Method for predicting photovoltaic power of fish-light system considering environmental changes and photovoltaic panel aging

Abstract

The present application relates to a kind of considering environmental change and photovoltaic panel aging fishlight system photovoltaic power prediction method, comprising: obtaining the data for fishlight system, power index analysis is carried out, so as to classify weather type, obtain the data set of each weather type;The correlation of each meteorological factor influencing photovoltaic power generation and photovoltaic power generation is analyzed, and the main influencing factor is screened out;Based on photovoltaic array mechanism model, the theoretical power value of parameter in data set is calculated, and the power abnormal point is identified and optimized;The relative aging degree of photovoltaic module is calculated, so as to modify photovoltaic output power in data set;According to the measured data of prediction day, find the historical data of similar day from data set, construct sample set and carry out iterative training to neural network;The measured data of prediction day is input into trained neural network, and photovoltaic power prediction result is obtained.Compared with prior art, the present application effectively improves the accuracy of fishlight system power prediction and the effectiveness of power supply.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power prediction technology, and in particular to a photovoltaic power prediction method for a fishery-solar system that takes into account environmental changes and photovoltaic panel aging. Background Technology

[0002] The fishery-solar-storage system includes solar photovoltaic panels, energy storage devices (such as battery packs), and intelligent control systems. The load includes various types of fishery-specific electrical equipment such as fishing boats, fishing gear, and aquaculture equipment. The photovoltaic output power of the fishery-solar system is significantly affected by weather conditions (such as light intensity, temperature, humidity, and wind speed) and environmental factors (such as wetlands, rivers, and lakes). On the one hand, changes in weather conditions, such as fluctuations in light intensity, temperature, humidity, and wind speed, directly affect the power generation of the photovoltaic system. Furthermore, in special environments such as wetlands, rivers, and lakes, high humidity and water reflection further exacerbate the fluctuations in photovoltaic module performance. On the other hand, long-term exposure to changing sunlight, temperature, humidity, and wind conditions leads to material degradation, increased contact resistance, and aging of encapsulation materials, manifesting as light attenuation, mechanical damage, and thermal stress, resulting in significant power loss.

[0003] The accuracy of short-term photovoltaic power forecasting is directly affected by meteorological factors. In-depth analysis of the impact of meteorological factors on short-term photovoltaic output and the search for historical similar days for the forecast date can help improve forecast accuracy. Some research has been conducted both domestically and internationally on finding historical similar days for the forecast date. Song Weiqiong et al. selected similar days by calculating the grey relational coefficient to improve forecast accuracy. Liu Hui et al. divided the feature data into linear and nonlinear types, and used the Pearson correlation coefficient (PCC) and maximum information coefficient to calculate the correlation between meteorological factors and power load, respectively. These methods all attempt to find similar day data using a single algorithm, resulting in low model prediction accuracy. Zhang Dahai et al. determined meteorological influencing factors through PCC, decomposed the historical dataset into intrinsic mode function (IMF) components at different frequencies, and used long-short-term time series networks and extreme learning machine models to predict high-frequency and low-frequency IMF components, respectively. Li Bin et al. combined grey relational coefficient, correlation weighting method, and maximum information coefficient to determine similar day curves to improve prediction accuracy.

[0004] Regarding the aging characteristics of photovoltaic modules, Ma Mingyao et al. used cell fill factor to determine the degree of aging failure and analyzed its current-voltage output characteristics. Existing photovoltaic power output prediction technologies lack consideration for module aging phenomena caused by uncertainties in the installation environment and changes in meteorological conditions. Aging data is often filtered out as outlier data to reduce its impact on output data. To improve the accuracy and reliability of photovoltaic power generation prediction, it is necessary to propose a prediction method that comprehensively considers different weather scenarios and aging degradation environments. By collecting and analyzing photovoltaic power generation data under different weather and environmental conditions, a high-precision power prediction model is established using a combination of data physics modeling and other methods.

