Household photovoltaic power generation grid-connected system DC module

Abstract

The invention discloses a direct current module of a household photovoltaic power generation grid-connected system, particularly relates to the technical field of household photovoltaic power generation grid connection, and is used for solving the problem that a global maximum power point is mistakenly locked due to voltage coupling when multiple modules are connected in parallel in a dynamic shadow environment. The method comprises the following steps: acquiring output voltage of a photovoltaic panel and voltage of a direct-current bus in real time to generate voltage gradient change data; frequency domain energy distribution analysis is carried out, and a voltage coupling over-limit frequency band interval is identified; dynamically scaling a voltage boundary condition of global maximum power point search based on a frequency band center frequency change rate; calculating the mutual information entropy of the direct-current arc noise envelope and the shadow jitter voltage gradient change standard deviation under the voltage boundary condition, and generating a system false lock probability weight through exponential decay mapping; when the probability weight exceeds a threshold value, switching to a global tracking mode, and synchronously outputting an impedance matching instruction to the DC-DC converter; the duty ratio of the switch is adjusted according to an instruction to realize impedance matching, and system false locking caused by dynamic shadow is avoided.

Description

Technical Field

[0001] This invention relates to the field of grid-connected residential photovoltaic power generation technology, and more particularly to a DC module for a grid-connected residential photovoltaic power generation system. Background Technology

[0002] In residential grid-connected photovoltaic (PV) systems, the DC power optimization module serves as the key interface between the PV panels and the grid-connected inverter, undertaking the core function of maximizing energy extraction. Current mainstream solutions employ a distributed architecture, with each PV panel equipped with an independent DC-DC converter (such as a power optimizer), improving the power generation efficiency of a single panel through real-time maximum power point tracking (MPPT). These modules are typically based on H-bridge or Boost topologies, with control strategies primarily using perturbation and observation (P&O) and incremental conductance (INC) methods, achieving high conversion efficiency under uniform illumination conditions.

[0003] However, when the system is deployed in residential environments with dynamic obstructions (such as trees or moving chimney projections), the output characteristics of the photovoltaic array exhibit multi-peak features. Although the existing MPPT control logic of the DC module can locate local peaks, when multiple modules are connected in parallel to the DC bus, the system as a whole is locked at a non-global maximum power point (GMPP) due to the difference in output impedance of each unit and the voltage coupling effect of the bus. This false locking phenomenon causes significant power generation loss, and traditional single-machine optimization algorithms cannot overcome the power optimization limitations caused by voltage coupling. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing a DC module for a household photovoltaic power generation grid-connected system.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] This invention provides the following technical solution:

[0007] A DC module for a household photovoltaic power generation grid-connected system includes:

[0008] The voltage acquisition unit is used to acquire the output voltage of each photovoltaic panel and the DC bus voltage in real time, and generate continuous time series voltage gradient change data.

[0009] The frequency domain analysis unit is used to perform frequency domain energy distribution analysis on voltage gradient change data and identify frequency bands where the voltage coupling strength exceeds the first threshold.

[0010] The boundary scaling unit is used to calculate the dynamic shadow movement speed based on the rate of change of the center frequency of the frequency band interval, and to scale the voltage boundary conditions of the global maximum power point search proportionally by combining the preset correspondence between the shadow movement speed and the voltage boundary scaling amplitude.

[0011] The weight generation unit is used to calculate the mutual information entropy of the DC arc noise envelope and the standard deviation of the voltage gradient change caused by shadow jitter based on the voltage boundary conditions, and to generate the probability weights for inducing system lock-up through exponential decay nonlinear mapping.

[0012] The mode switching unit is used to switch to the global maximum power point tracking mode when the probability weight exceeds the second threshold, and synchronously output impedance matching command to the DC-DC converter.

[0013] The duty cycle adjustment unit is used to adjust the switching duty cycle of the DC-DC converter according to the impedance matching command.

[0014] Furthermore, the output voltage of each photovoltaic panel and the DC bus voltage are collected in real time to generate continuous time series voltage gradient change data, including:

[0015] The output voltage values ​​of each photovoltaic panel and the DC bus voltage value are simultaneously acquired through a high-frequency differential sampling circuit.

[0016] A first-order differential calculation is performed on the photovoltaic panel output voltage value and DC bus voltage value obtained synchronously to obtain the voltage change between adjacent sampling points;

[0017] Arrange the voltage changes in chronological order to generate continuous time series voltage gradient change data.

[0018] Furthermore, frequency domain energy distribution analysis is performed on the voltage gradient change data to identify frequency bands where the voltage coupling strength exceeds a first threshold, including:

[0019] After applying a window function to the voltage gradient change data of the continuous time series, a Fourier transform is performed to obtain the frequency domain energy distribution;

[0020] Calculate the voltage coupling strength value for each frequency band. The voltage coupling strength value is the ratio of the frequency band energy to the total energy.

[0021] When the voltage coupling strength value of a certain frequency band exceeds the first threshold, the corresponding frequency band is determined to be a frequency band interval where the voltage coupling strength exceeds the first threshold.

[0022] Furthermore, based on the rate of change of the center frequency within the frequency band, the dynamic shadow movement speed is calculated. Combined with the preset correspondence between the shadow movement speed and the voltage boundary scaling amplitude, the voltage boundary conditions for the global maximum power point search are scaled proportionally, including:

[0023] The change in the center frequency of a frequency band interval per unit time is taken as the rate of change of the center frequency.

[0024] The dynamic shadow movement speed is calculated based on the rate of change of the center frequency using a linear proportional relationship.

[0025] Query the preset correspondence between shadow movement speed and voltage boundary scaling, and obtain the voltage boundary scaling corresponding to the current dynamic shadow movement speed;

[0026] Multiply the voltage boundary conditions of the global maximum power point search by a scaling factor, which is the sum of one and the voltage boundary scaling magnitude.

[0027] Furthermore, calculating the dynamic shadow movement speed includes multiplying the center frequency change rate by a preset scaling factor to obtain the dynamic shadow movement speed.

[0028] Furthermore, the pre-defined relationship between the shadow movement speed and the voltage boundary scaling magnitude is established in the following way:

[0029] The optimal voltage boundary scaling magnitude corresponding to different shadow movement speeds was measured under standard test conditions.

