Photovoltaic composite energy storage system control method and device based on dynamic adaptive frequency division

Abstract

The invention relates to the field of photovoltaic power generation, in particular to a photovoltaic composite energy storage system control method and device based on dynamic adaptive frequency division. A dynamic adaptive frequency division composite energy storage control algorithm is designed, the residual capacity and the power difference value of the lithium battery are used as input, dynamic frequency division is used as output of a fuzzy controller, the fuzzy controller is used for adaptively adjusting a frequency threshold value, power is subjected to fast Fourier transform, the high-frequency part of the power is absorbed / released by a super capacitor, and the high-frequency part of the power is absorbed / released by a super capacitor. And the low-frequency power part is absorbed / released by the lithium battery. The energy utilization is improved, the bus voltage is kept stable, frequent charging and discharging of the lithium battery are reduced, and the service life of the lithium battery is prolonged. The power distribution conflict is effectively solved, and the system stability is improved. An improved photovoltaic maximum power tracking algorithm is designed, global search of a genetic algorithm and local fine tuning of derivative acceleration are combined, efficiency and precision are considered, and global maximum power tracking is achieved.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic power generation, and in particular to a control method and device for a photovoltaic composite energy storage system based on dynamic adaptive frequency division. Background Art

[0002] Photovoltaic power generation is currently a common form of renewable energy generation. However, it is constrained by primary energy supply and struggles to meet the grid's peak and frequency regulation needs. Currently, my country faces a serious problem of curtailed wind and solar power generation. To improve the grid-connected utilization efficiency of wind and photovoltaic power generation, energy storage technology is the primary solution for smoothing fluctuations in renewable energy output. Currently, energy storage devices are primarily categorized as energy-based and power-based. Commonly used energy storage devices include lithium-ion batteries, lead-acid batteries, and supercapacitors. Batteries have high energy density but low power density, while supercapacitors have high power density but low energy density. Combining the advantages of these two types of energy storage can improve the power response speed of energy storage systems. Hybrid energy storage systems can store excess power in DC microgrids and provide power support to loads when renewable energy output is insufficient. With their rapid response, they complement renewable energy sources and are widely used to smooth power fluctuations in microgrids.

[0003] The power allocation control strategy for hybrid energy storage systems is key to achieving the complementary advantages of the two types of energy storage media. Existing hybrid energy storage power scheduling uses a fixed frequency threshold, meaning the high-frequency and low-frequency boundaries are fixed. This makes it difficult to adapt to sudden changes in illumination and load fluctuations, resulting in low synergistic efficiency between supercapacitors and lithium batteries. Furthermore, the system fails to distinguish between high and low power frequencies, leading to a decrease in the cycle life of lithium batteries due to frequent charging and discharging. Load power fluctuations are not considered during power allocation, resulting in poor control robustness and delayed response when the load fluctuates significantly. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a photovoltaic composite energy storage system control method and device based on dynamic adaptive frequency division. The frequency division threshold is dynamically and adaptively adjusted based on the remaining capacity and power difference of the lithium battery. The high-frequency component is borne by the supercapacitor and the low-frequency component is borne by the lithium battery, thereby improving energy utilization, reducing the frequent charging and discharging of the lithium battery, and increasing the life of the lithium battery.

[0005] To solve the above technical problems, the present invention adopts a technical solution: a photovoltaic composite energy storage system control method based on dynamic adaptive frequency division. The photovoltaic composite energy storage system includes a photovoltaic panel, an electrical load, a lithium battery, and a supercapacitor. The photovoltaic panel is connected to the electrical load via a boost converter and a DC bus. The lithium battery and supercapacitor are respectively connected to the DC bus via a buck-boost converter. The method includes the following steps: S01. Use the photovoltaic maximum power tracking algorithm to obtain the maximum power of the photovoltaic panel , using BP neural network to predict power load ; S02. Calculate power difference , , and obtain the remaining capacity of the lithium battery ; S03, the power difference , Remaining capacity of lithium battery Input fuzzy controller, fuzzy controller first calculates the power difference , Remaining capacity of lithium battery , Frequency division threshold adjustment Perform grade division and membership function calculation; then establish a fuzzy rule logic table based on fuzzy rules; use Maidani reasoning to obtain the fuzzy reasoning calculation value, use the center of gravity method to defuzzify the fuzzy reasoning calculation value, and obtain the frequency threshold adjustment value. ; S05, the frequency division threshold at the previous moment and the frequency division threshold adjustment amount Sum and get the frequency division threshold at the current moment: S06, use fast Fourier transform to convert the power difference Decompose into high-frequency components and low-frequency components : , in is a high-pass filter, is the frequency division threshold at the current moment; The high frequency components Assigned to the supercapacitor, the low frequency component Assigned to lithium batteries.

