Photovoltaic cell parameter identification method and system based on composite model

Abstract

The invention relates to the technical field of system optimization control, in particular to a photovoltaic cell parameter identification method and system based on a composite model. The method comprises the steps of obtaining real-time operation data of a photovoltaic cell, performing feature extraction, and generating preprocessed operation state data; calculating an initial parameter estimation value of the photovoltaic cell by adopting a single-diode model and a double-diode model, and generating an initial parameter sequence; performing parameter optimization by using a genetic algorithm and a particle swarm optimization algorithm to obtain an optimized photovoltaic cell parameter set; training an error compensation model of the long and short term memory network in combination with historical operation data, and correcting identification parameters; based on the optimized parameter set and the error compensation model, constructing a composite parameter identification model by adopting an adaptive fusion algorithm; and acquiring real-time monitoring data of the photovoltaic system, and calculating an optimal operation adjustment parameter. According to the invention, the photovoltaic cell parameter identification precision is improved, and the adaptability of the system to environmental changes is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of system optimization control, and in particular to a photovoltaic cell parameter identification method and system based on a composite model. Background Art

[0002] With the transformation of the global energy structure, photovoltaic power generation has been widely used as a clean and renewable energy source. However, because the power generation characteristics of photovoltaic cells are affected by multiple factors such as ambient temperature, light intensity, and battery aging, their operating state has strong nonlinear and time-varying characteristics. Traditional parameter identification methods have difficulty ensuring high accuracy and real-time performance in complex environments.

[0003] Existing technologies mainly use physical modeling methods (such as single-diode model and double-diode model) to calculate the key parameters of photovoltaic cells (such as photocurrent, saturation current, series resistance and parallel resistance). However, due to the low initial estimation accuracy of the parameters by the physical model, it is difficult to adapt to complex environmental changes, resulting in large identification errors. In order to improve the accuracy of parameter identification, some studies have introduced intelligent optimization algorithms, such as genetic algorithms and particle swarm optimization algorithms, to achieve global search and optimal solution of parameters through iterative optimization. However, the computational efficiency of intelligent optimization algorithms is low in dynamic environments, making it difficult to meet the real-time requirements of photovoltaic systems. In addition, deep learning technology has been used for photovoltaic system state prediction in recent years, but the use of deep learning alone for parameter identification is easily affected by the distribution of training data and is difficult to generalize to different operating conditions, resulting in insufficient system robustness.

[0004] Therefore, there is an urgent need for a composite model that integrates physical modeling, intelligent optimization and deep learning. It uses the real-time operating data, historical operating data and environmental change factors of photovoltaic cells to construct a high-precision parameter identification method, and realizes dynamic adjustment of the photovoltaic system through adaptive optimization strategies to improve the stability and overall power generation efficiency of the photovoltaic power generation system. Summary of the Invention

[0005] The present invention provides a photovoltaic cell parameter identification method and system based on a composite model. The technical problem actually solved by the patent of this invention is: how to accurately identify the key circuit parameters of photovoltaic cells based on the real-time operation data, historical operating data and environmental change factors of photovoltaic cells, combined with physical modeling, intelligent optimization algorithms and deep learning models, and dynamically adjust the operating status of the photovoltaic system through adaptive optimization strategies to improve the stability and overall power generation efficiency of the photovoltaic power generation system.

[0006] In order to solve the above problems, the present invention provides a photovoltaic cell parameter identification method based on a composite model, comprising:

[0007] Acquiring operating status data of the photovoltaic cell, including current, voltage, temperature, and light intensity data, and performing signal denoising, normalization, and feature extraction on the operating status data to generate pre-processed photovoltaic cell operating status data;

[0008] Based on the preprocessed photovoltaic cell operating state data, a single diode model and a double diode model are used to calculate preliminary parameter estimation values ​​of the photovoltaic cell, and an initial parameter sequence of the photovoltaic cell is generated;

[0009] Optimizing and calculating the initial parameter sequence of the photovoltaic cell, performing global search and optimal parameter solution using a genetic algorithm and a particle swarm optimization algorithm, and generating an optimized photovoltaic cell parameter set;

[0010] Based on the optimized photovoltaic cell parameter set, a deep neural network model is trained in combination with historical photovoltaic cell operation data to obtain a parameter prediction model with error compensation;

[0011] Based on the optimized photovoltaic cell parameter set and the error compensation parameter prediction model, an adaptive fusion algorithm is used to generate composite model identification parameters to construct an adaptive parameter module for the photovoltaic cell;

[0012] Based on the composite model identification parameters, real-time monitoring data of the photovoltaic system is obtained, and the photovoltaic system operation state adjustment parameters are dynamically calculated to generate an adaptive adjustment plan for the photovoltaic system;

[0013] Based on the photovoltaic system adaptive adjustment scheme, the control parameters of the photovoltaic system are optimized and adjusted, and the adjusted photovoltaic cell operating parameters are generated to realize the intelligent operation control of the photovoltaic cell.

[0014] Furthermore, the signal denoising includes:

[0015] The operating status data of the photovoltaic cell is subjected to wavelet transform denoising to eliminate high-frequency noise interference, and the denoised signal is optimized based on Kalman filtering to obtain the denoised photovoltaic cell time series data.

[0016] Furthermore, the calculation of the preliminary parameter estimation value of the photovoltaic cell includes:

[0017] The single diode model is used to calculate the equivalent circuit parameters of the photovoltaic cell, including photocurrent, saturation current, resistance parameters and series resistance, and the single diode model parameter sequence is obtained;

[0018] The double-diode model is used to calculate the current shunt parameters of the photovoltaic cell, and the nonlinear correction coefficient of the photovoltaic cell is calculated in combination with the shadow shielding effect to obtain the double-diode model parameter sequence;

[0019] A feature matching calculation is performed on the single diode model parameter sequence and the double diode model parameter sequence, a model with the smallest error is selected as a preliminary parameter estimation model, and an initial parameter sequence of the photovoltaic cell is generated.

[0020] Furthermore, the optimization calculation includes:

[0021] Based on the initial parameter sequence of the photovoltaic cell, a genetic algorithm is used to perform a global search, multiple parameter populations are initialized, and genetic mutation and selection operations are performed according to a target fitness function to obtain an optimized population parameter sequence;

[0022] Based on the optimized population parameter sequence, a particle swarm optimization algorithm is used to perform local fine-tuning, an optimal parameter value is searched based on a speed-position update mechanism, and key circuit parameters of the photovoltaic cell are iteratively optimized to obtain an optimized photovoltaic cell candidate parameter set;

[0023] The optimized photovoltaic cell candidate parameter set is evaluated for adaptability, an optimal solution is selected based on a multi-objective optimization criterion, and a final optimized photovoltaic cell parameter set is generated.

[0024] Furthermore, the training of the parameter prediction model for error compensation includes:

[0025] Based on the optimized photovoltaic cell parameter set, historical photovoltaic cell operation data is acquired, and the historical photovoltaic cell operation data is divided into time windows and feature label matched to generate a training data set;

[0026] Based on the training data set, a long short-term memory network deep neural network model is constructed to learn the time series characteristics of photovoltaic cell parameters and perform hyperparameter optimization to obtain a pre-trained error compensation model;

[0027] The optimized photovoltaic cell parameter set is input into the error compensation model, and a compensation value is calculated based on the historical error distribution of the photovoltaic cell to generate an error-corrected parameter prediction model.

[0028] Furthermore, the adaptive fusion algorithm includes:

[0029] Based on the optimized photovoltaic cell parameter set and the error-compensated parameter prediction model, a multi-model consistency analysis is performed to calculate the prediction error weights of different models and generate a preliminary fusion parameter set;

[0030] Based on the preliminary fusion parameter set, the Bayesian optimization algorithm is used to adaptively adjust the weights of different models, and nonlinear regression optimization is performed to obtain the optimal composite model identification parameters;

[0031] Based on the optimal composite model identification parameters, an adaptive parameter module of the photovoltaic cell is constructed, and the adaptive parameter module is updated online to adapt to changes in photovoltaic system parameters under different working conditions.