[0030] The measured data is stored as a velocity-amplitude mapping table.

[0031] Furthermore, based on voltage boundary conditions, the mutual information entropy of the DC arc noise envelope and the standard deviation of the voltage gradient change caused by shadow jitter is calculated. Probability weights for inducing system lock-up are generated through an exponentially decaying nonlinear mapping, including:

[0032] Extract the DC arc noise envelope data under scaled voltage boundary conditions;

[0033] Obtain the standard deviation data of voltage gradient changes caused by shadow jitter within the same time period;

[0034] Construct a joint probability distribution of DC arc noise envelope data and voltage gradient variation standard deviation data;

[0035] Calculate the mutual information entropy value based on the joint probability distribution;

[0036] The mutual information entropy value is input into an exponential decay function for nonlinear mapping, and the output is the probability weight that induces the system to lock falsely.

[0037] Furthermore, the nonlinear mapping of the mutual information entropy value into the exponential decay function includes:

[0038] Call the stored exponential decay function expression;

[0039] Substitute the mutual information entropy value into the exponential decay function to calculate the output value;

[0040] The output values ​​are normalized into probability weights.

[0041] Furthermore, when the probability weight exceeds the second threshold, the system switches to global maximum power point tracking mode and synchronously outputs impedance matching commands to the DC-DC converter, including:

[0042] Compare the probability weight with the second threshold. When the probability weight is greater than the second threshold, start the global maximum power point tracking mode and use a preset search algorithm to scan the global maximum power point.

[0043] Calculate the target impedance value based on the difference between the current output voltage of the photovoltaic panel and the DC bus voltage;

[0044] Generate an impedance matching command that makes the input impedance of the DC-DC converter approach the target impedance value;

[0045] Transmit impedance matching commands to the DC-DC converter.

[0046] Furthermore, adjusting the switching duty cycle of the DC-DC converter according to the impedance matching command includes:

[0047] Analyze the target impedance value in the impedance matching command; calculate the target duty cycle adjustment based on the difference between the target impedance value and the current input impedance value of the DC-DC converter;

[0048] Generate a PWM waveform control signal based on the target duty cycle adjustment;

[0049] The PWM waveform control signal is output to the switching transistor drive circuit of the DC-DC converter to adjust the switching duty cycle.

[0050] The beneficial effects of this invention are:

[0051] 1. A breakthrough in global power locking is achieved through dynamic frequency domain coupling analysis, accurately capturing voltage fluctuation characteristics caused by shadow jitter. The voltage boundary conditions are dynamically adjusted based on the center frequency change rate, so that the search range of the global maximum power point is adapted to the shadow movement speed in real time. This active boundary scaling effectively solves the voltage coupling effect when multiple modules are connected in parallel, ensuring that the global maximum power point can still be accurately located in a dynamic shadow environment, eliminating the risk of false locking of traditional single-machine optimization algorithms under multi-peak characteristics.

[0052] 2. Achieve feedforward control of system lock-up risk. Quantify the correlation strength between DC arc noise and shadow jitter through mutual information entropy. The weight value generated by nonlinear mapping accurately represents the system lock-up probability. Combined with dual threshold criteria to trigger global tracking mode, the synchronously output impedance matching command coordinates the DC-DC converter operating point in real time, so that each module maintains impedance coordination in the global optimization state, significantly improving the overall power generation efficiency in dynamic shading scenarios. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the structure of a DC module for a household photovoltaic power generation grid-connected system according to the present invention;

[0054] Figure 2This is a flowchart illustrating the voltage boundary conditions for the proportionally scaled global maximum power point search of this invention. Detailed Implementation

[0055] 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.

[0056] Example: Figure 1 A schematic diagram of the structure of a DC module for a household photovoltaic power generation grid-connected system is provided. The DC module for a household photovoltaic power generation grid-connected system includes:

[0057] The voltage acquisition unit is used to acquire the output voltage of each photovoltaic panel and the DC bus voltage in real time, and generate continuous time series voltage gradient change data.

[0058] The frequency domain analysis unit is used to perform frequency domain energy distribution analysis on voltage gradient change data and identify frequency bands where the voltage coupling strength exceeds the first threshold.

[0059] The boundary scaling unit is used to calculate the dynamic shadow movement speed based on the rate of change of the center frequency of the frequency band interval, and to scale the voltage boundary conditions of the global maximum power point search proportionally by combining the preset correspondence between the shadow movement speed and the voltage boundary scaling amplitude.

[0060] The weight generation unit is used to calculate the mutual information entropy of the DC arc noise envelope and the standard deviation of the voltage gradient change caused by shadow jitter based on the voltage boundary conditions, and to generate the probability weights for inducing system lock-up through exponential decay nonlinear mapping.

[0061] The mode switching unit is used to switch to the global maximum power point tracking mode when the probability weight exceeds the second threshold, and synchronously output impedance matching command to the DC-DC converter.

[0062] The duty cycle adjustment unit is used to adjust the switching duty cycle of the DC-DC converter according to the impedance matching command.

[0063] Real-time acquisition of output voltage from each photovoltaic panel and DC bus voltage generates continuous time-series voltage gradient change data. The specific implementation is as follows:

[0064] The output voltage values ​​of each photovoltaic panel and the DC bus voltage value are synchronously acquired through a high-frequency differential sampling circuit. The sampling rate of this high-frequency differential sampling circuit is set to no less than 10 kHz. The high-frequency differential sampling circuit includes a multi-channel differential amplifier and a high-speed analog-to-digital converter. The differential amplifier adopts an instrumentation amplifier architecture to eliminate common-mode interference. The positive terminal of each photovoltaic panel output voltage is connected to the non-inverting input terminal of the differential amplifier, and the negative terminal is connected to the inverting input terminal. The DC bus voltage measurement point is connected to the positive input terminal of an independent differential channel, and the negative input terminal is connected to the reference ground potential. The high-speed analog-to-digital converter generates the sampling clock based on a crystal oscillator. For example, when the sampling clock frequency is set to 10 kHz, the sampling interval is 100 microseconds. The sample-and-hold circuits of all input channels are triggered by the same clock edge. For example, the instantaneous voltage values ​​of each channel are latched simultaneously on the rising edge of the clock, and the sampling time deviation between channels is less than 10 nanoseconds. The data packet acquired at each sampling point includes: the photovoltaic panel output voltage value, the DC bus voltage value, and a timestamp accurate to the nanosecond level.