[0006] Furthermore, the photovoltaic maximum power tracking algorithm is used to obtain the maximum power of the photovoltaic panel The process is: S11. Derivative operation is performed on the power curve of the photovoltaic panel to preliminarily determine the location of the extreme point. The extreme point is the one where the derivative result is zero; S12. Describe the output characteristics of the photovoltaic cell using a single diode equivalent circuit. The description formula is: , Where I and V represent the output current and output voltage of the photovoltaic panel respectively. represents the photogenerated current, Represents the diode reverse saturation current, 、 represents the series resistance and parallel resistance, n represents the diode ideality factor, , k is the Boltzmann constant, T is the temperature, and q is the electron charge; The maximum power point condition is: , in is the maximum power point voltage, is the maximum power point current; S13. Design a genetic algorithm. First, initialize the population and use real number coding. Each individual represents a photovoltaic output voltage value. , the range is , is the open circuit voltage, the population size is N, Randomly generate initial voltage within the interval; Take the output power of photovoltaic panels as the fitness function, and calculate the fitness of each individual when generating the initial population. , individuals with derivatives of zero or close to zero are preferentially retained; Selection operation: the selection probability is allocated according to the fitness value ratio. Individuals with high fitness are more likely to be selected, and the current optimal individual is directly retained to the next generation; Crossover operation, crossover operation is performed on two parent individuals to generate offspring, the crossover probability is between; Mutation operation, for individual Add random perturbation, the mutation probability is between; Calculate the fitness value. If the maximum number of iterations is reached or the change in the optimal fitness for consecutive k generations is less than , If it is a predetermined condition for the termination of the genetic algorithm, the iteration is terminated.

[0007] Furthermore, the genetic algorithm also includes local gradient fine-tuning, and the best individual in each generation Perform gradient descent optimization to obtain a new optimal individual , ,in is the step length, P and V are the optimal individuals Corresponding power and voltage.

[0008] Furthermore, a genetic algorithm is used to optimize the BP neural network. The neural network parameters are used as individuals of the genetic algorithm, and the inverse of the BP neural network prediction error is used as the fitness function. The optimization of each individual is evaluated by the fitness function, and then a new generation of population is generated through selection, crossover, and mutation. The optimization is iterated until the stopping condition is met.

[0009] Furthermore, neural network parameters are weights and biases.

[0010] Furthermore, in step S03, the remaining capacity of the lithium battery is divided into three levels: low, medium, and high; the power difference is divided into five levels: negative large, negative small, zero, positive small, and positive large; the frequency division threshold adjustment amount is divided into five levels: very low, low, medium, high, and very high; and a triangular membership function is used to calculate the membership function of the remaining capacity of the lithium battery, the power difference, and the frequency division threshold adjustment amount. The fuzzy rule is: when the SOC is low: increase the frequency division threshold adjustment amount , let the supercapacitor bear more power (including some low frequency) to protect the lithium battery; when ΔP changes drastically (ΔP is negative or positive): increase the frequency division threshold adjustment amount , ensuring that the supercapacitor can handle high-frequency fluctuations; when the SOC is high and ΔP is stable (ΔP is small negative, zero, small positive): reduce the frequency division threshold adjustment amount , allowing lithium batteries to handle more low-frequency power.

[0011] Furthermore, the frequency division threshold at the current moment is low-pass filtered, and the final frequency division threshold is obtained by a first-order low-pass filter: , is the filter coefficient, ranging from 0.1 to 0.3, is the final frequency division threshold, is the final frequency division threshold of the previous moment, is the final frequency division threshold of the first two moments.

[0012] Furthermore, hysteresis control is adopted for the supercapacitor.

[0013] Furthermore, PID control is adopted for the lithium battery.

