A multi-model photovoltaic assembly array control method

Abstract

This invention discloses a control method for multi-model photovoltaic module arrays, belonging to the field of intelligent photovoltaic technology. The method includes: S1, acquiring current electrical parameters; S2, electrical parameter difference and threshold comparison analysis: calculating the difference between the current electrical parameters and the predicted electrical parameters, and determining whether the difference is greater than a preset threshold. If it is greater, proceed to step S3; otherwise, proceed to step S4; S3, adjusting the power supply control quantity and time series array values, then proceeding to step S5; S4, rearranging the time series array; S5, predicting the next time series electrical parameters; S6, two-dimensional FFT: performing a two-dimensional Fourier transform on the time series matrix to obtain the time-spectrum array; S7, predicting the electrical parameter increment; S8, MPPT algorithm control. By integrating frequency domain analysis, probabilistic prediction, and distributed parallel control, environmental abrupt changes and module differences are transformed into computable time-series spectral features, achieving "predictive" tracking of MPPT, which can significantly improve power generation efficiency under complex illumination and module difference scenarios.

Description

Technical Field

[0001] This invention relates to a control method for multi-model photovoltaic module arrays, belonging to the field of intelligent photovoltaic technology. Background Technology

[0002] In existing technologies, distributed installation of photovoltaic module arrays is quite common. Generally, photovoltaic modules of uniform specifications are used to form an array for a planned area; however, in practice, because a surplus of photovoltaic modules is reserved for material preparation during each project planning and installation, most photovoltaic projects end up with some surplus photovoltaic modules after installation. Over time, companies end up storing a large number of surplus photovoltaic modules in their warehouses, and these photovoltaic modules are of different models and specifications.

[0003] One obvious way to reuse these photovoltaic (PV) modules is to install mixed arrays of different models. However, due to the different models, this mixed installation easily leads to backflow of current. Generally, adding a smart optimizer to each PV module can solve this problem. However, for projects that reuse PV modules, cost is a crucial factor. If the cost of adding a smart optimizer is too high, the entire PV module reuse project becomes meaningless. Therefore, for such PV module reuse projects, it is necessary to minimize costs while avoiding backflow caused by mixed array installations of different models.

[0004] To this end, Chinese patent document with publication number CN121150607A discloses a control system for a multi-model photovoltaic module array. The system adjusts the PWM parameters by comparing the measured voltage with the voltage threshold, and the intelligent calculation module calculates the updated voltage threshold value to actively suppress current backflow and avoid energy backflow caused by mismatch of module parameters, as well as dynamically adjust the voltage threshold.

[0005] However, in practical applications, it has been found that when local shading occurs in the photovoltaic array, components age, or different types of components are mixed, the output power-voltage PV curve exhibits multiple peaks, i.e., a multi-peak phenomenon. Furthermore, when the intelligent computing module uses the traditional MPPT algorithm, it is prone to getting stuck in local power points and cannot track the global maximum power point, resulting in power generation losses. Simultaneously, when the light intensity changes rapidly, the traditional MPPT algorithm needs to re-search for the power point, and the fixed step size or complex parameter tuning leads to tracking delays, such as the computationally intensive variable step size perturbation method. In addition, under a centralized control architecture, there are also issues such as data processing delays from multiple components causing control command lag, and mismatch losses caused by differences in component parameters (such as temperature coefficient and aging degree). Summary of the Invention

[0006] To address the aforementioned technical issues, this invention provides a control method for multi-model photovoltaic module arrays. This method integrates frequency domain analysis, probabilistic prediction, and distributed parallel control to transform environmental abrupt changes and module differences into calculable time-series spectral characteristics, enabling "predictive" tracking of MPPT (Multi-Level Photovoltaic Spectrum Transformation). This significantly improves power generation efficiency under complex lighting and module difference scenarios.