[0065] First-order differential calculations are performed on the synchronously acquired photovoltaic panel output voltage and DC bus voltage values. The calculation process is executed in real-time within the digital signal processor according to the sampling order. Specifically, for any photovoltaic panel, the voltage values ​​of two consecutive timestamp-sequential sampling points are read. The voltage value of the subsequent sampling point is subtracted from the voltage value of the previous sampling point to obtain the voltage change within that time interval. For example, when the sampling interval is 100 microseconds, the voltage change within a 100-microsecond time window is calculated. The differential calculation of the DC bus voltage is performed independently, using the same timestamp alignment rule. The dimension of the voltage change is consistent with the original voltage value, and the unit is volts. The differential calculation process employs a dual-buffering mechanism: the first buffer stores the voltage value of the current sampling period, and the second buffer stores the voltage value of the previous period. The buffer data is updated after the calculation is completed.

[0066] The voltage changes calculated by differential sampling are arranged in timestamp order to generate a continuous time series of voltage gradient change data. The arrangement rule is as follows: the voltage change with the earliest timestamp is placed at the beginning of the sequence, and newly generated data are appended in ascending order of timestamp. This sequence is stored in a circular buffer, the buffer capacity of which is set according to the maximum storage duration. For example, if the data storage is set to 10 seconds, and the sampling rate is 10 kHz, the buffer capacity is 100,000 data points. Each data point contains three elements: voltage change value, device identifier, and timestamp. For multi-PV panel systems, the voltage change of each PV panel forms an independent subsequence, and the timestamps of each subsequence are strictly aligned. The final output voltage gradient change data structure includes: a time dimension array, a voltage change value matrix, and a list of device identifiers.

[0067] The input impedance of the instrumentation amplifier is set to be no less than 10 megohms, for example, using an AD8421 chip, with an input bias current of less than 1 nanoamp to avoid load effects on the photovoltaic circuit. The common-mode rejection ratio of the differential amplifier is set to be no less than 120 dB, for example, through a laser trimming resistor network. The DC bus voltage measurement uses an HCPL-7840 isolation amplifier with a bandwidth set to at least 5 times the sampling frequency; for example, a 100 kHz bandwidth is used when the sampling rate is 10 kHz. The timestamp generation circuit uses a GPS synchronous clock module, such as the UBLOX NEO-M8N module, which provides a time base with 1 microsecond accuracy. Subtraction operations in the first-order differential calculation are performed by the processor's arithmetic logic unit, for example, using hardware subtraction instructions from an ARM Cortex-M7 core. The data overwrite mechanism of the circular buffer is set so that when new data is written beyond the end of the buffer, it automatically returns to the starting address, overwrites historical data, and updates the sequence start pointer.

[0068] Frequency domain energy distribution analysis is performed on voltage gradient change data to identify frequency bands where voltage coupling strength exceeds a first threshold. Specifically, this is implemented as follows:

[0069] A window function, specifically a Hanning window, is applied to the voltage gradient change data of a continuous time series. The window function processing is as follows: a fixed-length data segment is extracted from the voltage gradient change data buffer of the continuous time series, for example, 1024 consecutive data points are extracted to form an analysis data frame; the Hanning window coefficient is multiplied by each data point in the data frame. The Hanning window coefficient is calculated using the formula: Hanning window coefficient = 0.5 × [1 - cos(2π × n / (L - 1)], where n is the index of the data point within the frame (ranging from 0 to L - 1), and L is the length of the data frame. The data frame after applying the window function retains the timestamp information of the original voltage gradient change data, and the sampling time stamp associated with each data point remains unchanged. The windowing operation is completed in the dedicated storage area of ​​the digital signal processor, and the processor maintains the data point order when copying data from the voltage gradient change data circular buffer.

[0070] A Fast Fourier Transform (FFT) is performed on the windowed data frame, with the number of FFT points set to be equal to the data frame length. The FFT calculation process includes: inputting the windowed data frame into the FFT algorithm, which uses radix-2 decimation-time method and generates a complex spectrum result through multi-stage butterfly operations; calculating the energy value for each complex frequency point, where energy value = 2^real part + 2^imaginary part; and arranging the energy values ​​of all frequency points in ascending order of frequency to form a frequency domain energy distribution. The frequency resolution is determined by the sampling rate and the number of FFT points, calculated as resolution = sampling rate / number of points. For example, when the sampling rate is 10 kHz and the number of points is 1024, the frequency resolution is approximately 9.766 Hz. The frequency domain energy distribution data is stored as a structure array, where each element contains a frequency point index, frequency value, energy value, and the corresponding original timestamp start time.

[0071] The process involves dividing frequency bands and calculating the voltage coupling strength for each band. A frequency band is defined as a continuous closed interval on the frequency axis. The frequency band division rule is as follows: the range from 0 Hz to the Nyquist frequency is divided into equal-width sub-intervals. For example, each 10 Hz width constitutes one frequency band, so [0, 10) Hz is the first frequency band, [10, 20) Hz is the second frequency band, and so on. The calculation of the voltage coupling strength value involves three steps: accumulating the energy values ​​of all frequency points within the target frequency band to obtain the frequency band energy; accumulating the energy values ​​of all frequency points across the entire frequency band (0 to the Nyquist frequency) to obtain the total energy; and calculating the ratio of the frequency band energy to the total energy as the voltage coupling strength value. The voltage coupling strength value is a dimensionless value, and the calculation result retains four decimal places of precision. The summation of the frequency band energy and the total energy uses 64-bit floating-point arithmetic to avoid the accumulation of rounding errors.

[0072] A first threshold is set, and frequency band interval determination is performed. The method for determining the first threshold is as follows: Under standard operating conditions without shadow interference, at least 100 sets of voltage gradient change data samples are collected; the voltage coupling strength value of each frequency band in each set of data is calculated; percentile statistics are performed on the voltage coupling strength values ​​of all samples, and the 95th percentile value is taken as the first threshold reference value. For example, if the 95th percentile is 0.25, the first threshold is set to 0.25. The determination process traverses all divided frequency bands and compares the voltage coupling strength value of each frequency band with the first threshold. When the voltage coupling strength value of a frequency band is greater than the first threshold, the start frequency, cutoff frequency, center frequency, and voltage coupling strength value of that frequency band are recorded to form a valid frequency band interval record. All marked valid frequency band intervals are arranged in ascending order of center frequency to generate a frequency band interval list.