[0014] The present invention also discloses a photovoltaic composite energy storage system control device based on dynamic adaptive frequency division, comprising a processor and a memory storing program instructions, wherein the processor is configured to execute the photovoltaic composite energy storage system control method based on dynamic adaptive frequency division as described above when running the program instructions.

[0015] Beneficial effects of the present invention: 1. A composite energy storage control algorithm with dynamic adaptive frequency division is designed, which solves the power allocation conflict and voltage stability problems through adaptive adjustment of the dynamic frequency division threshold and power balance control architecture. The composite energy storage control algorithm with dynamic adaptive frequency division designed by the present invention uses a fuzzy controller to adaptively adjust the frequency division threshold, performs a fast Fourier transform on the power, and the high-frequency part of the power is absorbed / released by the supercapacitor, and the low-frequency part of the power is absorbed / released by the lithium battery. It achieves the improvement of energy utilization, maintains the stability of the bus voltage, reduces the frequent charging and discharging of the lithium battery, and increases the life of the lithium battery. It effectively solves the power allocation conflict and improves the stability of the system.

[0016] 2. An improved photovoltaic maximum power point tracking (MPPT) algorithm was designed, combining the global search capabilities of a genetic algorithm with the local fine-tuning of derivative acceleration. This algorithm balances efficiency and accuracy, achieving global maximum power point tracking (MPPT). This allows photovoltaic power generation to collect more energy. This genetic algorithm-based MPPT control avoids the local oscillation issues of traditional methods through global optimization, and the integration of derivative acceleration further improves tracking speed and accuracy.

[0017] 3. A BP neural network optimized by a genetic algorithm predicts load power, enabling proactive control of photovoltaic and energy storage systems to ensure timely and reliable power delivery to the system. By optimizing the initial parameters of the BP neural network using a genetic algorithm, the accuracy of load forecasting can be significantly improved.

[0018] 4. A dynamic limiting filter is designed. To avoid frequent sudden changes in load energy storage control and limit the frequency division threshold to a reasonable range, the present invention adds a first-order low-pass filter to the algorithm to avoid frequent jumps.

[0019] 5. Based on the charge and discharge characteristics of lithium batteries and supercapacitors, a coordinated control system for the composite energy storage system was designed. Supercapacitors can charge and discharge quickly, so hysteresis control is used to avoid oscillation near the crossover point. Lithium batteries charge and discharge slowly, but discharge slowly and last longer, so PID control is used. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is the schematic diagram of the photovoltaic composite energy storage system; Figure 2 Flowchart of this method; Figure 3 Flowchart for realizing photovoltaic maximum power point tracking using improved genetic algorithm; Figure 4 is the membership function curve of the remaining capacity of the lithium battery; Figure 5 is the membership function curve of power difference; Figure 6 is the membership function curve of the frequency division threshold adjustment amount; Figure 7 This is the working flow diagram of the fuzzy controller. DETAILED DESCRIPTION

[0021] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0022] Example 1 This embodiment discloses a photovoltaic composite energy storage system control method based on dynamic adaptive frequency division, such as Figure 1As shown in the figure, the photovoltaic composite energy storage system includes photovoltaic panels, lithium batteries, supercapacitors and electrical loads. The photovoltaic panels are power generation components and are connected to the electrical loads through boost converters and DC busbars. The lithium batteries and supercapacitors are energy storage components and are connected to the DC busbars through buck-boost converters.

[0023] like Figure 2 As shown, the method includes the following steps: S01. Use the photovoltaic maximum power tracking algorithm to obtain the maximum power of the photovoltaic panel , using BP neural network to predict power load ; S02. Calculate power difference , , and obtain the remaining capacity of the lithium battery ; S03, the power difference , Remaining capacity of lithium battery Input fuzzy controller, fuzzy controller first calculates the power difference , Remaining capacity of lithium battery , Frequency division threshold adjustment Perform grade division and membership function calculation; then establish a fuzzy rule logic table based on fuzzy rules; use Maidani reasoning to obtain the fuzzy reasoning calculation value, use the center of gravity method to defuzzify the fuzzy reasoning calculation value, and obtain the frequency threshold adjustment value. ; S05, the frequency division threshold at the previous moment and the frequency division threshold adjustment amount Sum and get the frequency division threshold at the current moment: S06, use fast Fourier transform to convert the power difference Decompose into high-frequency components and low-frequency components : , in is a high-pass filter, is the frequency division threshold at the current moment; , is the cut-off angular frequency, ; The high frequency components Assigned to the supercapacitor, the low frequency component Assigned to lithium batteries.