[0007] This invention is achieved through the following technical solution: A method for controlling a multi-model photovoltaic module array includes the following steps: S1. Obtain current electrical parameters: Obtain the current electrical parameters from the photovoltaic string output circuit; S2. Comparison and analysis of electrical parameter difference and threshold: Calculate the difference between the current electrical parameter and the predicted electrical parameter, and determine whether the difference is greater than the preset threshold. If it is greater, proceed to step S3; otherwise, proceed to step S4. S3. Adjustment of power supply control quantity and timing array values: Adjust the power supply control quantity of the current photovoltaic string according to the ratio of the difference to the current electrical parameters, and adjust the values ​​in the timing array by a preset multiple of the ratio, and then proceed to step S5. S4. Reorder the timing array: Place the current electrical parameter at the beginning of the timing array and clear the last element of the timing array; S5. Predict the electrical parameters of the next time series: The time series arrays of multiple photovoltaic strings are calculated using the full probability weighting function to obtain the predicted electrical parameters of the next time series. S6. Two-dimensional FFT: Arrange the time series arrays of multiple photovoltaic strings into a time series matrix in order, and then perform a two-dimensional Fourier transform on the time series matrix to obtain the time spectrum matrix; S7. Predict electrical parameter increments: Calculate the electrical parameter increments for the next time series based on the time-frequency array; S8, MPPT algorithm control: The MPPT conductance incremental algorithm is used to control the photovoltaic string, and the electrical parameter increment of the next time sequence is used as the electrical parameter increment input in the MPPT conductance incremental algorithm.

[0008] The current electrical parameters and predicted electrical parameters both include current and voltage.

[0009] In step S3, the preset multiple of the ratio value is in the range of 0.5 to 1.0.

[0010] The preset ratio in step S3 is 0.63.

[0011] The full probability weighting function in step S5 refers to multiplying each numerical item in the time series array with a corresponding probability numerical item in the probability array, and then summing all the multiplication results.

[0012] The probability array is set for each photovoltaic string, and the number of items in the probability array is the same as the number of items in the time series array.

[0013] In step S6, the time series array of each photovoltaic string is arranged as a row in the time series matrix.

[0014] The number of items p in the time series array is not less than the number of photovoltaic strings n.

[0015] In step S7, the time spectrum array is first truncated according to the preset selected range, and the part of the time spectrum array that is not in the selected range is set to empty.

[0016] Steps S1 to S4 are performed on each photovoltaic string separately; The calculation of electrical parameters involved in steps S1 to S7 is performed separately for different types of electrical parameters.

[0017] The beneficial effects of this invention are as follows: 1. This invention integrates frequency domain analysis, probabilistic prediction, and distributed parallel control to transform environmental abrupt changes and component differences into computable time-series spectral characteristics, thereby achieving "predictive" tracking of MPPT and significantly improving power generation efficiency under complex lighting and component difference scenarios.

[0018] 2. This invention analyzes the frequency domain characteristics of each component group through the two-dimensional Fourier transform of the time-series matrix, and selects the dominant frequency components by combining the time-spectrum array to dynamically predict the electrical parameter increment of the next time series. Based on the full probability weight function, it integrates historical data and probability distribution to quantify the impact differences of environmental changes on photovoltaic modules and improve the prediction accuracy of the global optimal point. The predicted electrical parameter increment is directly input into the MPPT conductance increment algorithm, replacing the traditional increment calculation that relies on the current instantaneous value, thereby adjusting the operating point in advance, reducing the oscillation period, and adjusting the power supply control quantity by the ratio of the difference to the current electrical parameter to optimize the weight of the time-series array with a preset multiple, accelerating the system convergence to the new maximum power point. The time-series array is arranged into a matrix for 2D-FFT, and the information volume is compressed by frequency domain analysis to reduce computational complexity. An independent probability array is configured for each component group, which can effectively adapt to the output attenuation curves and other characteristics of different component models and improve the personalized prediction accuracy. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] The technical solution of the present invention is further described below, but the scope of protection is not limited to what is described.

[0021] First implementation method: like Figure 1As shown, the multi-model photovoltaic module array control method of the present invention includes the following steps: S1. Obtain current electrical parameters: In the t-th time period, obtain the current electrical parameters from the output circuit of the n-th photovoltaic string, that is, the current of the n-th photovoltaic string in the t-th time period. and the voltage of the nth photovoltaic string at time t .