[0073] Window function coefficients are stored in a pre-generated table containing 1024 32-bit floating-point values, with the index number corresponding to the data point number n. The Fast Fourier Transform (FFT) is implemented using processor hardware acceleration units, such as using the STM32H7 series chip's digital signal processing instruction set to perform butterfly operations. Frequency band boundaries are defined using a frequency boundary array; for example, the boundary array corresponding to a 10 Hz bandwidth is [0.0, 10.0, 20.0, ... 5000.0] Hz. An exception handling mechanism is implemented for the division operation in the voltage coupling strength calculation: when the total energy value is detected to be less than or equal to 0.001 volts², the data for this frame is discarded and the calculation process is re-initialized. The first threshold is dynamically updated as follows: every 24 hours, 100 frames of voltage gradient change data are re-acquired, and a new 95th percentile value is calculated; if the difference between the new value and the current threshold exceeds 0.05, it is gradually adjusted to the new value in steps of 0.01.

[0074] Figure 2 A flowchart of the voltage boundary conditions for the proportionally scaled global maximum power point search of this invention is provided. The dynamic shadow movement speed is calculated based on the rate of change of the center frequency of the frequency band interval. Combined with the preset correspondence between the shadow movement speed and the voltage boundary scaling amplitude, the voltage boundary conditions for the proportionally scaled global maximum power point search are implemented as follows:

[0075] The center frequency change rate is calculated based on a frequency band interval list, which contains a sequence of valid frequency band interval records arranged chronologically. The calculation process is as follows: Select frequency band interval records within a fixed time window preceding the current time, for example, all valid frequency band interval records within the most recent 1.0 second; for consecutive records with the same frequency band identifier, extract their center frequency values; calculate the change in center frequency between two adjacent time points, specifically the difference between the center frequency values ​​at the later time point and the center frequency value at the previous time point; divide the change by the corresponding time interval to obtain the instantaneous change rate value, where the time interval is calculated using timestamp differences, for example, the difference between the timestamps of the later and earlier records; take the arithmetic mean of all instantaneous change rate values ​​as the center frequency change rate for that frequency band interval. The unit of the center frequency change rate is Hertz per second, and the calculation result is retained to two decimal places. The minimum number of valid records in the time window is set to 3; if there are insufficient valid records, the previous value remains unchanged.

[0076] The dynamic shadow movement speed is calculated based on the rate of change of the center frequency, using a linear proportional relationship. This linear proportional relationship is embodied in a preset proportionality coefficient, which is determined through a standard calibration experiment at a light intensity of 1000 W / m². 2In a standard testing environment, a shadow object with a known moving speed is set up, for example, moving at three speeds: 0.5 m / s, 1.0 m / s, and 2.0 m / s. The rate of change of the center frequency corresponding to each speed is measured. The shadow moving speed is divided by the rate of change of the center frequency to obtain the scaling factor calibration value. The arithmetic mean of the calibration values ​​from multiple experiments is taken as the final preset scaling factor. The calculation is performed during extrapolation: Dynamic shadow moving speed = Rate of change of center frequency × Preset scaling factor. The unit of dynamic shadow moving speed is meters per second, and the calculation result is limited to the range of 0.1 m / s to 5.0 m / s. For example, when the rate of change of center frequency is 2.0 Hz / s and the preset scaling factor is 0.5 m / Hz, the dynamic shadow moving speed is 1.0 m / s.

[0077] This function queries the preset correspondence between shadow movement speed and voltage boundary scaling amplitude, stored in a speed-amplitude mapping table. The speed-amplitude mapping table is a two-dimensional lookup table; the first column is the shadow movement speed value (in meters per second), and the second column is the corresponding voltage boundary scaling amplitude value (dimensionless), with data sorted in ascending order of speed value. The query process uses a binary search algorithm: the current dynamic shadow movement speed is compared with the value in the first column of each row of the mapping table; if an exact match is found, the corresponding voltage boundary scaling amplitude value is directly returned; if no exact match is found, the voltage boundary scaling amplitude value is calculated by linear interpolation of two adjacent speed values ​​using the following formula: Amplitude value = Lower amplitude limit + (Upper amplitude limit - Lower amplitude limit) × (Current speed - Lower speed limit) / (Upper speed limit - Lower speed limit). The voltage boundary scaling amplitude value ranges from 0 to 0.5. For example, if the mapping table contains data pairs such as [0.5,0.1], [1.0,0.2], and [2.0,0.3], and the amplitude value is calculated to be 0.25 when the speed is 1.5 meters per second, through interpolation.

[0078] The voltage boundary conditions for the global maximum power point search are scaled. These conditions include two parameters: a minimum voltage boundary value and a maximum voltage boundary value. The scaling process is as follows: The scaling factor is obtained from the query; a scaling coefficient is calculated: scaling coefficient = 1 + voltage boundary scaling factor; the original minimum voltage boundary value is multiplied by the scaling coefficient to obtain the scaled minimum voltage boundary value; the original maximum voltage boundary value is multiplied by the scaling coefficient to obtain the scaled maximum voltage boundary value. The scaled voltage boundary values ​​are retained to two decimal places, and the unit is volts, consistent with the original values. For example, if the original voltage boundary conditions are a minimum of 200 volts and a maximum of 400 volts, and the voltage boundary scaling factor is 0.2, then the scaled voltage boundary conditions will be a minimum of 240 volts and a maximum of 480 volts. The scaled voltage boundary conditions are updated to the parameter register of the maximum power point tracking algorithm in real time, and the update process uses atomic operations to ensure data consistency.