[0032] Furthermore, the real-time monitoring data includes ambient temperature, light intensity, and battery aging:

[0033] Perform trend analysis on the real-time monitoring data to obtain real-time status parameters of the photovoltaic system.

[0034] Furthermore, the intelligent operation control of the photovoltaic system includes:

[0035] Based on the final operating parameters of the photovoltaic cells, the power output of the photovoltaic system is controlled, and the operating strategy is dynamically adjusted to achieve intelligent operation control of the photovoltaic cells.

[0036] Furthermore, a photovoltaic cell parameter identification system based on a composite model comprises:

[0037] Data acquisition and preprocessing module, which obtains key operating parameters from various data sources of photovoltaic cells;

[0038] Physical modeling and preliminary parameter calculation module, which models the equivalent circuit parameters of photovoltaic cells based on single-diode model and double-diode model, and calculates preliminary parameter estimates;

[0039] Intelligent optimization and parameter solution module, which performs global optimization on the preliminary parameter estimates and uses genetic algorithm and particle swarm optimization algorithm to find the optimal photovoltaic cell parameter set;

[0040] The error compensation and deep learning prediction module uses a deep neural network model to perform error compensation on optimized photovoltaic cell parameters to improve parameter identification accuracy;

[0041] The composite model fusion and adaptive identification module builds an adaptive photovoltaic cell parameter identification framework based on physical models, optimization algorithms, and deep learning models to improve system adaptability under complex working conditions;

[0042] The real-time monitoring and dynamic adjustment module combines the real-time monitoring data of the photovoltaic system to calculate the optimal operating adjustment parameters of the photovoltaic system and dynamically optimize the system operating status;

[0043] The feedback control and optimization execution module optimizes the control parameters of the photovoltaic system according to the real-time adjustment plan, realizing the intelligent operation control of the photovoltaic cells.

[0044] This invention provides a photovoltaic cell parameter identification method and system based on a composite model. By integrating physical modeling, intelligent optimization, and deep learning technologies, it achieves high-precision, low-computational-cost photovoltaic cell parameter identification and optimizes the intelligent control of photovoltaic systems. Compared with traditional photovoltaic cell parameter identification methods, this invention has the following advantages:

[0045] (1) Improve the accuracy of photovoltaic cell parameter identification. By combining the single-diode model and the double-diode model for preliminary parameter estimation and using the genetic algorithm and particle swarm optimization algorithm to globally optimize the parameters, the limitation of the physical modeling method's dependence on the initial parameters is reduced, and the calculation accuracy of the key parameters of the photovoltaic cell (such as photocurrent, saturation current, series resistance, and parallel resistance) is improved, making the identification error lower.

[0046] (2) Enhanced adaptability to complex working conditions. Traditional physical modeling methods have poor adaptability to complex working conditions such as shadow occlusion, battery aging, and temperature changes. This invention uses a long short-term memory network (LSTM) deep learning model to train historical operating data, construct an error compensation model, and correct parameter identification errors through an adaptive fusion strategy, thereby improving the applicability of the system under different environmental conditions and making photovoltaic cell parameter identification more robust.

[0047] (3) Improve computational efficiency and stability. Traditional optimization methods are computationally intensive and have slow convergence, making them difficult to apply to online monitoring of photovoltaic systems in real time. This invention uses a genetic algorithm for global search and combines it with particle swarm optimization for local fine-tuning, improving computational efficiency while ensuring identification accuracy. Furthermore, the Bayesian optimization algorithm is used to optimize model fusion parameters, further reducing computational resource consumption and enabling fast and efficient parameter identification.

[0048] (4) Realize intelligent dynamic adjustment of the photovoltaic system. This invention combines the composite model identification parameters with the real-time monitoring data of the photovoltaic system to calculate the optimal operating state adjustment parameters of the photovoltaic system, and optimizes the DC-DC converter (DC-DC) control strategy and maximum power point tracking (MPPT) algorithm. This enables the photovoltaic system to dynamically adjust operating parameters according to environmental changes, ensuring the long-term stable operation of photovoltaic cells and improving overall power generation efficiency.

[0049] (5) Enhance the real-time performance of photovoltaic cell parameter identification. By building an adaptive parameter module and combining it with an online update mechanism, the system can continuously optimize the identification parameters during operation to adapt to different operating conditions and improve the response speed of the photovoltaic system to sudden environmental changes.

[0050] The key innovations of the present invention are:

[0051] (1) Composite model parameter identification framework. A method combining a single diode model and a double diode model is used for preliminary parameter estimation, and an intelligent optimization algorithm is used for optimization calculation. Compared with the traditional single modeling method, the accuracy and adaptability of parameter identification are improved.

[0052] (2) Deep learning error compensation strategy. The LSTM (Long Short-Term Memory) deep neural network is used to learn historical photovoltaic cell operation data and construct an error compensation model to reduce the calculation errors of the physical model and optimization algorithm and improve the robustness of the identification parameters.

[0053] (3) Adaptive model fusion based on Bayesian optimization. The Bayesian optimization algorithm is used to fuse the identification parameters of different models, calculate the optimal weight distribution, and improve the stability and computational efficiency of the identification parameters through nonlinear regression optimization.

[0054] (4) Real-time online monitoring and intelligent adjustment. Based on the real-time operating status of the photovoltaic system, the optimal operating adjustment parameters are calculated, and the DC-DC converter and MPPT algorithm are optimized to enable the photovoltaic cell operating parameters to be dynamically adjusted, thereby improving the intelligent control capabilities of the photovoltaic system.

[0055] (5) Multi-objective optimization strategy. An optimization strategy combining genetic algorithm and particle swarm optimization algorithm is adopted to improve the accuracy of photovoltaic cell parameter identification while reducing computational complexity, thus ensuring the high efficiency of the system in large-scale applications.

[0056] By integrating multiple innovative technologies, the present invention improves the accuracy and efficiency of photovoltaic cell parameter identification, and realizes adaptive optimization and intelligent control of photovoltaic systems, which has wide application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A schematic diagram of a flow chart of a photovoltaic cell parameter identification method based on a composite model provided in an embodiment of the present application;

[0058] Figure 2 This is a structural block diagram of a photovoltaic cell parameter identification system based on a composite model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the art to which this application belongs. The terms used in the specification of the application are for the purpose of describing specific embodiments only and are not intended to limit this application. The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0060] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0061] Example 1: Reference Figure 1 , is a flow chart of a photovoltaic cell parameter identification method based on a composite model provided by an embodiment of the present invention. The flow chart may include at least steps S100-S700:

[0062] S100, based on a first computing node in cloud computing, obtaining operating status data of a photovoltaic cell, including current, voltage, temperature, and light intensity data, and performing signal denoising, normalization, and feature extraction on the operating status data to generate preprocessed photovoltaic cell operating status data;

[0063] S200, based on a second computing node in cloud computing, modeling the photovoltaic cell operating state data, calculating preliminary parameter estimates of the photovoltaic cell using a single diode model and a double diode model, and generating an initial parameter sequence of the photovoltaic cell;

[0064] S300, based on a third computing node in the cloud computing, performing optimization calculation on the initial parameter sequence, performing global search and solving the optimal parameters using a genetic algorithm and a particle swarm optimization algorithm, and generating an optimized photovoltaic cell parameter set;

[0065] S400, based on a deep learning computing node in cloud computing, obtaining the optimized photovoltaic cell parameter set, and training a deep neural network model in combination with historical photovoltaic cell operation data to obtain a parameter prediction model for error compensation;

[0066] S500, based on the adaptive computing node in the cloud computing, obtaining the optimized photovoltaic cell parameter set and the parameter prediction model for error compensation, using an adaptive fusion algorithm to generate composite model identification parameters, and constructing an adaptive parameter module for the photovoltaic cell;

[0067] S600, based on the real-time monitoring node in the cloud computing, obtaining the composite model identification parameters, and combining the real-time monitoring data of the photovoltaic system, dynamically calculating the photovoltaic system operation state adjustment parameters, and generating the photovoltaic system adaptive adjustment plan;

[0068] S700 , based on the feedback control node in the cloud computing, obtain the photovoltaic system adaptive adjustment plan, optimize and adjust the control parameters of the photovoltaic system, generate adjusted photovoltaic cell operating parameters, and realize intelligent operation control of the photovoltaic cell.