[0022] S2. Comparison and analysis of electrical parameter differences and thresholds: Calculate the current electrical parameters { , } and predicted electrical parameters { , The difference between} , }, and determine the difference { , Is it greater than the preset threshold? , If the value is greater than}, proceed to step S3; otherwise, proceed to step S4. S3. Adjustment of power supply control quantity and timing array value: Based on the ratio of the difference to the current electrical parameter s={ , Adjust the current power supply control quantity of the photovoltaic string, and use a preset multiple us of the proportional value s, i.e., { , Adjust timing array O n-1 ={[ , , ,……, ],[ , , ,……, The value in ]} is then used to proceed to step S5.

[0023] S4. Reorder the timing array: rearrange the current electrical parameters { , If the first element is inserted into the timing array and the last element is cleared, then the timing array will have an O(n) value. n ={[ , , ,……, ],[ , , ,……, ]}.

[0024] S5. Predicting the next time-series electrical parameters: using the full probability weighting function K(O) n ,P nThe time-series array O for multiple photovoltaic strings n Calculations are performed to obtain the predicted electrical parameters for the next time series {[ , , ,……, ],[ , , ,……, ]}.

[0025] S6. Two-dimensional FFT: Arranges the time-series arrays of multiple photovoltaic strings into a time-series matrix in order, N={{[ , , ,……, ], [ , , ,……, ],……,[ , , ,……, ]},{[ , , ,……, ], [ , , ,……, ],……, [ , , ,……, Then, a two-dimensional Fourier transform is performed on the time series matrix N to obtain the time-spectrum matrix F = {{[ , , ,……, ], [ , , ,……, ],……, [ , , ,……, ]},{[ , , ,……, ],[ , , ,……, ],……, [ , , ,……, ]}}.

[0026] S7. Predicting electrical parameter increments: Calculate the electrical parameter increment d for the next time series based on the time-frequency array F. t+1 ={[ , , ,……, ],[ , , ,……, In practice, the size of the time spectrum array u=n can be controlled to ensure that each row in the time spectrum array corresponds to each row in the time series matrix. Then, by performing an inverse FFT transformation on each row and taking the calculated value of the next time series, the electrical parameter increment of the corresponding row can be calculated. Alternatively, other methods can be used, such as taking the average of every three rows in the time spectrum array and then corresponding to a row in the time series matrix, and then performing an inverse FFT transformation to achieve the calculation in this step.

[0027] S8, MPPT Algorithm Control: The photovoltaic string is controlled using the MPPT conductance incremental algorithm, with the electrical parameter increment of the next time series used as the input term for the electrical parameter increment in the MPPT conductance incremental algorithm. Generally, the MPPT conductance incremental algorithm is a relatively ready-made packaged module, and its input term is the electrical parameter increment d of the current time series. t This step essentially involves replacing the input items and then directly calling the existing MPPT conductance incremental algorithm module.

[0028] Second implementation method: The second embodiment of the present invention differs from the first embodiment mainly in that: The full probability weighting function in step S5 refers to multiplying each numerical item in the time series array by a corresponding probability numerical item in the probability array, and then summing all the multiplication results. , In the formula, As probability weights, one set is cached for each group of photovoltaic strings.

[0029] The current and predicted electrical parameters both include current and voltage. As can be seen from the foregoing, power and other electrical parameters can also be used, but current and voltage are relatively easier to measure and process, so current and voltage are preferred.

[0030] In step S3, the preset multiple of the ratio value is in the range of 0.5 to 1.0.

[0031] The preset ratio in step S3 is 0.63.

[0032] The probability array is set for each photovoltaic string, and the number of items in the probability array is the same as the number of items in the time series array.

[0033] In step S6, the time series array of each photovoltaic string is arranged as a row in the time series matrix.

[0034] Third implementation method: The main difference between the third embodiment of the present invention and the first embodiment is that: The number of items p in the time series array is not less than the number of photovoltaic strings n. This ensures that the calculation is relatively simple when performing two-dimensional Fourier transforms and predicting electrical parameter increments. It is easy to see that the ideal case is p=n.

[0035] In step S7, the time spectrum array is first truncated according to a preset selected range, and the portion of the time spectrum array outside the selected range is completely empty. Since the time spectrum array is a two-dimensional frequency domain spectrum, its meaning is relatively definite. Therefore, a preset selected range can be set in advance to filter out frequency fluctuations (especially high-frequency jitter) that are not within the control consideration range, so as to achieve more precise control.

[0036] Steps S1 to S4 are performed on each photovoltaic string separately; The calculation of electrical parameters involved in steps S1 to S7 is performed separately for different types of electrical parameters.