[0079] The process of establishing the velocity-amplitude mapping table was completed through standard test experiments: under an illumination intensity of 1000 W / m². 2 In a standard test environment, a shadowed object is set to move at a constant speed, such as 0.1 m / s, 0.5 m / s, 1.0 m / s, 1.5 m / s, and 2.0 m / s. At each speed, the voltage boundary scaling value is adjusted from 0 to 0.5 in steps of 0.05. The output power of the photovoltaic system is measured in real time. The optimal voltage boundary scaling value that maximizes the output power is recorded. The test speed values ​​and the optimal scaling value are paired and stored in a mapping table. The mapping table update mechanism is as follows: a new test is performed monthly, which includes three new speed points. When the deviation between the new data and the mapping table interpolation result exceeds 10%, a full retest is triggered to rebuild the mapping table.

[0080] The center frequency change rate calculation employs a sliding window management mechanism with a window length of 1.0 second. Records within the window are stored in order of timestamp. The calibration experiment requirements for the preset proportional coefficient are: each velocity point test duration is no less than 30 seconds, with a sampling interval of 100 milliseconds; outlier data points outside the ±3 standard deviation range are removed; the final coefficient is the average of 20 valid experiments. The binary search algorithm for the mapping table lookup is set to a maximum iteration depth of 10 times, stopping iteration when the difference between adjacent velocities is less than 0.01 meters per second. Overflow protection is implemented for voltage boundary scaling calculation: when the scaled minimum voltage boundary value is lower than the hardware allowable lower limit (e.g., 50 volts), it is automatically corrected to the hardware lower limit; when the scaled maximum voltage boundary value exceeds the hardware allowable upper limit (e.g., 600 volts), it is automatically corrected to the hardware upper limit. Invalid input handling mechanism: when the center frequency change rate is ≤0, the dynamic shadow movement speed is forcibly set to 0.1 meters per second; when the query speed exceeds the maximum value of the mapping table, the amplitude value corresponding to the maximum speed is used.

[0081] The mutual information entropy of the DC arc noise envelope and the standard deviation of the voltage gradient change caused by shadow jitter is calculated based on the voltage boundary conditions. Probability weights for inducing system lock-up are generated through an exponentially decaying nonlinear mapping. Specifically, the implementation is as follows:

[0082] The DC arc noise envelope data under scaled voltage boundary conditions is extracted through a three-stage signal processing process: First, raw DC bus voltage waveform data is acquired at a sampling rate of no less than 10 kHz within the voltage operating range determined by the scaled minimum and maximum voltage boundary values. Second, the raw data is input into a bandpass filter with a passband range of 500 Hz to 10000 Hz. This filter employs fourth-order attenuation characteristics, suppressing signals outside the passband by at least 40 dB. Then, the filtered AC component undergoes absolute value conversion, converting all negative values ​​to positive values. Finally, the converted signal is input into a low-pass filter with a cutoff frequency of 100 Hz for smoothing, outputting a DC arc noise envelope data sequence. The envelope data is stored in a time-series structure, with each data point containing a timestamp with microsecond-level precision and an amplitude in volts. For example, when the voltage boundary conditions are set to a minimum of 240 volts and a maximum of 480 volts, continuous acquisition for 100 milliseconds yields 1000 envelope data points arranged in chronological order.

[0083] This method acquires standard deviation data of voltage gradient changes caused by shadow jitter, which is strictly time-aligned with the DC arc noise envelope data. The acquisition process includes precise matching and dynamic calculation: Voltage gradient change data sequences that perfectly correspond to the timestamps of the envelope data are retrieved from the circular buffer of the voltage gradient change data, with time synchronization deviation controlled within ±100 microseconds; the standard deviation is calculated using a sliding time window method, with a fixed window length of 200 milliseconds and approximately 2000 data points within the window; the standard deviation calculation process is as follows: first, the arithmetic mean of all voltage gradient changes within the window is calculated; then, the squared deviation of each change from the mean is calculated; then, the sum of all squared deviations is calculated; this sum is divided by the number of data points minus one; finally, the square root of the quotient is taken to obtain the standard deviation; the window slides in 10-millisecond steps, and the standard deviation is recalculated after each slide; a continuous standard deviation data sequence is output. For example, within the interval from start time t1 to end time t2, 50 standard deviation data points can be obtained through sliding window calculation.

[0084] A joint probability distribution of DC arc noise envelope data and voltage gradient variation standard deviation data is constructed. The construction process consists of data pairing and statistical distribution: Envelope data points and standard deviation data points aligned with timestamps are paired into two-dimensional data pairs; the envelope amplitude range is divided into M equal intervals (e.g., 20 intervals), with interval width = (maximum envelope value - minimum envelope value) / 20; the standard deviation range is divided into N equal intervals (e.g., 20 intervals), with interval width = (maximum standard deviation - minimum standard deviation) / 20; this forms an M x N two-dimensional grid; the number of data pairs falling into each grid cell is counted; and the joint probability value P(i,j) is obtained by dividing the count value of each cell by the total number of data pairs. For example, when there are 1000 total data pairs and the count value in a grid cell (i = 5, j = 8) is 42, the joint probability P(5,8) = 0.042.

[0085] The mutual information entropy value is calculated based on the joint probability distribution. The calculation process is performed in three steps: First, the marginal probability distribution of the envelope data is calculated. For each envelope amplitude interval, the joint probability value of that interval combined with all standard deviation intervals is accumulated. Second, the marginal probability distribution of the standard deviation data is calculated. For each standard deviation interval, the joint probability value of that interval combined with all envelope amplitude intervals is accumulated. Third, the mutual information entropy value is calculated. For each two-dimensional cell, the quotient of the joint probability value divided by the corresponding marginal probability product is calculated. Then, the natural logarithm of the quotient is calculated. The joint probability value is then multiplied by the logarithm. Finally, the products of all cells are summed and the result is negative. When the joint probability is zero, the contribution value of that cell is set to zero. When the product of marginal probabilities is less than one in a million, the value is set to one in a million to avoid division by zero errors. The calculation result is retained to four decimal places. For example, when dividing into a 20x20 grid, the contribution values ​​of 400 cells need to be calculated and accumulated.