[0024] Commonly used maximum power point tracking (MPPT) control strategies include hill climbing, perturbation and observation, and conductance increment. These algorithms offer the advantage of simple control, finding peak power by searching for extreme points along a curve. However, they can be prone to becoming trapped in local extreme points, which deviate from the global maximum power point. To address these shortcomings of existing MPPT algorithms, this embodiment employs a genetic algorithm with pilot operation acceleration to achieve photovoltaic MPPT control.

[0025] like Figure 3 As shown, the genetic algorithm accelerated by the pilot operation is used to obtain the maximum power of the photovoltaic panel The process is: S11. Perform a derivative operation on the power curve of the photovoltaic panel to preliminarily determine the location of the extreme point. This operation speeds up the search speed, and only the point where the derivative is zero is needed to determine the extreme point.

[0026] The photovoltaic power curve has multiple extreme points, and the maximum power point is always at an extreme point. Therefore, we first take the derivative to preliminarily determine the extreme points, and then use the genetic algorithm to compare these extreme points to determine the maximum power point.

[0027] S12. Describe the output characteristics of the photovoltaic cell using a single diode equivalent circuit. The description formula is: , This formula is a universal mathematical model for photovoltaic cells, which can reflect the point where the output voltage and output power of photovoltaic power generation reach their maximum values. Where I and V represent the output current and output voltage of the photovoltaic panel respectively. represents the photogenerated current, Represents the diode reverse saturation current, 、 represents the series resistance and parallel resistance, n represents the diode ideality factor, , k is the Boltzmann constant, T is the temperature, and q is the electron charge; The maximum power point condition is: , in is the maximum power point voltage, is the maximum power point current.

[0028] S13. Design a genetic algorithm. First, initialize the population and use real number coding. Each individual represents a photovoltaic output voltage value. , the range is , is the open circuit voltage, the population size is N, N is set according to the resources, and in this embodiment, N is set between [20,50]. The initial voltage is randomly generated within the interval.

[0029] Take the output power of photovoltaic panels as the fitness function, and calculate the fitness of each individual when generating the initial population. , prioritize individuals whose derivatives are zero or close to zero to reduce invalid searches. In the above description, close to zero means that the difference from zero is within the set range, such as [-0.1, 0.1] is close to zero.

[0030] The selection operation allocates the selection probability according to the proportion of fitness value. Individuals with high fitness are more likely to be selected, and the current optimal individual is directly retained to the next generation to avoid the loss of high-quality genes.

[0031] Crossover operation, for two parent individuals and Perform crossover operation to generate offspring , , the crossover probability is between.

[0032] Mutation operation, for individual Add random perturbations, , is the individual after adding random disturbance, is a random perturbation, is the variation strength, the value range is [0.05V oc , 0.1V oc ], the mutation probability is between.

[0033] Calculate the fitness value. If the maximum number of iterations is reached or the change in the optimal fitness for consecutive k generations is less than , It is a predetermined condition for the termination of the genetic algorithm. In this embodiment, If the value is 0.5, the iteration is terminated. The maximum number of iterations is in the range of [50, 100].

[0034] Local gradient fine-tuning, in order to accelerate the convergence to the precise maximum power point, local gradient fine-tuning is performed, and the optimal individual in each generation Perform gradient descent optimization to obtain a new optimal individual , ,in is the step length, P and V are the optimal individuals Corresponding power and voltage.

[0035] Load power (electricity load) forecasting uses historical electricity consumption data, meteorological information, date type, and other factors to predict future electricity demand. Accurate load forecasting is crucial for power grid scheduling and energy management. Neural networks, due to their powerful nonlinear modeling capabilities and advantages in processing time series data, have become a core method for load forecasting. This example uses a genetic algorithm to optimize the BP neural network load forecasting algorithm.

[0036] First, the BP neural network is a multi-layer feedforward network that uses a backpropagation algorithm to adjust weights and biases. Its basic structure consists of an input layer, a hidden layer, and an output layer. The input layer receives feature data (such as historical load, temperature, and date), the hidden layer performs nonlinear transformations, and the output layer produces predictions. During training, the error between the predicted and actual values ​​is calculated, and parameters are adjusted through backpropagation to minimize the loss function.