[0005] Against this backdrop, in order to improve the management level of fishery-solar systems in special environments such as wetlands, rivers and lakes, it is urgent to study a method for predicting the photovoltaic power of fishery-solar systems that takes into account the impact of environmental changes and the aging characteristics of photovoltaic panels, so as to achieve all-weather and reliable power supply, ensure the normal operation of fish farms, and provide reference for optimizing energy management and energy storage scheduling. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a photovoltaic power prediction method for a fishery-solar system that takes into account environmental changes and photovoltaic panel aging. It analyzes weather conditions such as light intensity, temperature, humidity, and wind speed, as well as the aging characteristics and power-related characteristics of photovoltaic modules, and uses a BP neural network to accurately predict photovoltaic power generation.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] A method for predicting photovoltaic power in a fishery-solar system that takes into account environmental changes and photovoltaic panel aging includes the following steps:

[0009] Data for predicting photovoltaic power in fishery solar power systems is acquired, power index analysis is performed, and weather types are classified to obtain datasets for various weather types.

[0010] Based on the dataset, the correlation between various meteorological factors affecting photovoltaic power generation and photovoltaic power generation is analyzed, thereby identifying the main influencing factors that have a significant impact on power.

[0011] Based on the photovoltaic array mechanism model, theoretical power values ​​are calculated for the irradiance and temperature parameters in the dataset, and compared with the corresponding measured power values ​​to identify and optimize power anomalies.

[0012] The IV curve of the photovoltaic module is tested by an IV tester, and the relative aging degree of the photovoltaic module is calculated, thereby correcting the photovoltaic output power in the data.

[0013] Based on the measured data of the predicted day, analyze the power index, classify the weather type, and find historical data of similar days from the dataset to construct a sample set;

[0014] The neural network is iteratively trained using a sample set until a well-trained neural network is obtained;

[0015] Based on the measured data of the predicted date, the features corresponding to the main influencing factors are extracted and input into the trained neural network to obtain the photovoltaic power prediction results of the fishery-solar system.

[0016] Furthermore, the process of classifying weather types specifically involves:

[0017] Instantaneous power output and daily cumulative power generation are used as power indicators. The similarity between different weather types is calculated by Euclidean distance, thereby classifying the weather types.

[0018] The Euclidean distance is calculated as follows:

[0019]

[0020] In the formula, and E represents the average power generation at time i for two different weather types; p and E q These represent the cumulative daily power generation for the two different weather types; n is the number of time periods divided by the selected day.

[0021] Furthermore, the method employs a grey relational analysis method to analyze the correlation between various meteorological factors and photovoltaic power generation.

[0022] The method for analyzing factors affecting photovoltaic power generation based on grey relational analysis includes:

[0023] Normalize the data in the dataset;

[0024] For the normalized data, construct a feature vector with each meteorological factor as a similarity factor;

[0025] Based on the feature vectors, a reference sequence and a comparison sequence are constructed, and the difference is calculated to obtain the difference sequence;

[0026] The minimum value of each difference sequence is defined as the minimum range, and the maximum value of each difference sequence is defined as the maximum range; thus, the correlation coefficient between the comparison sequence and the reference sequence is calculated to reflect the shape differences of each output time series curve.

[0027] The correlation coefficients of each comparison sequence are normalized to obtain the corresponding factor weight coefficients, which describe the relative effects of each meteorological factor on the system, thereby screening out the main influencing factors that have a greater impact on power.

[0028] Furthermore, the calculation expression for the normalization process is as follows:

[0029]

[0030] In the formula, x i For the i-th sequence; x i (k) represents the k-th data point in sequence i; x i ′(k) is the normalized value of the k-th data point; Let be the maximum and minimum data points in the i-th sequence, respectively; m is the total number of data points in the sequence; n is the total number of data points in the sequence.

[0031] The expression for calculating the difference sequence is:

[0032] Δ i (k)=|x′0(k)-x′ i (k)|(i=0,1,2,...,m; k=1,2,3,...,n)

[0033] In the formula, Δ i =(Δ i (1),Δ i (2),…,Δ i (n)); x0=(x0(1),x0(2),…,x0(n)) is the reference sequence. When i>0, then x is called i =(x i (1),x i (2),…,x i (n)) is the comparison sequence; while x′0 and x′ i All are normalized sequences;

[0034] The formula for calculating the correlation coefficient is as follows:

[0035]

[0036] In the formula, ξ 0i ρ is the correlation coefficient; ρ is the resolution coefficient. The minimum range; The maximum range;

[0037] The formula for calculating the factor weight coefficient is as follows:

[0038]

[0039] In the formula, ξi is the factor weight coefficient of the i-th comparison sequence.