[0086] The intermediate value is calculated using a predefined exponential decay function, which is a negative exponential form of the natural exponential function. The exponent is the decay coefficient multiplied by the mutual information entropy value. The decay coefficient is determined through system calibration experiments: in a test environment simulating photovoltaic system lock-up, the mutual information entropy value and the actual lock-up frequency are measured under different operating conditions; the decay coefficient is adjusted to maximize the statistical correlation between the function output value and the lock-up frequency; the final decay coefficient is determined to be 8.0. The function expression is stored in a parsable string format. During calculation, the mutual information entropy value is multiplied by the decay coefficient, the negative value is taken, and then the intermediate value of the natural exponential function output is calculated.

[0087] The median value is converted into a probability weight using a linear normalization method: Dynamically adjustable lower and upper thresholds are set. When the median value is less than the lower threshold, the probability weight is set to zero; when the median value is greater than the upper threshold, the probability weight is set to 1.0; when the median value is between the two thresholds, the probability weight is equal to the median value minus the lower threshold, divided by the upper threshold minus the lower threshold. The lower and upper thresholds are determined through historical data analysis: during 100 hours of continuous system operation, all median values ​​under normal operating conditions are recorded; the minimum median value is taken as the lower threshold; the maximum median value is taken as the upper threshold. The probability weight is a dimensionless value, ranging from 0 to 1. For example, when the lower threshold is 0.001, the upper threshold is 0.999, and the median value is 0.85, the probability weight is (0.85-0.001) / (0.999-0.001)≈0.851.

[0088] The bandpass filter in the signal processing stage is implemented using two cascaded second-order filter units, each providing 12 dB of attenuation per octave. The sliding window standard deviation calculation employs an incremental update strategy: maintaining three state variables—the cumulative sum of data within the window, the cumulative sum of squares, and the number of data points; updating these state variables and recalculating the standard deviation when a new data point is added. A minimum count constraint is set for probability distribution statistics: when the number of data pairs in a unit is less than 5, it is merged into an adjacent unit. The natural logarithm in the mutual information entropy calculation uses a multinomial approximation algorithm, approximating the calculation with a six-term expansion. The exponential function calculation uses pre-calculated values ​​combined with linear interpolation, storing sampled values ​​in the range of 0 to 10 at intervals of 0.01. The normalization threshold update mechanism is as follows: every 24 hours, the most recent 1000 sets of intermediate values ​​are re-analyzed, updating the lower and upper thresholds, with the adjustment range limited to ±0.05.

[0089] When the probability weight exceeds the second threshold, the system switches to global maximum power point tracking mode and synchronously outputs an impedance matching command to the DC-DC converter. Specifically, the implementation is as follows:

[0090] The system compares the probability weight with the second threshold in real time. The probability weight is a dimensionless value calculated in the previous steps, ranging from 0 to 1. The second threshold is determined through system calibration experiments: in a test environment simulating various shadow interference conditions, the correspondence between changes in the probability weight and the actual occurrence of false locks is recorded, and the critical weight value at which the false lock rate significantly increases is used as the benchmark value for the second threshold. The comparison process is as follows: read the current probability weight value; retrieve the pre-stored second threshold value from non-volatile memory; perform a numerical comparison operation, and output a logical truth signal when the probability weight is greater than the second threshold. For example, if the second threshold is set to 0.7, and the current probability weight is 0.85, then the judgment condition is met. The comparison operation is performed at a fixed period, for example, once every 100 milliseconds, and the result is updated to the system status register in real time.

[0091] When the probability weight exceeds the second threshold, the system switches to global maximum power point tracking (MPPT) mode. This mode uses a pre-defined perturbation-observation algorithm to scan the global maximum power point. The mode switching process includes state transitions and parameter initialization: First, the system operating mode flag is updated from local maximum power point tracking (MPPT) mode to global maximum power point tracking (MPPT) mode; second, the scanning parameters are initialized, setting the starting scanning voltage to the minimum voltage boundary value of the current voltage boundary condition, the ending scanning voltage to the maximum voltage boundary value, and the scanning step size to 1% of the voltage boundary range width; finally, the perturbation-observation algorithm is started to perform the scan. The execution flow of the perturbation-observation algorithm is as follows: a voltage perturbation is applied to the photovoltaic panel output voltage. The perturbation direction is determined by the current power change trend. For example, when a power increase is detected, the original perturbation direction is maintained, and when the power decreases, the perturbation direction is reversed; the perturbation amplitude is dynamically adjusted according to the light intensity, for example, set to 1% of the voltage value under standard illumination; the scan continues until the entire voltage boundary range is traversed. During the scan, the maximum power value and its corresponding voltage value are recorded in real time.

[0092] The target impedance value is calculated based on the difference between the instantaneous output voltage of the photovoltaic panel and the instantaneous DC bus voltage. The calculation process includes data acquisition and numerical computation: the photovoltaic panel output voltage is acquired in real time using a voltage sensor; the DC bus voltage is acquired in real time using another voltage sensor; the absolute value of the algebraic difference between the two voltage values ​​is calculated; and this voltage difference is divided by a preset reference current value to obtain the target impedance value. The reference current value is set according to the rated parameters of the photovoltaic system, typically taking 90% of the maximum power point current of the photovoltaic panel. The unit of the target impedance value is ohms, and the calculation result is retained to two decimal places. For example, when the measured output voltage of the photovoltaic panel is 300 volts, the measured DC bus voltage is 280 volts, and the reference current is 8 amperes, the target impedance value is equal to 20 volts divided by 8 amperes, which equals 2.5 ohms.

[0093] An impedance matching command is generated to bring the input impedance of the DC-DC converter closer to the target impedance value. The command generation process is based on the closed-loop control principle: the actual input impedance value at the input terminal of the DC-DC converter is measured in real time by dividing the instantaneous input voltage value by the instantaneous input current value; the algebraic difference between the actual input impedance value and the target impedance value is calculated; this difference is input to the proportional-integral (PI) controller, whose proportional coefficient is determined to be 0.5 through system step response testing, and whose integral time constant is determined to be 0.1 seconds through system stability testing; the PI controller outputs an impedance adjustment amount; and the adjustment amount is converted into a standardized impedance matching command data packet. For example, when the target impedance value is 2.5 ohms and the actual measured impedance value is 3.0 ohms, the PI controller outputs a negative adjustment amount, generating a command to reduce the input impedance.