[0037] A genetic algorithm is a heuristic optimization algorithm that simulates natural selection and heredity. It searches for the optimal solution in the solution space through selection, crossover, and mutation. Here, a genetic algorithm can be used to optimize the initial weights and biases of a BP neural network, avoiding the problem of traditional gradient descent methods falling into local optima.

[0038] This example combines the two. A genetic algorithm is used to optimize the parameters of the BP neural network to improve prediction accuracy. Specifically, each individual in the genetic algorithm represents a set of neural network parameters (weights and biases). The fitness function (the inverse of the prediction error) evaluates the quality of each individual. A new generation of populations is generated through selection, crossover, and mutation. Optimization is then iterated until a stopping condition is met.

[0039] Optimizing the initial parameters of the BP neural network using a genetic algorithm significantly improves the accuracy of power load forecasting. The model, through the synergy of global search and local fine-tuning, effectively addresses complex nonlinear relationships and is suitable for short- to medium- to long-term load forecasting needs in power systems.

[0040] This embodiment is directed to a composite energy storage system composed of supercapacitors and lithium batteries. The composite energy storage system combines the excellent features of supercapacitors and lithium batteries. Supercapacitors have high power density and can exchange power quickly, but have low energy density and cannot discharge for a long time. Lithium batteries have high energy density and can discharge for a long time, but cannot exchange power quickly. Although composite energy storage combines the excellent features of supercapacitors and lithium batteries, lithium batteries and supercapacitors also have their own defects, and their control is a difficult point. See the background technology section for details. To address this, this embodiment designs a composite energy storage control algorithm with dynamic adaptive frequency division, which uses a fuzzy controller to adaptively adjust the frequency division threshold and perform a fast Fourier transform on the power. The high-frequency part of the power is absorbed / released by the supercapacitor, and the low-frequency part of the power is absorbed / released by the lithium battery. This achieves improved energy utilization, maintains bus voltage stability, reduces the frequent charging and discharging of lithium batteries, and increases the life of lithium batteries. It effectively resolves power allocation conflicts and improves system stability.

[0041] The dynamic frequency division threshold is achieved by real-time monitoring of system status parameters (lithium battery remaining capacity SOC, power difference) and adaptively adjusting the frequency division threshold to achieve reasonable power distribution between supercapacitors and lithium batteries.

[0042] According to the overall power balance equation of the photovoltaic energy storage system, the relationship between power balance control and DC bus voltage stability can be known: , in is the photovoltaic power generation power, and are the power of supercapacitor and lithium battery respectively, is the load power, is the DC bus power.

[0043] DC bus voltage The dynamic equation is: , When the power is balanced, the bus voltage is stable. ,but .

[0044] in Indicates the DC bus capacitance 、 、 、 They represent the photovoltaic panel circuit, the current on the supercapacitor, the lithium battery current and the load current respectively.

[0045] As long as power balance is achieved, the DC bus voltage can be kept stable, thereby ensuring constant power quality for the loads. This ensures a constant voltage for the loads. Therefore, the overall goal of this method is to achieve power balance and voltage stability by dynamically coordinating the power distribution (charging and discharging) between the supercapacitor and lithium battery.

[0046] First, the frequency division threshold is the critical frequency that decomposes the power signal into high-frequency and low-frequency components. Power signals above the frequency division threshold are called high-frequency components and are processed by supercapacitors to address rapid fluctuations (such as sudden load changes and photovoltaic fluctuations). Power signals below or equal to the frequency division threshold are called low-frequency components and are processed by lithium batteries to provide stable energy.

[0047] The purpose of dynamic adaptive frequency division is to optimize the frequency division threshold according to the system status and balance the response speed and energy storage life. Dynamic adaptive frequency division adopts fuzzy logic and is realized through a fuzzy controller. The input of the fuzzy controller is the remaining capacity of the lithium battery. , power difference , the output is the frequency division threshold adjustment The fuzzy controller mainly includes three processes: fuzzification, fuzzy reasoning and defuzzification. The working process of the fuzzy controller is as follows: Figure 7 shown.