[0069] Step S100 at least includes steps S110-S130:

[0070] S110 , obtaining operating status data of the photovoltaic cell, including current, voltage, temperature, and light intensity data, and performing time synchronization processing on the operating status data to generate time-aligned original operating data of the photovoltaic cell.

[0071] Specifically, the current I of the photovoltaic cell is collected t , voltage V t , temperature T t and light intensity G t Data, where t represents the time series index, I t is the output current of the photovoltaic cell at time t, V t is the output voltage at the corresponding moment, T t is the ambient temperature recorded by the temperature sensor, G t The light intensity recorded by the light sensor.

[0072] Furthermore, for the operating status data of photovoltaic cells, a time synchronization algorithm is used to align them to ensure the consistency of timestamps of different data sources. Assume that the data sampling time intervals of different sensors are Δt I , Δt V , Δt T and Δt G , then the main time base t ref (Usually select the sensor with the highest sampling rate) to perform data interpolation so that all data are unified to t ref Moment, that is:

[0073] D t ={I t ,V t ,T t ,Gt},t∈T ref

[0074] Among them, D t is the original operating data of the photovoltaic cell after time alignment, T ref Represents a unified time series after time synchronization.

[0075] S120 , performing signal denoising processing on the original operation data of the photovoltaic cell, using a wavelet transform filtering method to eliminate high-frequency noise, and performing Kalman filtering optimization to obtain the photovoltaic cell time series data after noise reduction.

[0076] Specifically, from the time-aligned photovoltaic cell raw operating data D t Extract the current I t , voltage V t , temperature T t and light intensity G t Signal, the wavelet transform method is used to denoise each signal. Define the wavelet transform decomposition signal S t for:

[0077] S t =A t +D t

[0078] Among them, A t is the low-frequency component of the signal (i.e. the main trend part), D t is the high-frequency noise component, D′ t is the noise signal after threshold processing. For the signal after wavelet decomposition, set the noise threshold θ for denoising:

[0079]

[0080] Furthermore, the Kalman filter method is used to smooth and optimize the denoised current, voltage, temperature and light data. The observation state vector is defined as:

[0081] X t =[I t ,V t ,T t ,G t ] T

[0082] Among them, X t represents the true state vector at time t, [·] TThis is a column vector that represents the operating status of the photovoltaic system at time t. It is composed of a combination of key parameters collected by real-time monitoring nodes. Its functions are as follows: serving as an input feature vector for the neural network, Kalman filter, and modeling algorithm; ensuring the consistency of the data structure for subsequent processing; and serving as the information starting point for the entire identification system, used to support the entire parameter estimation process.

[0083] The state transfer equation is:

[0084] X t+1 =FX t +W t

[0085] Among them, X t+1 is the true state vector at time t+1, F is the state transfer matrix, W t is the process noise. The observation equation is defined as follows:

[0086] Y t =HX t +V t

[0087] Among them, Y t is the observation value, H is the measurement matrix, V t is the measurement noise. The estimated state is updated iteratively through Kalman filtering, and finally the smoothed and noise-reduced photovoltaic cell time series data is obtained:

[0088]

[0089] in, is the photovoltaic cell time series data after noise reduction, They are the denoised current, voltage, temperature and light data respectively.

[0090] S130 , normalizing the noise-reduced photovoltaic cell time series data, and extracting key characteristic parameters of the photovoltaic cell, including maximum power point voltage, short-circuit current, and temperature compensation coefficient, to generate pre-processed photovoltaic cell operating status data.

[0091] Specifically, the photovoltaic cell time series data after noise reduction is Normalization is performed to reduce the impact of numerical scale differences between different physical quantities on the calculation. The Min-Max normalization method is used for normalization:

[0092]

[0093] in, represents the denoised data, X min and X max are the minimum and maximum values ​​of the data, respectively. The data are normalized.

[0094] Furthermore, based on the normalized data, key characteristic parameters of photovoltaic cells are extracted, including:

[0095] Maximum power point voltage V mp : Obtained by calculating the power extreme point of the photovoltaic cell IV curve:

[0096]

[0097] Short-circuit current I sc : When the voltage is zero, extract the maximum short-circuit current of the photovoltaic cell:

[0098]

[0099] Temperature compensation coefficient α T : Calculate the temperature coefficient of the open circuit voltage of the photovoltaic cell according to the temperature change:

[0100]

[0101] Among them, V oc Indicates the open circuit voltage of the photovoltaic cell, ΔV oc It represents the change of open circuit voltage at different temperatures, and ΔT is the temperature change.

[0102] Finally, the pre-processed photovoltaic cell operating status data is generated:

[0103]

[0104] Connection Instructions

[0105] D calculated by S110 t As the input of S120, denoising is performed to reduce the impact of measurement errors on subsequent calculations.

[0106] S120 calculated As the input of S130, normalization processing is performed and key characteristic parameters of the photovoltaic cell are extracted.

[0107] Photovoltaic cell operating status data calculated by S130 It will be used as input to S200 (physical modeling module) to build single diode and double diode models.

[0108] The implementation steps in this section synchronized, denoised, normalized, and extracted features from the photovoltaic cell's current, voltage, temperature, and illumination data, generating high-quality input data. This data not only provided accurate basic parameters for subsequent physical modeling but also enhanced the accuracy and stability of the entire photovoltaic cell parameter identification system.

[0109] Step S200 at least includes steps S210-S230:

[0110] S210 , based on the photovoltaic cell operating status data, using a single diode model to calculate equivalent circuit parameters of the photovoltaic cell, including photocurrent, saturation current, resistance parameters, and series resistance, to obtain a single diode model parameter sequence.

[0111] Specifically, based on the photovoltaic cell operating status data generated in S130 The single diode model is used to calculate the equivalent circuit parameters of the photovoltaic cell.

[0112] Output current I of the single diode model t,sd It is given by the following formula:

[0113]

[0114] Among them: I ph is the photocurrent, reflecting the contribution of light intensity to the current; I s is the saturation current, which is related to temperature and material properties; q is the electron charge constant; V t is the terminal voltage of the photovoltaic cell; R s is the series resistance; R sh is the parallel resistance, reflecting the influence of leakage current; n is the diode ideal factor, indicating the nonlinearity of the relationship between current and voltage; k is the Boltzmann constant; T t is the operating temperature of the photovoltaic cell, e represents the base of the natural logarithm, which is a mathematical constant with a value of approximately 2.71828.

[0115] I t,sd This formula is the core output current expression of the single diode equivalent circuit model, which is used to calculate the output current I of the photovoltaic cell at time t. t,sd It takes into account: the current I generated by light ph , the reverse leakage characteristics of the diode (saturation current I s and ideality factor n), the series internal resistance of the battery R s and parallel resistor R sh , temperature T t The thermal sensitivity of the output. This model can be used to fit the actual IV curve, which is an important basis for subsequent parameter identification, model fitting and error calculation, and is the core expression in physical modeling. Based on the formula, the Newton iteration method is used to ph , I s 、R s and R sh Solve and obtain the single diode model parameter sequence:

[0116] D sd ={I ph ,I s ,R s ,R sh ,n}

[0117] Among them, D sd It is a parameter set and a core identification object in the composite model structure of the present invention. It directly affects the accuracy of photovoltaic cell modeling, the accuracy of power prediction, and the effect of subsequent adaptive control strategies. It serves as an important input parameter for the subsequent S230 calculation model error matching.

[0118] S220 , based on the photovoltaic cell operating status data, using a dual-diode model to calculate the current shunt parameters of the photovoltaic cell, and combining the shadow shielding effect to calculate the nonlinear correction coefficient of the photovoltaic cell, to obtain a dual-diode model parameter sequence.