[0037] In summary, this invention analyzes the frequency domain characteristics of each component group through two-dimensional Fourier transform of the time-series matrix, selects dominant frequency components by combining time-spectrum array truncation, dynamically predicts the electrical parameter increment of the next time series, and quantifies the impact differences of environmental abrupt changes on photovoltaic modules based on the full probability weight function, thereby improving the prediction accuracy of the global optimal point. The predicted electrical parameter increment is directly input into the MPPT conductance increment algorithm, replacing the traditional increment calculation that relies on the current instantaneous value, thereby adjusting the operating point in advance, reducing the oscillation period, and adjusting the power supply control quantity by the ratio of the difference to the current electrical parameter, optimizing the weight of the time-series array with a preset multiple, and accelerating the system convergence to the new maximum power point. The time-series array is arranged into a matrix for 2D-FFT, and the information volume is compressed by frequency domain analysis to reduce computational complexity. An independent probability array is configured for each component group, which can effectively adapt to the output attenuation curves and other characteristics of different component models, improving the personalized prediction accuracy.

Claims

1. A method for controlling a multi-model photovoltaic module array, characterized in that: Includes the following steps: S1. Obtain current electrical parameters: Obtain the current electrical parameters from the photovoltaic string output circuit; S2. Comparison and analysis of electrical parameter difference and threshold: Calculate the difference between the current electrical parameter and the predicted electrical parameter, and determine whether the difference is greater than the preset threshold. If it is greater, proceed to step S3; otherwise, proceed to step S4. S3. Adjustment of power supply control quantity and timing array values: Adjust the power supply control quantity of the current photovoltaic string according to the ratio of the difference to the current electrical parameters, and adjust the values ​​in the timing array by a preset multiple of the ratio, and then proceed to step S5. S4. Reorder the timing array: Place the current electrical parameter at the beginning of the timing array and clear the last element of the timing array; S5. Predict the electrical parameters of the next time series: The time series arrays of multiple photovoltaic strings are calculated using the full probability weighting function to obtain the predicted electrical parameters of the next time series. S6. Two-dimensional FFT: Arrange the time series arrays of multiple photovoltaic strings into a time series matrix in order, and then perform a two-dimensional Fourier transform on the time series matrix to obtain the time spectrum matrix; S7. Predict electrical parameter increments: Calculate the electrical parameter increments for the next time series based on the time-frequency array; S8, MPPT algorithm control: The MPPT conductance incremental algorithm is used to control the photovoltaic string, and the electrical parameter increment of the next time sequence is used as the electrical parameter increment input in the MPPT conductance incremental algorithm.

2. The multi-model photovoltaic module array control method as described in claim 1, characterized in that: The current electrical parameters and predicted electrical parameters both include current and voltage.

3. The multi-model photovoltaic module array control method as described in claim 1, characterized in that: In step S3, the preset multiple of the ratio value is in the range of 0.5 to 1.

0.

4. The multi-model photovoltaic module array control method as described in claim 3, characterized in that: The preset ratio in step S3 is 0.

63.

5. The multi-model photovoltaic module array control method as described in claim 1, characterized in that: The full probability weighting function in step S5 refers to multiplying each numerical item in the time series array with a corresponding probability numerical item in the probability array, and then summing all the multiplication results.

6. The multi-model photovoltaic module array control method as described in claim 5, characterized in that: The probability array is set for each photovoltaic string, and the number of items in the probability array is the same as the number of items in the time series array.

7. The multi-model photovoltaic module array control method as described in claim 1, characterized in that: In step S6, the time series array of each photovoltaic string is arranged as a row in the time series matrix.

8. The multi-model photovoltaic module array control method as described in claim 7, characterized in that: The number of items p in the time series array is not less than the number of photovoltaic strings n.

9. The multi-model photovoltaic module array control method as described in claim 1, characterized in that: In step S7, the time spectrum array is first truncated according to the preset selected range, and the part of the time spectrum array that is not in the selected range is set to empty.

10. The multi-model photovoltaic module array control method as described in claim 1, characterized in that: Steps S1 to S4 are performed on each photovoltaic string separately; The calculation of electrical parameters involved in steps S1 to S7 is performed separately for different types of electrical parameters.

Images