[0048] Fuzzy, first divide the remaining capacity of the lithium battery into three levels: low, medium, and high. The three levels are expressed as {low L, medium M, high H}. The power difference is divided into five levels: negative large, negative small, zero, positive small, and positive large. The five levels are expressed as {negative large NB, negative small NS, zero 0, positive small PS, positive large PB}. The frequency division threshold adjustment amount is divided into five levels: extremely low, low, medium, high, and extremely high. The five levels are expressed as {extremely low VL, low L, medium M, high H, extremely high VH}. The triangular membership function is used to calculate the membership function of the remaining capacity of the lithium battery, the power difference, and the frequency division threshold adjustment amount. The specific membership function curve is as follows: Figure 4 、 5 , as shown in 6.

[0049] The fuzzy rules are: (1) When the remaining capacity of the lithium battery is low: increase the frequency division threshold adjustment amount , allowing supercapacitors to bear more power (including some low-frequency) and protect lithium batteries; (2) When ΔP changes dramatically (ΔP is large in negative or positive): increase the frequency division threshold adjustment amount , ensuring that the supercapacitor can handle high frequency fluctuations (3) When the remaining capacity of the lithium battery is high and ΔP is stable (ΔP is small negative, zero, and small positive): reduce the frequency division threshold adjustment amount , allowing lithium batteries to handle more low-frequency power The fuzzy rule table established based on fuzzy rules is: Table 1 Fuzzy rule table Fuzzy reasoning, fuzzy reasoning adopts Maidani reasoning, and the calculation formula is: , in Calculate values ​​for fuzzy inference; is the membership value of the remaining capacity of the lithium battery; is the power difference membership value; To take the smaller operation.

[0050] Defuzzification,defuzzification adopts the centroid method, and the calculation formula is: , where j represents the jth fuzzy rule.

[0051] To avoid frequent sudden changes in load energy storage control, the frequency division threshold is limited to a reasonable range (such as 0.5 Hz to 5 Hz). The frequency division threshold at the current moment is low-pass filtered. The final frequency division threshold obtained by the first-order low-pass filter is: , in is the filter coefficient, ranging from 0.1 to 0.3, is the final frequency division threshold, is the final frequency division threshold of the previous moment, is the final frequency division threshold of the first two moments.

[0052] This embodiment adopts cooperative control for the composite energy storage system. The supercapacitor absorbs high-frequency components and requires a fast response time. Hysteresis control is used to achieve millisecond-level response. The control formula is: , in is the supercapacitor reference current, is the high-frequency component of power, is the DC bus voltage.

[0053] Lithium batteries absorb low-frequency components and require small fluctuations and a stable system. PID algorithm is used for smooth charging and discharging. The control formula is: , in is the reference current of lithium battery, is the low-frequency component of power, is the lithium battery power, is the proportional control coefficient, is the integral control coefficient, is the differential control coefficient.

[0054] To protect lithium batteries and avoid excessive temperature rise, the charge and discharge rate is limited: , C is the capacity of the lithium battery.

[0055] Example 2 This embodiment provides a control device for a photovoltaic composite energy storage system based on dynamic adaptive frequency division, including a processor and memory. Optionally, the device may also include a communication interface and a bus. The processor, communication interface, and memory 301 may communicate with each other via bus 303. The communication interface may be used for information transmission. Processor 304 may invoke logic instructions in the memory to execute the control method for a photovoltaic composite energy storage system based on dynamic adaptive frequency division described in the above embodiment.

[0056] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product.

[0057] Memory, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as the program instructions / modules corresponding to the methods in the embodiments of the present disclosure. The processor executes the program instructions / modules stored in the memory to perform functional applications and data processing, thereby implementing the photovoltaic composite energy storage system control method based on dynamic adaptive frequency division in the above-mentioned embodiments.

[0058] The memory may include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data generated based on the use of the terminal device. Furthermore, the memory may include high-speed random access memory and non-volatile memory.

[0059] The present invention designs an improved photovoltaic maximum power point tracking algorithm, which combines the global search of genetic algorithm and the local fine-tuning of derivative acceleration, takes into account both efficiency and accuracy, realizes global maximum power point tracking, and photovoltaic power generation collects more light energy through maximum power point tracking.