[0119] Specifically, for the photovoltaic cell operating status data The double diode model is further used to model the output current I t,dd It is expressed by the following formula:

[0120]

[0121] Among them: I t,dd is the output current of the dual-diode model at time t; I ph is the photocurrent; q is the elementary charge constant; V t is the output voltage; R s is the series resistance; T t is the photovoltaic cell temperature; R sh is the parallel resistance; k is the Boltzmann constant; I s1 and I s2 They represent the saturation currents of the two diodes respectively; n1 and n2 are the ideal factors of the two diodes, n1 corresponds to the diffusion process, and n2 corresponds to the recombination process.

[0122] Furthermore, in order to consider the nonlinear characteristics of photovoltaic cells, the nonlinear correction coefficient β of photovoltaic cells is calculated in combination with the shadow shielding effect. s , whose expression is:

[0123]

[0124] Among them: I mp is the current at the maximum power point; I sc is the short-circuit current; V oc is the open circuit voltage; V mp is the voltage at the maximum power point.

[0125] Estimate I using the least squares method s1 , I s2 , n1, n2 and R sh The optimal value of , the double diode model parameter sequence is obtained:

[0126] D dd ={I ph ,I s1 ,I s2 ,R s ,R sh ,n1,n2,β s}

[0127] Among them, D dd A parameter set defined in the dual-diode model that describes the modeling behavior of photovoltaic cells under more complex nonlinear characteristics. It serves as an important input parameter for the subsequent S230 calculation model error matching.

[0128] S230 , performing feature matching calculation on the single-diode model parameter sequence and the dual-diode model parameter sequence, selecting the model with the smallest error as a preliminary parameter estimation model, and generating an initial parameter sequence of the photovoltaic cell.

[0129] Specifically, based on the single diode model parameter sequence D calculated in S210 sd The double diode model parameter sequence D calculated by S220 dd , and the optimal model is selected using error matching calculation.

[0130] The error measurement function of the photovoltaic cell is defined as:

[0131]

[0132] Where: E m is the error measurement function; I t,exp is the current value measured in the experiment; I t,model is the predicted current value calculated by the single-diode model or the double-diode model; N is the total number of measured data points.

[0133] By calculating E m , if the error E of the single diode model sd Less than the error E of the double diode model dd , the parameters of the single diode model are selected as the preliminary parameter estimation model; otherwise, the parameters of the double diode model are selected:

[0134]

[0135] Finally, the initial parameter sequence D of the photovoltaic cell is generated init, as the input of the S300 module (intelligent optimization solution).

[0136] Connection Instructions

[0137] The single diode model parameter D calculated by S210 sd The double diode model parameter D calculated by S220 dd As input to the S230 error matching calculation;

[0138] S230 calculates the initial parameter sequence D of the photovoltaic cell init As the input of S300 (intelligent optimization solution module), it is used to further optimize the parameter estimation results and improve the identification accuracy.

[0139] This section describes the photovoltaic cell modeling process in detail, covering the calculation of single diode model and double diode model, current shunt parameter estimation and model error matching. The final initial parameter sequence D init As the input for subsequent optimization calculations, it provides a basis for photovoltaic cell parameter identification.

[0140] Step S300 at least includes steps S310-S330:

[0141] S310 , based on the initial parameter sequence, a genetic algorithm is used to perform a global search, multiple parameter populations are initialized, and genetic variation and selection operations are performed according to a target fitness function to obtain an optimized population parameter sequence.

[0142] Specifically, based on the photovoltaic cell initial parameter sequence D calculated in S230 init As the input of the optimization solution, a global search is performed using a genetic algorithm to initialize multiple parameter populations. The population of photovoltaic cell parameters is set as:

[0143]

[0144] Where: P (0) represents the initial population; is the parameter vector of the i-th individual, i.e., the key circuit parameter combination of the photovoltaic cell:

[0145]

[0146] N is the number of individuals in the population.

[0147] Based on the initial population, the fitness of each individual is calculated using the target fitness function. Define the fitness function f(X i ) Calculate the model error:

[0148]

[0149] Where: I t,exp is the output current of the photovoltaic cell measured in the experiment; I t,model (X i ) is based on individual X i The calculated current value of the photovoltaic cell model; M is the number of experimental data points.

[0150] Furthermore, performing genetic variation and selection operations on the initial population includes:

[0151] Selection operation: Use the roulette wheel method to select individuals with higher fitness to form the next generation crossover population;

[0152] Crossover operation: Perform uniform crossover on the selected individuals to generate new candidate solutions;

[0153] Mutation operation: Randomly perturb the parameters of some individuals to increase population diversity.

[0154] Iterate the genetic operation until the fitness converges or the maximum number of iterations G is reached. max , and finally the optimized population parameter sequence is obtained:

[0155]

[0156] in, As input for further optimization calculations in S320.

[0157] S320 , based on the optimized population parameter sequence, using a particle swarm optimization algorithm to perform local fine-tuning, searching for optimal parameter values ​​based on a speed-position update mechanism, and iteratively optimizing key circuit parameters of the photovoltaic cell to obtain an optimized photovoltaic cell candidate parameter set.

[0158] Specifically, based on the optimized population parameter sequence calculated in S310 The particle swarm optimization algorithm is used for local fine-tuning to improve the accuracy of parameter identification. The particle swarm optimization parameter vector is defined as:

[0159]

[0160] The particle velocity update equation is:

[0161]

[0162] in: is the velocity of particle i in generation t+1; ω is the inertia weight, which controls the balance of particle movement; c1 and c2 are learning factors, which control the influence of individual optimal value and global optimal value respectively; r1 and r2 are random factors, which enhance the search ability; P best,i is the best historical position of particle i; G bestis the global optimal position.

[0163] The particle position update equation is:

[0164]

[0165] in, is the new position of particle i at time t+1, is the position of particle i at time t, is the velocity of particle i at time t+1. By iteratively calculating the velocity and position update equations, the global optimal parameter combination is gradually approached, ultimately obtaining the optimized set of candidate photovoltaic cell parameters. This formula is the core position update formula in the particle swarm optimization (PSO) algorithm. It represents the position of the current i-th solution vector (i.e., a parameter set) at time t+1. It is obtained by superimposing its current position and velocity. It reflects the algorithm's iterative search for optimal parameters in the solution space. In parameter identification, it is used to continuously approach the true parameters from the initial value.

[0166]

[0167] Among them, P opt is the optimized parameter candidate set, is the optimal parameter vector of the i-th particle / individual, and N is the number of particles.

[0168] S330 , performing adaptability evaluation on the optimized photovoltaic cell candidate parameter set, selecting an optimal solution based on a multi-objective optimization criterion, and generating a final optimized photovoltaic cell parameter set.

[0169] Specifically, based on the optimized photovoltaic cell candidate parameter set P calculated in S320 opt , evaluate the adaptability of each candidate solution, and select the optimal solution using a multi-objective optimization criterion. Define the objective optimization function F(X) as the final optimization criterion:

[0170] F(X)=w1f1(X)+w2f2(X)+w3f3(X)

[0171] Where: w1, w2, and w3 are optimization weights; f1(X) is the parameter identification error, which calculates the error between the optimized parameters and experimental data; f2(X) is parameter stability, which measures the volatility of the optimized parameters under different test conditions; and f3(X) is the computational convergence, which evaluates the convergence rate of the optimization algorithm. This function is used to measure the comprehensive quality of a candidate parameter solution X, including: f1: fitting error (accuracy); f2: parameter stability (robustness); and f3: optimization convergence (efficiency). Through weighted combination, multi-objective evaluation is transformed into a single-objective optimization, which is a key criterion for finding the optimal battery modeling parameters.

[0172]

[0173] in, is the current value calculated by substituting the current parameter X into the model, is the measured current data, T is the length of the data time series;

[0174]

[0175] Where N is the number of working conditions (such as different environmental scenarios), σ i (X) is the standard deviation of parameter X in the i-th scenario;

[0176]

[0177] Among them, F k (X) is the fitness function value at the kth iteration, and K is the total number of iterations.

