT sampling error compensation method according to claim 3, characterized in that, Also includes: Determine whether the amount of data in the running window is less than a first preset value; If so, the sampling window is dynamically expanded to a second preset value, and the abnormal data is removed again; Otherwise, continue with the following steps.
5. The MPPT sampling error compensation method according to claim 1, characterized in that, The hierarchical filtering strategy is specifically as follows: Determine whether the coefficient of variation is less than or equal to a first threshold; If so, then perform single-point median filtering on the effective window data to obtain the denoised data; Otherwise, determine whether the coefficient of variation is less than or equal to the second threshold; If so, then perform weighted median filtering on the effective window data to obtain the denoised data; Otherwise, cascaded median filtering is performed on the effective window data to obtain the denoised data.
6. The MPPT sampling error compensation method according to claim 5, characterized in that, The weighted median filtering specifically refers to: The data is sorted in descending order based on the size of the effective window data to obtain the sorted data. The denoised data is obtained by weighting the sorted data according to the Pascal's triangle value corresponding to the number of rows of the sorted data by dividing it by a set value.
7. The MPPT sampling error compensation method according to claim 5, characterized in that, The cascaded median filtering specifically refers to: Based on the current valid window data, a median value is taken every preset number of times to obtain the first new array; Based on the next valid window data, a median value is taken every preset number of times to obtain a second new array; And so on, to obtain a new array of the preset number; The denoised data is obtained by reorganizing and sorting all the values in the new array and taking the median value.
8. The MPPT sampling error compensation method according to claim 4 or 5, characterized in that, Also includes: Obtain the slope of the anomaly coefficients for a set number of the sampling windows; When the slope is greater than a set threshold, the weighted median filtering is triggered.
9. The MPPT sampling error compensation method according to claim 2, characterized in that, The formula for determining the current step size is: ； in, U n+1 This represents the size of the (n+1)th perturbation step, i.e., the current step size. K base Indicates the basic step size. CV Represents the coefficient of variation. dV It represents the amount of voltage change. U n This represents the size of the nth perturbation step.
10. An MPPT sampling error compensation system, used to perform an MPPT sampling error compensation method as described in any one of claims 1-9, characterized in that, include: Data acquisition module, sliding sampling module, anomaly removal module, hierarchical filtering module, control output module; The data acquisition module is used to acquire voltage or current data collected by the target sensor as target data; The sliding sampling module is used to obtain multiple window data by using dynamic sliding window sampling based on the target data; The anomaly removal module is used to remove abnormal data based on the data mean and data variance obtained from the window data, so as to obtain valid window data. The hierarchical filtering module is used to obtain the coefficient of variation based on the data mean and the data variance; and to apply a hierarchical filtering strategy to the effective window data based on the coefficient of variation to obtain denoised data. The control output module is used to determine the current step size based on the coefficient of variation; and to obtain the duty cycle of the control IGBT drive signal based on the current step size and the denoised data combined with the variable step size perturbation observation method.