[0060] A composite energy storage control algorithm with dynamic adaptive frequency division was designed. A fuzzy controller was used to adaptively adjust the frequency division threshold. The power was fast Fourier transformed, with the high-frequency portion of the power absorbed / released by the supercapacitor and the low-frequency portion by the lithium battery. This algorithm improved energy utilization, maintained bus voltage stability, reduced frequent charging and discharging of the lithium battery, and increased its lifespan. It effectively resolved power allocation conflicts and improved system stability. Adaptive frequency division power adjustment effectively resolved power allocation conflicts, improving system stability and energy storage lifespan. A neural network optimized by a genetic algorithm predicted power load, based on which the future power change rate was determined, allowing for proactive control intervention. The fuzzy controller took the remaining capacity and power difference of the lithium battery as inputs, and dynamically divided the frequency as its output. The high-frequency component was absorbed / released by the supercapacitor and the low-frequency component by the lithium battery. A dynamic limiting filter was designed to limit frequent sudden changes in load energy storage control and to limit the frequency division threshold to a reasonable range. A first-order low-pass filter was incorporated into the algorithm to prevent frequent jumps. Based on the charge and discharge characteristics of lithium batteries and supercapacitors, a coordinated control system for the composite energy storage system was designed. Supercapacitors can charge and discharge quickly, so hysteresis control is used to avoid oscillation near the crossover point. Lithium batteries charge and discharge slowly, but discharge slowly and last longer, so PID control is used.

[0061] The above description and the accompanying drawings sufficiently illustrate the embodiments of the present disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible variations. Unless expressly required, individual components and functions are optional, and the order of operations may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. Moreover, the terms used in this application are only used to describe the embodiments and are not used to limit the scope of protection. As used in the description herein, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to also include the plural forms. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of one or more associated listings. In addition, when used in this application, the term "comprise" and its variations "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, an element defined by the sentence "comprising a..." does not exclude the presence of other identical elements in the process, method or device that includes the element. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments can be referenced to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can be referred to the description of the method part.

[0062] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. The technicians will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0063] In the embodiments disclosed herein, the disclosed methods and products (including but not limited to devices and equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units may be merely a logical functional division. In actual implementation, other division methods may be used, such as combining or integrating multiple units or components into another system, or omitting or disabling some features. In addition, the coupling or direct coupling or communication connection shown or discussed between each other may be through some interface, or the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to implement the present embodiments according to actual needs. In addition, the functional units in the embodiments of the present disclosure may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit.

Claims

1. A control method for a photovoltaic composite energy storage system based on dynamic adaptive frequency division. The photovoltaic composite energy storage system includes a photovoltaic panel, an electrical load, a lithium battery, and a supercapacitor. The photovoltaic panel is connected to the electrical load via a boost converter and a DC bus. The lithium battery and supercapacitor are each connected to the DC bus via a buck-boost converter. The method is characterized by: This method comprises the following steps: S01. Use the photovoltaic maximum power tracking algorithm to obtain the maximum power of the photovoltaic panel , using BP neural network to predict power load ; S02. Calculate power difference , , and obtain the remaining capacity of the lithium battery ; S03, the power difference , Remaining capacity of lithium battery Input fuzzy controller, fuzzy controller first calculates the power difference , Remaining capacity of lithium battery , Frequency division threshold adjustment Perform grade division and membership function calculation; then establish a fuzzy rule logic table based on fuzzy rules; use Maidani reasoning to obtain the fuzzy reasoning calculation value, use the center of gravity method to defuzzify the fuzzy reasoning calculation value, and obtain the frequency threshold adjustment value. ; S05, the frequency division threshold at the previous moment and the frequency division threshold adjustment amount Sum and get the frequency division threshold at the current moment: S06, use fast Fourier transform to convert the power difference Decompose into high-frequency components and low-frequency components : ， in is a high-pass filter, is the frequency division threshold at the current moment; The high frequency components Assigned to the supercapacitor, the low frequency component Assigned to lithium batteries.