[0178] Furthermore, we select the parameter vector X with the smallest F(X) * As the final optimized photovoltaic cell parameter set:

[0179]

[0180] The final generated optimization parameter set X * The optimal equivalent circuit parameters of the photovoltaic cell include:

[0181]

[0182] Among them, D final The optimized photovoltaic cell parameter set is used as the input for S400 (deep learning error compensation) calculation to further improve the accuracy of photovoltaic cell parameter identification.

[0183] Connection Instructions

[0184] The optimized population parameter sequence calculated by S310 As input for local fine-tuning of S320 using particle swarm optimization;

[0185] The photovoltaic cell candidate parameter set P calculated by S320 opt As input to the S330 suitability evaluation;

[0186] The final optimized parameter set D calculated by S330 final As the input of S400 deep learning error compensation, it improves the accuracy of parameter prediction.

[0187] This section describes in detail the photovoltaic cell parameter optimization process based on genetic algorithm and particle swarm optimization algorithm, including global search, local fine-tuning and multi-objective optimization, and finally obtains the optimized photovoltaic cell parameter set D final , providing input for subsequent deep learning corrections.

[0188] Step S400 at least includes steps S410-S430:

[0189] S410 , based on the optimized photovoltaic cell parameter set, obtaining historical photovoltaic cell operation data, and performing time window division and feature label matching on the data to generate a training data set.

[0190] Specifically, based on the optimized photovoltaic cell parameter set calculated in S330 Obtain historical operating data of photovoltaic cells to enhance the generalization ability of parameter prediction models. Set the historical data set to:

[0191] H t ={I t1 ,V t1 ,T t1 ,G t1 ,P t}

[0192] Among them: H t For historical photovoltaic cell operation dataset; I t1 is the output current of the photovoltaic cell; V t1 is the output voltage of the photovoltaic cell; T t1 is the ambient temperature; G t1 is the light intensity; P t =I t1 V t1 is the output power of the photovoltaic cell.

[0193] Furthermore, based on the historical data, the operation data of the photovoltaic cells is divided into time windows, the time window size is set to W, and the time series input data is constructed:

[0194] X t2 ={H t-W+1 ,H t-W+2 ,...,H t}

[0195] Among them, X t2 Represents the input feature vector at time t, which contains the historical running data of the latest W time steps.

[0196] In addition, feature label matching is performed on the historical data to define the target output variable Y t As error compensation target:

[0197]

[0198] in: is the parameter estimate calculated based on the theoretical model of photovoltaic cells; D final is the optimized photovoltaic cell parameter set; Y t As the error compensation value, it represents the deviation between the theoretical calculation parameters and the optimized parameters. It represents the construction of deep learning target labels in error compensation model training: Parameters estimated by the physical model; D final : The actual or expected parameter obtained by the optimization algorithm; Y t : is their difference, which is used as the “regression label” in supervised learning; through this formula, the LSTM network can learn the “distribution law of model estimation deviation” for subsequent error correction.

[0199] Finally, a training dataset is generated for training the deep neural network model:

[0200] D train ={X t2 ,Y t},t∈T

[0201] Among them, D train is the training data set, X t2 represents the input feature vector at time t, Y t As the error compensation value, it represents the deviation between the theoretical calculation parameters and the optimized parameters, and T is the total length of the time series data.

[0202] S420. Based on the training data set, construct an LSTM deep neural network model, learn the time series characteristics of photovoltaic cell parameters, and perform hyperparameter optimization to obtain a pre-trained error compensation model.

[0203] Specifically, based on the training data set D generated in S410 train , a long short-term memory (LSTM) deep neural network model is constructed to learn the time series characteristics of photovoltaic cell parameters and perform error compensation.

[0204] Set the input of the LSTM model to:

[0205]

[0206] The hidden state update equation of the LSTM network is:

[0207]

[0208] Where: h tis the hidden state of the LSTM network; W h and W x are the weight matrices of the hidden layer and the input layer respectively; b h is the bias term; f(·) is the activation function.

[0209] The output layer of the LSTM network calculates the error compensation value

[0210]

[0211] Where: W o is the output layer weight matrix; b o is the output layer bias term.

[0212] Furthermore, to optimize the hyperparameters of the LSTM model, a Bayesian optimization algorithm is used to search for the optimal hyperparameter combination, including:

[0213] Learning rate η: used for gradient descent to update weights;

[0214] Number of hidden layer units N h : Determines the modeling capability of the LSTM network;

[0215] Batch size B: affects the stability of model training.

[0216] Define the loss function Calculate the mean square error (MSE) of the error compensation value:

[0217]

[0218] Among them, N2 is the number of training data samples.

[0219] By minimizing the loss function through the gradient descent method and adjusting the weight parameters of the LSTM network, we can finally obtain the pre-trained error compensation model:

[0220]

[0221] in, As input to the parameter prediction model after calculating the error correction in S430.

[0222] S430: Input the optimized photovoltaic cell parameter set into the error compensation model, calculate the compensation value based on the historical error distribution of the photovoltaic cell, and generate an error-corrected parameter prediction model.

[0223] Specifically, based on the LSTM error compensation model obtained by the S420 training The optimized photovoltaic cell parameter set D calculated by S330 final As input, calculate the error correction value

[0224]

[0225] in: The error correction value output by the LSTM error compensation model.

[0226] Furthermore, the final corrected parameter prediction value is calculated based on the historical error distribution of the photovoltaic cell

[0227]

[0228] in, Represents the parameter prediction value after error correction, including:

[0229]

[0230] in, is the error-corrected photocurrent, is the corrected saturation current of the first diode, is the corrected saturation current of the second diode, is the corrected series resistance, is the corrected parallel resistance, is the first ideal factor after correction, is the modified second ideal factor, is the corrected scaling compensation coefficient.

[0231] Finally, the error-corrected parameter prediction model M is generated pred :

[0232]

[0233] Among them, M pred As the input of S500 (composite model fusion) to improve the accuracy of photovoltaic cell parameter identification.

[0234] Connection Instructions

[0235] The training data set D calculated by S410 train As input to the S420 training LSTM error compensation model;

[0236] LSTM model trained by S420 as input for calculating the error correction value in S430;

[0237] Among them, the error compensation model Output the correction parameters in S430 This result is used as the composite model fusion input in S500, rather than the entire model structure M pred .

[0238] This section describes in detail the process of photovoltaic cell parameter error compensation based on the LSTM deep learning model, covering data acquisition, training data construction, model training and error correction, and finally obtains the error-corrected parameter prediction model M. pred , providing input for subsequent composite model fusion and improving the accuracy of photovoltaic cell parameter identification.

[0239] Step S500 at least includes steps S510-S530:

[0240] S510 , performing multi-model consistency analysis based on the optimized photovoltaic cell parameter set and the error-compensated parameter prediction model, calculating prediction error weights of different models, and generating a preliminary fusion parameter set.

[0241] Specifically, a multi-model consistency analysis is performed on the output parameters of different models to determine the relative weight of the prediction error of each model.

[0242] Set the multi-model prediction output to:

[0243]

[0244] in: is the overall prediction parameter set, D t,sd The photovoltaic cell parameters output by the single diode model; D t,dd The photovoltaic cell parameters output by the double diode model; D t,lstm Photovoltaic cell parameters predicted by the LSTM error compensation model.

[0245] Furthermore, the prediction error E of each model is calculated m , define the error calculation formula:

[0246]

[0247] in: are the actual photovoltaic cell parameters measured experimentally; are the photovoltaic cell parameters predicted by different models m; N is the number of experimental data points.

[0248] Based on the error calculation results, the prediction error weight w of each model is normalized and calculated m :

[0249]

[0250] Among them, w mRepresents the weight of model m, and M3 is the total number of models. Finally, calculate the initial fusion parameter set:

[0251]

[0252] in, is the initial fusion parameter set, w m is the weight, The predicted value of the mth model is used as the input for calculating the optimal composite model identification parameters in S520. This is the calculation formula for the multi-model fusion output, which means that at time t, the output results of each sub-model are The fusion parameters generated by weighted combination.

[0253] S520: Based on the preliminary fusion parameter set, the Bayesian optimization algorithm is used to adaptively adjust the weights of different models, and nonlinear regression optimization is performed to obtain the optimal composite model identification parameters.