[0040] Furthermore, the photovoltaic array mechanism model includes a current source, a diode, an equivalent parallel resistance reflecting the non-ideal characteristics of the PN junction and the influence of nearby impurities, a contact resistance composed of electrodes and semiconductors, and a series resistance composed of semiconductor and electrode resistances.

[0041] The diode and the equivalent parallel resistor are connected in parallel across the current source, and the contact resistor and the series resistor are connected in series and then connected in parallel across the current source.

[0042] Furthermore, based on the IV relationship expression of the photovoltaic array mechanism model, the theoretical power value is calculated. The IV relationship expression is as follows:

[0043]

[0044] In the formula, I is the load current; p I is the current flowing through the parallel resistor; ph Photocurrent; I d For dark current; I sd R is the reverse saturation current of the diode; U is the load voltage; s and R p These represent the series resistance and the equivalent parallel resistance, respectively; 'a' is the diode ideality factor; T emp Here, k is the temperature of the photovoltaic cell; k0 is the Boltzmann constant, k0 = 1.381 × 10⁻⁶. -23 J / K; q is the electron charge, q = 1.6021 × 10 -19 C;

[0045] in,

[0046]

[0047] In the formula, G is the actual irradiance; G ref For reference irradiance; T ref Reference temperature; I sc_ref For the short-circuit current at reference irradiance and reference temperature; α i The short-circuit current temperature coefficient;

[0048]

[0049] In the formula, E g For band gap energy; I os The reverse saturation current of the diode under reference temperature conditions;

[0050]

[0051] In the formula, I sc and U oc These represent the short-circuit current and open-circuit voltage of the photovoltaic cell at the reference temperature, respectively.

[0052] Furthermore, the calculation process for the relative aging degree of the photovoltaic module includes:

[0053] Based on the tested IV curve of the photovoltaic module, the measured open-circuit voltage, short-circuit current, and maximum power point are obtained, and the fill factor is calculated. The expression for the fill factor is as follows:

[0054]

[0055] In the formula, σ FF U is the fill factor; ocm I is the open-circuit voltage. scm This is the short-circuit current; U max I is the voltage at the maximum power point. max This is the current corresponding to the maximum power point;

[0056] Based on the characteristic that the fill factor of photovoltaic modules is generally greater than 0.7 during normal operation, the relative aging degree is calculated, and the corresponding calculation expression is as follows:

[0057]

[0058] In the formula, d e This refers to the relative degree of aging.

[0059] The calculation expression for correcting the photovoltaic output power in the dataset is as follows:

[0060] P s =P t (1-K loss )

[0061] In the formula, P t P represents the photovoltaic output power without considering aging effects before correction. s To account for the effects of aging on photovoltaic output power; K loss This is the power loss factor.

[0062] Furthermore, the process of finding historical data of similar days from the dataset specifically involves:

[0063] First, similar days to the predicted date are found using the highest and lowest temperatures and weather conditions. Then, the day with the closest characteristics is selected as the similar day, and a comprehensive similarity index is used as the evaluation criterion. The formula for calculating the comprehensive similarity index is as follows:

[0064]

[0065] In the formula, This is the calculated value of the comprehensive similarity index; ξ 0i (k) is the correlation coefficient between the current sequence i and the reference sequence; N0 is the number of data in the sequence.

[0066] Furthermore, the neural network is a BP neural network. The BP neural network normalizes the collected data and then predicts the photovoltaic power of the fishery-solar system. The normalization error is evaluated using the normalized absolute mean error and the normalized root mean square error.

[0067] Furthermore, the evaluation metrics used during the training of the BP neural network include:

[0068]

[0069] In the formula, MAPE is the mean absolute percentage error; MAE is the mean absolute error; MBE is the mean error; RMSE is the root mean square error; x′ i x represents the predicted output value of the BP neural network. i is the actual value; N is the number of data points in the sequence.

[0070] Compared with the prior art, the present invention has the following advantages:

[0071] (1) The present invention first classifies the weather type of the data according to the power index, and analyzes the correlation of photovoltaic power generation power for various meteorological factors affecting photovoltaic power generation, extracts the main influencing factors, and can select similar day data of the corresponding weather type for the predicted day data under different weather conditions, and can further select similar day data based on the main influencing factors and the comprehensive similarity index for network training.

[0072] On the other hand, by testing the IV curve of photovoltaic modules, the relative aging degree of photovoltaic modules can be calculated, thereby correcting the centralized photovoltaic output power of the data. When the aging degree of photovoltaic modules is severe, the measured power data is treated as a separate category for power prediction, thereby improving the accuracy of the prediction results.