2. The photovoltaic composite energy storage system control method based on dynamic adaptive frequency division according to claim 1 is characterized in that: Use the photovoltaic maximum power point tracking algorithm to obtain the maximum power of the photovoltaic panel The process is: S11. Derivative operation is performed on the power curve of the photovoltaic panel to preliminarily determine the location of the extreme point. The extreme point is the one where the derivative result is zero; S12. Describe the output characteristics of the photovoltaic cell using a single diode equivalent circuit. The description formula is: ， Where I and V represent the output current and output voltage of the photovoltaic panel respectively. represents the photogenerated current, Represents the diode reverse saturation current, 、 represents the series resistance and parallel resistance, n represents the diode ideality factor, , k is the Boltzmann constant, T is the temperature, and q is the electron charge; The maximum power point condition is: ， in is the maximum power point voltage, is the maximum power point current; S13. Design a genetic algorithm. First, initialize the population and use real number coding. Each individual represents a photovoltaic output voltage value. , the range is , is the open circuit voltage, the population size is N, Randomly generate initial voltage within the interval; Take the output power of photovoltaic panels as the fitness function, and calculate the fitness of each individual when generating the initial population. , individuals with derivatives of zero or close to zero are preferentially retained; Selection operation: the selection probability is allocated according to the fitness value ratio. Individuals with high fitness are more likely to be selected, and the current optimal individual is directly retained to the next generation; Crossover operation, crossover operation is performed on two parent individuals to generate offspring, the crossover probability is between; Mutation operation, for individual Add random perturbation, the mutation probability is between; Calculate the fitness value. If the maximum number of iterations is reached or the change in the optimal fitness for consecutive k generations is less than , If it is a predetermined condition for the termination of the genetic algorithm, the iteration is terminated.

3. The photovoltaic composite energy storage system control method based on dynamic adaptive frequency division according to claim 2, characterized in that: The genetic algorithm also includes local gradient fine-tuning, and the best individual in each generation Perform gradient descent optimization to obtain a new optimal individual , ,in is the step length, P and V are the optimal individuals Corresponding power and voltage.

4. The photovoltaic composite energy storage system control method based on dynamic adaptive frequency division according to claim 1, characterized in that: Use genetic algorithm to optimize BP neural network, take neural network parameters as individuals of genetic algorithm, take the inverse of BP neural network prediction error as fitness function, evaluate the optimization of each individual through fitness function, then generate a new generation of population through selection, crossover and mutation, and iterate optimization until the stopping condition is met.

5. The photovoltaic composite energy storage system control method based on dynamic adaptive frequency division according to claim 4 is characterized in that: Neural network parameters are weights and biases.

6. The photovoltaic composite energy storage system control method based on dynamic adaptive frequency division according to claim 1, characterized in that: In step S03, the remaining capacity of the lithium battery is divided into three levels: low, medium, and high; the power difference is divided into five levels: negative large, negative small, zero, positive small, and positive large; the frequency division threshold adjustment amount is divided into five levels: very low, low, medium, high, and very high; the triangular membership function is used to calculate the membership function of the remaining capacity of the lithium battery, the power difference, and the frequency division threshold adjustment amount. The fuzzy rule is: when the remaining capacity of the lithium battery is low: increase the frequency division threshold adjustment amount , allowing the supercapacitor to bear more power and protect the lithium battery; when ΔP is negative or positive, increase the frequency division threshold adjustment amount , to ensure that the supercapacitor can handle high-frequency fluctuations; when the remaining capacity of the lithium battery is high and ΔP is negative, zero, or positive, reduce the frequency division threshold adjustment amount , allowing lithium batteries to handle more low-frequency power.

7. The photovoltaic composite energy storage system control method based on dynamic adaptive frequency division according to claim 1, characterized in that: Perform low-pass filtering on the frequency division threshold at the current moment, and obtain the final frequency division threshold through the first-order low-pass filter: ， is the filter coefficient, ranging from 0.1 to 0.3, is the final frequency division threshold, is the final frequency division threshold of the previous moment, is the final frequency division threshold of the first two moments.

8. The photovoltaic composite energy storage system control method based on dynamic adaptive frequency division according to claim 1, characterized in that: Hysteresis control is used for supercapacitors.

9. The photovoltaic composite energy storage system control method based on dynamic adaptive frequency division according to claim 1, characterized in that: PID control is used for lithium batteries.

10. A photovoltaic composite energy storage system control device based on dynamic adaptive frequency division, comprising a processor and a memory storing program instructions, characterized in that: The processor is configured to execute the photovoltaic composite energy storage system control method based on dynamic adaptive frequency division according to any one of claims 1 to 9 when running the program instructions.

Images