[0254] Specifically, based on the preliminary fusion parameter set calculated in S510 The Bayesian optimization algorithm is used to adaptively adjust the weights of each model to maximize the recognition accuracy. Define the target optimization function F(w):

[0255]

[0256] Where: F(w) represents the optimization objective function, which measures the model identification error; w m Represents the weight of model m.

[0257] Bayesian optimization is used to search for the optimal weight combination Minimize F(w). Define the optimal fusion parameters:

[0258]

[0259] Furthermore, the optimal fusion parameters are subjected to nonlinear regression optimization, and a nonlinear regression model is set:

[0260]

[0261] Where: f(·) is the regression function; θ is the parameter to be optimized in the regression model.

[0262] Gradient descent is used to optimize θ, and the optimal composite model identification parameters are finally obtained:

[0263]

[0264] Among them, θ * is the adaptive fusion coefficient, As the input of S530 to build the photovoltaic cell adaptive parameter module. The dynamic weight (θ * ) is combined with the fusion function f(·) to output the parameter set that is ultimately used for online monitoring and regulation, which is the “final landing value” of the entire identification process.

[0265] S530 : Based on the optimal composite model identification parameters, construct an adaptive parameter module for the photovoltaic cell, and update the model online to adapt to changes in photovoltaic system parameters under different working conditions.

[0266] Specifically, based on the optimal composite model identification parameters calculated in S520 Construct an adaptive parameter module for photovoltaic cells to adapt to the operating state of the photovoltaic system under different working conditions. Define the adaptive parameter update equation:

[0267]

[0268] in: is the photovoltaic cell identification parameter after adaptive update; α is the dynamic adjustment coefficient, which controls the update speed; is the identification parameter of the previous moment.

[0269] Furthermore, in order to enhance the real-time adaptability of the model, a sliding window strategy is used to dynamically adjust the parameters and set the sliding window size W2:

[0270]

[0271] in, As the final photovoltaic cell parameter identification result, it can be used to calculate the operation adjustment strategy of the photovoltaic system in S600 (real-time online monitoring).

[0272] Connection Instructions

[0273] The preliminary fusion parameter set calculated by S510 As input for Bayesian optimization tuning of S520;

[0274] S520 calculates the optimal composite model identification parameters As input to the S530 adaptive identification model;

[0275] S530 Calculate the adaptive update parameters As the input of S600 (real-time online monitoring), it improves the intelligent regulation capability of the photovoltaic system.

[0276] This section describes in detail the process of photovoltaic cell parameter identification and optimization based on the adaptive fusion algorithm, including multi-model consistency analysis, Bayesian optimization weight adjustment and adaptive online update, and finally generates the optimal photovoltaic cell parameter identification model. It can dynamically adapt to different operating conditions and improve the identification accuracy and stability of the photovoltaic system.

[0277] Step S600 at least includes steps S610-S630:

[0278] S610 , based on the composite model identification parameters, obtain real-time monitoring data of the photovoltaic system, including ambient temperature, light intensity, and battery aging, and perform trend analysis to obtain real-time status parameters of the photovoltaic system.

[0279] Specifically, based on the adaptive parameter module parameters calculated in S530 Get real-time monitoring data of the photovoltaic system to update the operating status of the photovoltaic system. Set the real-time monitoring data set as:

[0280] M t ={T t3 ,G t3 ,A t3 ,I t3 ,V t3 ,P t3}

[0281] Where: M t is a real-time monitoring dataset; T t3 is the ambient temperature; G t3 is the light intensity; A t3 is the aging coefficient of the photovoltaic cell, indicating the degree of cell aging; I t3 is the output current of the photovoltaic cell; V t3 is the output voltage of the photovoltaic cell; P t3 =I t3 V t3 is the real-time power output of the photovoltaic cell.

[0282] Furthermore, the monitoring data is trend analyzed to predict the short-term state change of the photovoltaic system. Set the trend change rate:

[0283]

[0284] Where: R m is the variable M t The rate of change of M t Represents the monitoring variable of the photovoltaic system at time t; M t-1 represents the monitoring variable of the photovoltaic system at time t-1; Δt is the sampling time interval.

[0285] By calculating R x , determine the real-time status parameters of the photovoltaic system:

[0286] S t ={T t3 ,G t3 ,A t3 ,R T3 ,R G3 ,R A}

[0287] Among them, S t is the real-time status parameter of the photovoltaic system, based on real-time monitoring data and parameter identification model; T t3 is the real-time ambient or component temperature; G t3 is the real-time light intensity (irradiance); A t3 is the battery aging indicator; R T3 is the temperature change rate, Indicates the temperature fluctuation trend per unit time; R G3 is the rate of change of illumination, Used to capture dynamic fluctuations in irradiance; R A is the aging change trend rate, Used to estimate the system degradation rate and assist in updating dynamic control strategies; set the power optimization objective function and solve the optimal adjustment parameters

[0288] S620 : Based on the real-time state parameters of the photovoltaic system and in combination with an adaptive parameter module, calculate the optimal operating state adjustment parameters of the photovoltaic system and generate a dynamic adjustment sequence for the photovoltaic system.

[0289] Specifically, the optimal operating state adjustment parameters of the photovoltaic system are calculated.

[0290] Set the power optimization objective function of the photovoltaic system:

[0291]

[0292] Among them: F(U t ) is the power optimization objective function, which represents the power difference before and after system adjustment; is the maximum theoretical power of the photovoltaic system under current environmental conditions; P t (U t ) is the given adjustment parameter U t The output power of the photovoltaic system at time . The gradient descent method is used to optimize the adjustment parameter U t , set the update equation:

[0293]

[0294] in: is the optimized adjustment parameter; η is the learning rate, which controls the adjustment step size; Calculate the value for the gradient.

[0295] Finally, the dynamic adjustment sequence of the photovoltaic system is generated:

[0296]

[0297] in, Serves as input for S630 to calculate the photovoltaic system adaptive adjustment plan.

[0298] S630: Based on the photovoltaic system dynamic adjustment sequence, optimize the voltage and current regulation strategies of the photovoltaic system, and generate a photovoltaic system adaptive adjustment plan.

[0299] Specifically, the photovoltaic system dynamic adjustment sequence calculated based on S620 Optimize the voltage and current regulation strategy of the photovoltaic system to generate the final adaptive adjustment plan.

[0300] Defines the tuning rules for the Maximum Power Point Tracking (MPPT) algorithm:

[0301]

[0302] in: is the maximum power point voltage at time t+1; is the maximum power point voltage at time t; Adjust parameters for optimization; is the baseline adjustment parameter; λ is the adjustment factor.

[0303] Furthermore, the adjusted photovoltaic cell output power is calculated:

[0304]

[0305] in, is the adjusted photovoltaic cell power.

[0306] Finally, the photovoltaic system adaptive adjustment plan is generated:

[0307]

[0308] in, As the final photovoltaic system adjustment plan, it can be used in S700 (feedback control and optimization execution) to calculate the optimal control strategy of the photovoltaic system.

[0309] Connection Instructions

[0310] The real-time status parameter S of the photovoltaic system calculated by S610 tAs input for calculating the optimal operating state adjustment parameters in S620;

[0311] S620 Calculated dynamic adjustment sequence As input for S630 to calculate the adaptive adjustment scheme;

[0312] Adaptive adjustment solution calculated by S630 As the input of S700 (feedback control and optimization execution), it improves the dynamic regulation capability of the photovoltaic system.

[0313] This section describes in detail the process of optimizing photovoltaic system parameters based on real-time monitoring data and adaptive identification models, including state monitoring, optimal adjustment calculation and adaptive optimization, and finally generates an adaptive adjustment plan for the photovoltaic system. Improve the intelligent regulation capability of photovoltaic systems.

[0314] Step S700 at least includes steps S710-S730:

[0315] S710 : Based on the photovoltaic system adaptive adjustment scheme, optimize the DC-DC converter control strategy of the photovoltaic system, calculate the power tracking adjustment parameters, and obtain the dynamic power adjustment parameters of the photovoltaic cells.