[0094] Impedance matching commands are transmitted to the control unit of the DC-DC converter via an industry-standard serial communication protocol. Command data packets are encapsulated in a general frame format, including a start symbol, device address, command type, data field, checksum, and end symbol. They are sent at a rate of 19200 bits per second via an asynchronous serial interface. The DC-DC converter has a receive buffer and executes the command after verifying data integrity using the checksum. Transmission assurance mechanisms include: each command is sent three times consecutively; the receiving end returns an acknowledgment within 10 milliseconds; if no acknowledgment is received within the timeout period, a retransmission mechanism is triggered. Command transmission priority is set to the highest level to ensure a transmission delay of no more than 1 millisecond.

[0095] The dynamic calibration mechanism for the second threshold is as follows: Monthly statistics are compiled on the actual number of false lock-up events when the probability weight exceeds the current threshold. If the false lock-up rate exceeds 5%, the threshold is lowered by 0.05; if the false lock-up rate is below 1%, the threshold is raised by 0.03. The update strategy for the reference current value is as follows: a full-range current-voltage characteristic scan is performed every 24 hours, updating the maximum power point current value and recalculating the reference current to 90% of the latest maximum power point current value. The anti-integral saturation mechanism of the proportional-integral controller is set to pause integral term accumulation when the absolute value of the output regulation exceeds 20% of the rated range. The communication anomaly handling procedure is as follows: after three consecutive transmission failures, the system automatically switches to the pulse width modulation signal direct control mode, sending analog control signals via a dedicated signal line.

[0096] Adjusting the switching duty cycle of the DC-DC converter according to the impedance matching command is specifically implemented as follows:

[0097] The process of parsing the target impedance value from the received impedance matching instruction data packet is executed in the microcontroller of the DC-DC converter. The impedance matching instruction data packet consists of four parts: a start identifier field, an instruction type code field, a target impedance value data field, and a cyclic redundancy check (CRC) field. The parsing process is as follows: First, it checks whether the start identifier field conforms to a predefined format, for example, checking if the first two bytes are the hexadecimal value 0xAA55; second, it extracts the four bytes of binary data from the target impedance value data field; it converts this binary data into the target impedance value according to the IEEE 754 single-precision floating-point format; finally, it calculates the CRC value and compares it with the check field in the data packet. If the check fails, a retransmission request is sent through the communication interface. The target impedance value is fixed in ohms, and the parsing result is stored in the impedance instruction register. For example, when the data field byte content is 40200000, the converted target impedance value is 2.5 ohms. A timeout monitoring mechanism is set for the parsing operation: if the parsing time exceeds 500 microseconds, the current instruction is abandoned and the previous valid instruction value is restored.

[0098] The target duty cycle adjustment is calculated based on the difference between the target impedance and the current input impedance of the DC-DC converter. This calculation is achieved through a proportional-integral (PI) control algorithm. The current input impedance is obtained through real-time measurement: a voltage sensor is used to collect the instantaneous voltage value at the DC-DC converter input port; a current sensor is used to collect the instantaneous current value at the input port; the instantaneous input port voltage value is divided by the instantaneous input port current value to obtain the current input impedance value. The calculation steps for the target duty cycle adjustment are as follows: calculate the numerical difference between the target impedance and the current input impedance; input this numerical difference into the PI controller; the PI controller performs proportional and integral operations and then outputs the target duty cycle adjustment. The proportional gain of the PI controller is determined to be 0.2 through open-loop response testing, and the integral time constant is determined to be 50 milliseconds through closed-loop stability testing. The target duty cycle adjustment is a dimensionless value, ranging from -0.1 to +0.1. For example, when the target impedance is 2.5 ohms and the current input impedance is 3.0 ohms, the PI controller may output an adjustment of -0.05.

[0099] The process of generating a PWM waveform control signal based on the target duty cycle adjustment includes two stages: duty cycle update and waveform generation. The reference duty cycle value is stored in non-volatile memory, with an initial default value of 50%. The duty cycle update process is as follows: read the current reference duty cycle value from memory; add the target duty cycle adjustment to the reference duty cycle value to obtain the new duty cycle value; perform limiting processing on the new duty cycle value, limiting it to between 10% and 90%; and write the limited duty cycle value back to memory. In the waveform generation stage: a carrier signal is generated through a clock divider, with the carrier frequency set to 20 kHz; the duty cycle value is converted into a timer comparison value; for example, when the timer period count is 1000, a 40% duty cycle corresponds to a comparison value of 400; and the PWM waveform signal is generated through the timer's comparison matching function. For example, when the new duty cycle is 45%, a square wave signal with a high level occupying 45% of the cycle time is generated.

[0100] The PWM waveform control signal is output to the switching transistor drive circuit of the DC-DC converter. The output process includes level adaptation and timing control. The drive circuit requires an input signal of 3.3 volts logic level, therefore level conversion is performed: the 1.8 volt logic level output by the controller is boosted to 3.3 volts by a level conversion chip. Timing synchronization control: the rising edge of the PWM signal is precisely aligned with the falling edge of the carrier signal, with an alignment deviation not exceeding 10 nanoseconds. The output protection mechanism includes: when the current detection circuit reports an overcurrent event, the PWM signal is immediately forced to a low level; a dead time is inserted during PWM signal state switching, with a dead time length set to 100 nanoseconds; the dead time setting is based on 1.5 times the turn-off delay characteristic of the switching transistor. Physical signal transmission uses twisted-pair shielded cable, with a maximum transmission distance limited to within 0.5 meters.

[0101] The parameter calibration method for the proportional-integral controller is as follows: Apply an impedance step disturbance under standard test load conditions and record the system response curve; adjust the proportional coefficient to ensure the response overshoot does not exceed 10%; adjust the integral time constant to ensure the system settling time is less than 100 milliseconds. The duty cycle limiting threshold is set based on: determining the safe operating area through power device thermal stress testing; when the duty cycle is below 10%, the inductor current may be intermittent, and when it is above 90%, switching losses increase significantly. The drive circuit protection strategy includes: real-time monitoring of the drive chip output current; triggering current limiting protection when it exceeds 2 amps; monitoring the drive chip junction temperature; automatically reducing the PWM frequency when it exceeds 125 degrees Celsius. Instruction parsing error handling: After three consecutive parsing failures, switch to safe mode with a fixed duty cycle of 50%.