[0073] Overall, this invention effectively improves the accuracy of power prediction and the effectiveness of power supply, helps to optimize the scheduling of energy storage devices and power distribution in complex environments, improves the overall efficiency of the system, ensures a stable power supply for fisheries under various weather conditions, and enhances fishery production efficiency. Attached Figure Description

[0074] Figure 1 This is a flowchart illustrating a method for predicting photovoltaic power in a fishery-solar system that takes into account environmental changes and photovoltaic panel aging, provided in an embodiment of the present invention.

[0075] Figure 2 This is an equivalent circuit diagram of a single diode model of a photovoltaic cell provided in an embodiment of the present invention;

[0076] Figure 3 This is a flowchart illustrating a grey relational analysis method provided in an embodiment of the present invention;

[0077] Figure 4 This is a schematic diagram of the IV curves of a photovoltaic module before and after aging, provided in an embodiment of the present invention. The solid line is the normal IV curve, and the dashed line is the measured IV curve after aging.

[0078] Figure 5 This is a schematic diagram of a photovoltaic power prediction process based on a BP neural network provided in an embodiment of the present invention. Detailed Implementation

[0079] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0080] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0081] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0082] Example 1

[0083] like Figure 1 As shown, this embodiment provides a method for predicting the photovoltaic power of a fishery-solar system that takes into account environmental changes and photovoltaic panel aging, including the following steps:

[0084] S1: Obtain data for photovoltaic power prediction of the fishery solar system, perform power index analysis, and classify weather types to obtain datasets for various weather types;

[0085] S2: Based on the dataset, analyze the correlation between various meteorological factors affecting photovoltaic power generation and photovoltaic power generation output, thereby identifying the main influencing factors that have a significant impact on power output;

[0086] S3: Based on the photovoltaic array mechanism model, calculate the theoretical power value for the irradiance and temperature parameters in the dataset, compare it with the corresponding measured power value, and identify and optimize power anomalies.

[0087] S4: Test the IV curve of the photovoltaic module using an IV tester, calculate the relative aging degree of the photovoltaic module, and thus correct the photovoltaic output power in the data set;

[0088] S5: Analyze the power index based on the measured data of the predicted day, classify the weather type, and find historical data of similar days from the dataset to construct a sample set;

[0089] S6: Iteratively train the neural network using a sample set until a well-trained neural network is obtained;

[0090] S7: Extract the features corresponding to the main influencing factors based on the measured data of the predicted date, and input them into the trained neural network to obtain the photovoltaic power prediction results of the fishery-solar system.

[0091] The following is a detailed description of each step:

[0092] 1. Weather type classification considering environmental differences in step S1

[0093] The process of classifying weather types is as follows:

[0094] Instantaneous power output and daily cumulative power generation are used as power indicators. The similarity between different weather types is calculated by Euclidean distance, thereby classifying the weather types.

[0095] The expression for calculating Euclidean distance is:

[0096]

[0097] In the formula, This represents the average power generation during the i-th time period under standard weather conditions. E represents the average power generation of the weather type to be classified during the i-th time period; p and E q These represent the daily cumulative power generation for two different weather types; n is the number of time periods divided into the selected days.

[0098] Specifically, on sunny days, solar radiation is strongest, allowing photovoltaic modules to reach their highest power output. On cloudy days, partial cloud cover obstructs sunlight, leading to unstable radiation intensity and significant fluctuations in power output. On overcast days, complete cloud cover significantly reduces radiation intensity, resulting in lower power output. Rainy days not only involve cloud cover but also raindrops further reducing sunlight while increasing humidity and cooling effects, leading to a significant decrease in power output. Foggy days are characterized by the scattering and absorption of sunlight by suspended water droplets, weakening radiation intensity and reducing the power output of photovoltaic modules. Weather conditions include various meteorological factors such as cloud cover, ambient temperature, wind speed, and air pressure.

[0099] By monitoring and calculating power indicators, a comprehensive understanding of the system's operating status and efficiency can be obtained. Key power indicators include instantaneous power output, daily cumulative power generation, and capacity factor. Instantaneous power output reflects the photovoltaic system's power generation capacity at a specific moment and is a crucial parameter for real-time monitoring; daily cumulative power generation is a measure of the total power generation throughout the day and can assess the system's daily performance.