[0316] Specifically, the photovoltaic system adaptive adjustment scheme calculated based on S630 The control strategy of the DC-DC converter of the photovoltaic system is optimized. The input and output relationship of the DC-DC converter is set as:

[0317]

[0318] in: is the output voltage of the DC-DC converter; is the input voltage of the DC-DC converter; D t is the duty cycle, which represents the control parameter of the converter.

[0319] Furthermore, the parameters are adjusted based on the maximum power point and Calculate the optimized DC-DC converter duty cycle:

[0320]

[0321] in, As input for calculating maximum power point tracking (MPPT) optimization parameters at S720 .

[0322] Finally, the dynamic power regulation parameters of the photovoltaic cells are generated:

[0323]

[0324] in, As input for S720 to calculate the photovoltaic system power output strategy.

[0325] S720: Based on the dynamic power adjustment parameters of the photovoltaic cell, adjust the maximum power point tracking (MPPT) algorithm, optimize the power output strategy of the photovoltaic system, and generate the final operating parameters of the photovoltaic cell.

[0326] Specifically, the photovoltaic cell dynamic power adjustment parameter calculated in S710 is The MPPT algorithm of the photovoltaic system is optimized to improve the power output efficiency. The update equation of the MPPT algorithm is defined as:

[0327]

[0328] in: is the maximum power point voltage at the next moment; λ is the adjustment step size; is the rate of change of power with respect to voltage.

[0329] Furthermore, the final operating parameters of the optimized photovoltaic cells are calculated:

[0330]

[0331] in: is the final adjusted photovoltaic cell current; The final power output.

[0332] Finally, the final operating parameters of the photovoltaic cell are generated Serves as input for S730 to calculate the dynamic operation control strategy of the photovoltaic system.

[0333] S730 . Based on the final operating parameters of the photovoltaic cell, control the power output of the photovoltaic system and dynamically adjust the operating strategy to achieve intelligent operation control of the photovoltaic cell.

[0334] Specifically, the final operating parameters of the photovoltaic cell calculated based on S720 Control the power output of the photovoltaic system and adjust the operation strategy. Set the output power control equation of the photovoltaic system:

[0335]

[0336] in: is the final output power of the photovoltaic system; η conv is the conversion efficiency of the system.

[0337] Furthermore, a dynamic adjustment strategy is adopted to adaptively adjust the MPPT algorithm step size λ according to environmental changes:

[0338]

[0339] in: is the adjustment step size after dynamic optimization; λ min ,λ max are the minimum and maximum adjustment steps respectively; R G is the rate of change of light intensity; γ is the adjustment factor.

[0340] Ultimately, intelligent operation control of photovoltaic cells is achieved, enabling them to dynamically adapt to different operating conditions and improve power output stability.

[0341] Example 2: Figure 2 FIG. 1 shows a structural block diagram of a photovoltaic cell parameter identification system based on a composite model according to an embodiment of the present invention. Figure 2 As shown, the structure may include:

[0342] The data acquisition and preprocessing module 10 is used to obtain key operating parameters from multiple sources of photovoltaic cells, including current, voltage, ambient temperature, light intensity, and cell aging information. This module collects data in real time using high-precision sensors and employs data synchronization technology to ensure the consistency of timestamps across different data sources. Furthermore, this module performs signal denoising on the raw data, including wavelet transform denoising and Kalman filter optimization, to reduce environmental noise interference. Normalization is then used to standardize the data, ensuring consistency and usability, and providing high-quality input data for subsequent model calculations.

[0343] The physical modeling and preliminary parameter calculation module 20 is used to model the equivalent circuit parameters of the photovoltaic cell based on the single diode model and the double diode model, and calculate the preliminary parameter estimates. This module uses the physical modeling method to calculate key cell parameters such as photocurrent, saturation current, series resistance, parallel resistance and temperature compensation factor. The single diode model provides basic circuit parameter calculation results by solving the current-voltage (IV) relationship of the photovoltaic cell. The double diode model further considers the carrier recombination effect to improve the accuracy of parameter identification. Finally, the module performs matching analysis on the calculated parameter sequence, generates preliminary parameter estimates of the photovoltaic cell, and provides basic data for optimization solution.

[0344] The intelligent optimization and parameter solving module 30 is used to perform global optimization on the preliminary parameter estimation values, and adopts genetic algorithm and particle swarm optimization algorithm to find the optimal photovoltaic cell parameter set.

[0345] This module first uses a genetic algorithm to initialize the parameter population and optimizes the population parameters through crossover, mutation, and selection operations to improve parameter identification accuracy. It then uses a particle swarm optimization algorithm to locally fine-tune the parameters optimized by the genetic algorithm and further optimizes the parameter solution through a speed-position iterative update mechanism.

[0346] Finally, the module evaluates the adaptability of the optimized parameter candidate set based on the multi-objective optimization criterion, selects the parameter set with the smallest error, and forms the final optimized parameter value of the photovoltaic cell.

[0347] The error compensation and deep learning prediction module 40 is used to perform error compensation on the optimized photovoltaic cell parameters using a deep neural network model to improve the parameter identification accuracy.

[0348] This module builds a long short-term memory (LSTM) model based on historical photovoltaic cell operating data and learns the time series characteristics of cell parameters. Through training, the model can predict the error distribution of photovoltaic cell parameters under different environmental conditions and generate error compensation values.

[0349] The module then inputs the optimized photovoltaic cell parameters into the LSTM model, calculates the compensation value based on the historical error distribution, and finally outputs the error-corrected parameter prediction results to ensure the long-term stability and high precision of the parameter identification system.

[0350] The composite model fusion and adaptive identification module 50 is used to construct an adaptive photovoltaic cell parameter identification framework based on physical models, optimization algorithms and deep learning models to improve the system adaptability under complex working conditions.

[0351] This module first performs a multi-model consistency analysis, calculates the prediction error weights of different models, and generates a preliminary fusion parameter set. It then uses a Bayesian optimization algorithm to adaptively adjust the model weights and performs nonlinear regression optimization to calculate the optimal composite model identification parameters.

[0352] Finally, the module constructs an adaptive parameter module for photovoltaic cells based on the optimal fusion parameters and supports online updates to adapt to changes in photovoltaic system parameters under different environments and working conditions, thereby improving the robustness and generalization ability of photovoltaic cell parameter identification.

[0353] The real-time monitoring and dynamic adjustment module 60 is used to combine the real-time monitoring data of the photovoltaic system, calculate the optimal operation adjustment parameters of the photovoltaic system, and dynamically optimize the system operation status.

[0354] This module first obtains ambient temperature, light intensity, and battery aging information from the PV system's monitoring equipment and performs trend analysis to generate real-time status parameters for the PV system. Then, combined with the adaptive parameter module, it calculates the optimal operating adjustment parameters for the PV system and generates a dynamic adjustment sequence to optimize the voltage and current regulation strategy for the PV cells.

[0355] Finally, the module optimizes the operating parameters of the PV system according to the adjustment strategy and generates an adaptive adjustment plan for the PV system to improve the overall power generation efficiency and stability.

[0356] The feedback control and optimization execution module 70 is used to optimize the control parameters of the photovoltaic system according to the real-time adjustment plan to achieve intelligent operation control of the photovoltaic cells.

[0357] This module first optimizes the DC-DC converter control strategy based on the PV system's adaptive adjustment scheme and calculates the power tracking adjustment parameters. It then optimizes the maximum power point tracking (MPPT) algorithm, incorporating dynamic power regulation parameters, and calculates the PV system's final operating parameters.

[0358] Ultimately, the module dynamically adjusts the power output of the photovoltaic system based on the optimized operating parameters and optimizes the operation control strategy to achieve intelligent operation and management of the photovoltaic system and improve the overall efficiency of photovoltaic power generation.

[0359] The present invention provides a photovoltaic cell parameter identification method and system based on a composite model, which has the following beneficial effects:

[0360] High-precision parameter identification: This invention combines physical models, intelligent optimization algorithms and deep learning technology to accurately identify photovoltaic cell parameters under different environmental conditions and improve identification accuracy.