[0102] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0103] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0104] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0105] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0106] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0107] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0108] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0109] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion 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 described in the various embodiments of this application. 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.

[0110] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0111] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A DC module for a household photovoltaic power generation grid-connected system, characterized in that, include: The voltage acquisition unit is used to acquire the output voltage of each photovoltaic panel and the DC bus voltage in real time, and generate continuous time series voltage gradient change data. The frequency domain analysis unit is used to perform frequency domain energy distribution analysis on voltage gradient change data and identify frequency bands where the voltage coupling strength exceeds the first threshold. The boundary scaling unit is used to calculate the dynamic shadow movement speed based on the rate of change of the center frequency of the frequency band interval, and to scale the voltage boundary conditions of the global maximum power point search proportionally by combining the preset correspondence between the shadow movement speed and the voltage boundary scaling amplitude. The weight generation unit is used to calculate the mutual information entropy of the DC arc noise envelope and the standard deviation of the voltage gradient change caused by shadow jitter based on the voltage boundary conditions, and to generate the probability weights for inducing system lock-up through exponential decay nonlinear mapping. The mode switching unit is used to switch to the global maximum power point tracking mode when the probability weight exceeds the second threshold, and synchronously output impedance matching command to the DC-DC converter. The duty cycle adjustment unit is used to adjust the switching duty cycle of the DC-DC converter according to the impedance matching command.

2. The DC module of a household photovoltaic power generation grid-connected system according to claim 1, characterized in that, Real-time acquisition of output voltage from each photovoltaic panel and DC bus voltage generates continuous time-series voltage gradient change data, including: The output voltage values ​​of each photovoltaic panel and the DC bus voltage value are simultaneously acquired through a high-frequency differential sampling circuit. A first-order differential calculation is performed on the photovoltaic panel output voltage value and DC bus voltage value obtained synchronously to obtain the voltage change between adjacent sampling points; Arrange the voltage changes in chronological order to generate continuous time series voltage gradient change data.

3. A DC module for a household photovoltaic power generation grid-connected system according to claim 2, characterized in that, Frequency domain energy distribution analysis was performed on voltage gradient change data to identify frequency bands where voltage coupling strength exceeds a first threshold, including: After applying a window function to the voltage gradient change data of the continuous time series, a Fourier transform is performed to obtain the frequency domain energy distribution; Calculate the voltage coupling strength value for each frequency band. The voltage coupling strength value is the ratio of the frequency band energy to the total energy. When the voltage coupling strength value of a certain frequency band exceeds the first threshold, the corresponding frequency band is determined to be a frequency band interval where the voltage coupling strength exceeds the first threshold.

4. A DC module for a household photovoltaic power generation grid-connected system according to claim 3, characterized in that, The dynamic shadow movement speed is calculated based on the rate of change of the center frequency within the frequency band. Combined with the preset correspondence between the shadow movement speed and the voltage boundary scaling amplitude, the voltage boundary conditions for the global maximum power point search are scaled proportionally, including: The change in the center frequency of a frequency band interval per unit time is taken as the rate of change of the center frequency. The dynamic shadow movement speed is calculated based on the rate of change of the center frequency using a linear proportional relationship. Query the preset correspondence between shadow movement speed and voltage boundary scaling, and obtain the voltage boundary scaling corresponding to the current dynamic shadow movement speed; Multiply the voltage boundary conditions of the global maximum power point search by a scaling factor, which is the sum of one and the voltage boundary scaling magnitude.

5. A DC module for a household photovoltaic power generation grid-connected system according to claim 4, characterized in that, The calculation of dynamic shadow movement speed includes multiplying the rate of change of the center frequency by a preset scaling factor to obtain the dynamic shadow movement speed.

6. A DC module for a household photovoltaic power generation grid-connected system according to claim 4, characterized in that, The preset correspondence between shadow movement speed and voltage boundary scaling is established in the following way: The optimal voltage boundary scaling magnitude corresponding to different shadow movement speeds was measured under standard test conditions. The measured data is stored as a velocity-amplitude mapping table.

7. A DC module for a household photovoltaic power generation grid-connected system according to claim 4, characterized in that, The mutual information entropy of the DC arc noise envelope and the standard deviation of the voltage gradient change caused by shadow jitter is calculated based on voltage boundary conditions. Probability weights for inducing system lock-up are generated through an exponentially decaying nonlinear mapping, including: Extract the DC arc noise envelope data under scaled voltage boundary conditions; Obtain the standard deviation data of voltage gradient changes caused by shadow jitter within the same time period; Construct a joint probability distribution of DC arc noise envelope data and voltage gradient variation standard deviation data; Calculate the mutual information entropy value based on the joint probability distribution; The mutual information entropy value is input into an exponential decay function for nonlinear mapping, and the output is the probability weight that induces the system to lock falsely.

8. A DC module for a household photovoltaic power generation grid-connected system according to claim 7, characterized in that, The nonlinear mapping of mutual information entropy values ​​into an exponential decay function includes: Call the stored exponential decay function expression; Substitute the mutual information entropy value into the exponential decay function to calculate the output value; The output values ​​are normalized into probability weights.

9. A DC module for a household photovoltaic power generation grid-connected system according to claim 7, characterized in that, When the probability weight exceeds the second threshold, switch to global maximum power point tracking mode and synchronously output impedance matching commands to the DC-DC converter, including: Compare the probability weight with the second threshold. When the probability weight is greater than the second threshold, start the global maximum power point tracking mode and use a preset search algorithm to scan the global maximum power point. Calculate the target impedance value based on the difference between the current output voltage of the photovoltaic panel and the DC bus voltage; Generate an impedance matching command that makes the input impedance of the DC-DC converter approach the target impedance value; Transmit impedance matching commands to the DC-DC converter.

10. A DC module for a household photovoltaic power generation grid-connected system according to claim 9, characterized in that, Adjusting the switching duty cycle of the DC-DC converter according to impedance matching instructions includes: Analyze the target impedance value in the impedance matching command; calculate the target duty cycle adjustment based on the difference between the target impedance value and the current input impedance value of the DC-DC converter; Generate a PWM waveform control signal based on the target duty cycle adjustment; The PWM waveform control signal is output to the switching transistor drive circuit of the DC-DC converter to adjust the switching duty cycle.

Images