[0100] By combining key power indicators with weather quality and conditions, the weather types of sunny, cloudy, overcast, rainy, and foggy are classified, thereby enabling power classification and forecasting.

[0101] Instantaneous power output (e.g., deriving a power value every 15 minutes) and daily cumulative power generation are used as power indicators. The similarity between different types is calculated using Euclidean distance, thereby classifying weather types.

[0102] 2. Correlation analysis between multiple factors and photovoltaic power generation in step S2

[0103] The method employs a grey relational analysis of factors influencing photovoltaic power generation to analyze the correlation between various meteorological factors and photovoltaic power generation.

[0104] The analysis methods for factors affecting photovoltaic power generation based on grey relational analysis include:

[0105] Normalize the data in the dataset;

[0106] For the normalized data, construct a feature vector with each meteorological factor as a similarity factor;

[0107] Based on the feature vectors, a reference sequence and a comparison sequence are constructed, and the difference is calculated to obtain the difference sequence;

[0108] The minimum value of each difference sequence is defined as the minimum range, and the maximum value of each difference sequence is defined as the maximum range; thus, the correlation coefficient between the comparison sequence and the reference sequence is calculated to reflect the shape differences of each output time series curve.

[0109] The correlation coefficients of each comparison sequence are normalized to obtain the corresponding factor weight coefficients, which describe the relative effects of each meteorological factor on the system, thereby screening out the main influencing factors that have a greater impact on power.

[0110] Specifically, this involves analyzing the correlation between irradiance, ambient temperature, wind speed, relative humidity, geographical location, and photovoltaic output power, and establishing a grey relational analysis method for analyzing factors affecting photovoltaic power generation. The data is then normalized to the 0-1 range.

[0111]

[0112] In the formula, x i Let x be the i-th sequence in the gray system factor set; i (k) represents the k-th data point in sequence i; x i ′(k) is the normalized value of the k-th data point; These are the maximum and minimum data points in the i-th sequence, respectively.

[0113] Grey relational analysis is used to determine the degree to which each influencing factor interferes with the system's changes based on the strength of their correlation. The steps of grey relational analysis are as follows: Figure 3 As shown.

[0114] 2.1 Selection of eigenvectors.

[0115] Construct an eigenvector X = [E,T] using solar irradiance, photovoltaic panel temperature, and weather conditions as similarity factors. m [,...]. Where E is solar irradiance; T m This refers to the temperature of the solar panel.

[0116] 2.2 Calculate the correlation coefficient between the feature vector and each time period in the historical dataset, i.e., the similarity index.

[0117] Define the difference Δ between the feature vector and the historical data being compared. i (k), as shown in the formula

[0118] Δ i (k)=|x′0(k)-x′ i (k)|(i=0,1,2,...,m; k=1,2,3,...,n) (3)

[0119] In the formula Δ i =(Δ i (1),Δ i (2),…,Δ i (n)); x0=(x0(1),x0(2),…,x0(n)) is the reference sequence. When i>0, then x is called i =(xi (1),x i (2),…,x i (n)) is the comparison sequence; while x′0 and x′ i This is the normalized sequence.

[0120] The minimum value of each difference sequence is defined as the minimum range. The maximum value in each difference sequence is the maximum range. Calculate the correlation coefficient ξ between the i-th comparison sequence and the reference sequence. 0i (k) is used to reflect the shape differences of multiple output time-series curves of a photovoltaic panel, and can be expressed as:

[0121]

[0122] In the formula, ρ is the resolution coefficient, which ranges from 0 to 1, and is usually taken as 0.5.

[0123] 2.3 Calculation of Grey Relational Degree.

[0124] Normalizing the grey relational degree yields the factor weight coefficient ξ for the i-th comparison sequence. i for:

[0125]

[0126] Factor weight coefficients describe the relative magnitude of the influence of each factor on the system.

[0127] 3. Anomaly analysis based on mechanistic model and power comparison in step S3

[0128] The photovoltaic array mechanism model includes a current source, a diode, an equivalent parallel resistance reflecting the non-ideal characteristics of the PN junction and the influence of nearby impurities, a contact resistance consisting of electrodes and semiconductors, and a series resistance consisting of semiconductor and electrode resistances.

[0129] The diode and the equivalent parallel resistor are connected in parallel across the current source, and the contact resistance and the series resistance are connected in series and then connected in parallel across the current source.