[0361] Strong adaptive optimization capability: Adopting adaptive fusion model, through Bayesian optimization and error compensation, the system's adaptability to complex working conditions is improved, ensuring the stability and real-time performance of parameter estimation.

[0362] High computational efficiency: By combining genetic algorithm with particle swarm optimization, the speed of parameter optimization calculation is improved, the computational cost is reduced, and efficient parameter solution is achieved.

[0363] Online monitoring and dynamic adjustment: The system supports real-time data collection and status monitoring, and optimizes photovoltaic system operating parameters based on adaptive adjustment strategies to improve the power output and stability of the photovoltaic system.

[0364] Intelligent operation control: Through DC-DC converter optimization and MPPT adjustment, the operation strategy of the photovoltaic system is optimized, intelligent control is achieved, and the long-term operation performance of photovoltaic cells is improved.

[0365] In summary, the present invention realizes a parameter identification method with high precision, high adaptability and efficient calculation by constructing a photovoltaic cell parameter identification system based on a composite model, which provides key technical support for the intelligent monitoring and optimization of photovoltaic systems and has broad application value.

[0366] Obviously, the embodiments described above are only some of the embodiments of the present application, rather than all of the embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned specific embodiments, or to make equivalent replacements for some of the technical features therein. Any equivalent structure made using the contents of the present application specification and the accompanying drawings, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present application.

Claims

1. A photovoltaic cell parameter identification method based on a composite model, characterized in that: The following steps are involved: Acquiring operating status data of the photovoltaic cell, including current, voltage, temperature, and light intensity data, and performing signal denoising, normalization, and feature extraction on the operating status data to generate pre-processed photovoltaic cell operating status data; Based on the preprocessed photovoltaic cell operating state data, a single diode model and a double diode model are used to calculate preliminary parameter estimation values ​​of the photovoltaic cell, and an initial parameter sequence of the photovoltaic cell is generated; Optimizing and calculating the initial parameter sequence of the photovoltaic cell, performing global search and optimal parameter solution using a genetic algorithm and a particle swarm optimization algorithm, and generating an optimized photovoltaic cell parameter set; Based on the optimized photovoltaic cell parameter set, a deep neural network model is trained in combination with historical photovoltaic cell operation data to obtain a parameter prediction model with error compensation; Based on the optimized photovoltaic cell parameter set and the error compensation parameter prediction model, an adaptive fusion algorithm is used to generate composite model identification parameters to construct an adaptive parameter module for the photovoltaic cell; Based on the composite model identification parameters, real-time monitoring data of the photovoltaic system is obtained, and the photovoltaic system operation state adjustment parameters are dynamically calculated to generate an adaptive adjustment plan for the photovoltaic system; Based on the photovoltaic system adaptive adjustment scheme, the control parameters of the photovoltaic system are optimized and adjusted, and the adjusted photovoltaic cell operating parameters are generated to realize the intelligent operation control of the photovoltaic cell.

2. The photovoltaic cell parameter identification method based on the composite model according to claim 1, characterized in that: The signal denoising comprises: The operating status data of the photovoltaic cell is subjected to wavelet transform denoising to eliminate high-frequency noise interference, and the denoised signal is optimized based on Kalman filtering to obtain the denoised photovoltaic cell time series data.

3. The photovoltaic cell parameter identification method based on the composite model according to claim 1, characterized in that: The calculation of the preliminary parameter estimates of the photovoltaic cell includes: The single diode model is used to calculate the equivalent circuit parameters of the photovoltaic cell, including photocurrent, saturation current, resistance parameters and series resistance, and the single diode model parameter sequence is obtained; The double-diode model is used to calculate the current shunt parameters of the photovoltaic cell, and the nonlinear correction coefficient of the photovoltaic cell is calculated in combination with the shadow shielding effect to obtain the double-diode model parameter sequence; A feature matching calculation is performed on the single diode model parameter sequence and the double diode model parameter sequence, a model with the smallest error is selected as a preliminary parameter estimation model, and an initial parameter sequence of the photovoltaic cell is generated.

4. The photovoltaic cell parameter identification method based on the composite model according to claim 1, characterized in that: The optimization calculation includes: Based on the initial parameter sequence of the photovoltaic cell, a genetic algorithm is used to perform a global search, multiple parameter populations are initialized, and genetic mutation and selection operations are performed according to a target fitness function to obtain an optimized population parameter sequence; Based on the optimized population parameter sequence, a particle swarm optimization algorithm is used to perform local fine-tuning, an optimal parameter value is searched based on a speed-position update mechanism, and key circuit parameters of the photovoltaic cell are iteratively optimized to obtain an optimized photovoltaic cell candidate parameter set; The optimized photovoltaic cell candidate parameter set is evaluated for adaptability, an optimal solution is selected based on a multi-objective optimization criterion, and a final optimized photovoltaic cell parameter set is generated.

5. The photovoltaic cell parameter identification method based on the composite model according to claim 1, characterized in that: The training of the parameter prediction model for error compensation includes: Based on the optimized photovoltaic cell parameter set, historical photovoltaic cell operation data is acquired, and the historical photovoltaic cell operation data is divided into time windows and feature label matched to generate a training data set; Based on the training data set, a long short-term memory network deep neural network model is constructed to learn the time series characteristics of photovoltaic cell parameters and perform hyperparameter optimization to obtain a pre-trained error compensation model; The optimized photovoltaic cell parameter set is input into the error compensation model, and a compensation value is calculated based on the historical error distribution of the photovoltaic cell to generate an error-corrected parameter prediction model.

6. The photovoltaic cell parameter identification method based on the composite model according to claim 1, characterized in that: The adaptive fusion algorithm includes: Based on the optimized photovoltaic cell parameter set and the error-compensated parameter prediction model, a multi-model consistency analysis is performed to calculate the prediction error weights of different models and generate a preliminary fusion parameter set; Based on the preliminary fusion parameter set, the Bayesian optimization algorithm is used to adaptively adjust the weights of different models, and nonlinear regression optimization is performed to obtain the optimal composite model identification parameters; Based on the optimal composite model identification parameters, an adaptive parameter module of the photovoltaic cell is constructed, and the adaptive parameter module is updated online to adapt to changes in photovoltaic system parameters under different working conditions.

7. The photovoltaic cell parameter identification method based on the composite model according to claim 1, characterized in that: The real-time monitoring data includes ambient temperature, light intensity, and battery aging: Perform trend analysis on the real-time monitoring data to obtain real-time status parameters of the photovoltaic system.

8. The photovoltaic cell parameter identification method based on the composite model according to claim 1, characterized in that: The intelligent operation control of the photovoltaic system includes: Based on the final operating parameters of the photovoltaic cells, the power output of the photovoltaic system is controlled, and the operating strategy is dynamically adjusted to achieve intelligent operation control of the photovoltaic cells.

9. A photovoltaic cell parameter identification system based on a composite model, applied to the photovoltaic cell parameter identification method based on a composite model according to any one of claims 1 to 8, characterized in that: Include: Data acquisition and preprocessing module, which obtains operating status data from various data sources of photovoltaic cells; Physical modeling and preliminary parameter calculation module, which models the equivalent circuit parameters of photovoltaic cells based on single-diode model and double-diode model, and calculates preliminary parameter estimates; Intelligent optimization and parameter solution module, which performs global optimization on the preliminary parameter estimates and uses genetic algorithm and particle swarm optimization algorithm to find the optimal photovoltaic cell parameter set; Error compensation and deep learning prediction module, which uses a deep neural network model to perform error compensation on the optimized photovoltaic cell parameter set; The composite model fusion and adaptive identification module builds an adaptive photovoltaic cell parameter identification framework based on physical models, optimization algorithms, and deep learning models; The real-time monitoring and dynamic adjustment module combines the real-time monitoring data of the photovoltaic system to calculate the optimal operating adjustment parameters of the photovoltaic system and dynamically optimize the system operating status; The feedback control and optimization execution module optimizes the control parameters of the photovoltaic system according to the real-time adjustment plan, realizing the intelligent operation control of the photovoltaic cells.

Images