[0130] In other words, in terms of mechanistic modeling, the equivalent circuit of a single-diode model of a photovoltaic cell is as follows: Figure 2 As shown, its IV relationship expression is shown in equation (2). It includes a current source in parallel with the diode, an equivalent parallel resistance reflecting the non-ideal characteristics of the PN junction and the influence of nearby impurities, and a series resistance consisting of the contact resistance between the electrode and the semiconductor and the resistance of the semiconductor and the electrode.

[0131]

[0132] In the formula, I is the load current;p I is the current flowing through the parallel resistor; ph Photocurrent; I d For dark current; I sd U is the reverse saturation current of the diode; U is the load voltage, which is the open-circuit voltage of the photovoltaic cell when no load is connected; R s and R p These are the equivalent series and parallel resistances, respectively; 'a' is the diode ideality factor; T emp Here, k is the temperature of the photovoltaic cell; k0 is the Boltzmann constant, k0 = 1.381 × 10⁻⁶. -23 J / K; q is the electron charge, q = 1.6021 × 10 -19 C.

[0133] The photogenerated current I in equation (6) ph It varies with irradiance and temperature, and its relationship with irradiance and temperature is expressed as follows:

[0134]

[0135] In the formula, G is the actual irradiance; G ref For reference irradiance, G is typically used. ref =1000W / m 2 ;T ref For reference temperature, T is usually used. ref =25℃; I sc_ref For the short-circuit current at reference irradiance and reference temperature; α i This is the temperature coefficient of the short-circuit current.

[0136] The diode reverse saturation current I in equation (6) sd The expression is:

[0137]

[0138] In the formula, E g This is the bandgap energy, which is approximately 1.12 eV for silicon at a standard temperature of 25°C; I os This represents the reverse saturation current of the diode under reference temperature conditions.

[0139] Reverse saturation current I of the diode at reference temperature os for:

[0140]

[0141] In the formula, I sc and U oc These represent the short-circuit current and open-circuit voltage of the photovoltaic cell at the reference temperature, respectively.

[0142] As shown in equations (6)-(9), light intensity and temperature have a significant impact on photovoltaic power generation, and are direct influencing factors. In contrast, salt spray in the environment primarily alters the temperature of the solar cells, and seawater primarily alters the received irradiance, thus affecting the output characteristics and consequently the output power.

[0143] 4. Photovoltaic output power correction taking into account module aging characteristics in step S4

[0144] The calculation process for the relative aging degree of photovoltaic modules includes:

[0145] Based on the tested IV curve of the photovoltaic module, the measured open-circuit voltage, short-circuit current and maximum power point are obtained, and the fill factor is calculated.

[0146] Based on the characteristic that the fill factor of photovoltaic modules is generally greater than 0.7 when they are working normally, the relative aging degree is calculated.

[0147] The photovoltaic output power in the dataset is corrected based on the relative aging degree.

[0148] Specifically, the schematic diagram of the IV curves of photovoltaic modules before and after aging is shown below. Figure 4 As shown. The IV curve of the component is tested using an IV tester, based on the measured open-circuit voltage V. ocm Short-circuit current I scm and the maximum power point P max The fill factor is calculated using the following expression:

[0149]

[0150] The fill factor of components during normal operation is generally greater than 0.7. The relative aging degree d is then calculated. e The expression is:

[0151]

[0152] Considering the aging characteristics of the components, a power correction based on the degradation rate is introduced into the photovoltaic output power:

[0153] P s =P t (1-K loss (12)

[0154] In the formula, P t P represents the photovoltaic output power without considering aging effects before correction. s To account for the effects of aging on photovoltaic output power; K loss This is the power loss factor.

[0155] As usage time and environmental factors affect the modules, a drop below 0.5 is considered a sign of severe aging. When the aging level is deemed severe, the power data measured under severe aging conditions is treated as a separate category for power prediction.

[0156] 5. Selection of similar days based on grey relational analysis in step S5

[0157] The process of finding historical data with similar dates from a dataset is as follows:

[0158] First, find similar days to the predicted day by taking the highest and lowest temperatures and weather conditions. Then, select the day with the closest characteristics as the similar day and use the comprehensive similarity index as the evaluation standard.

[0159] Specifically, the power generation of photovoltaic panels is mainly determined by factors such as irradiance and atmospheric visibility. However, for a specific region's fishery-solar hybrid project, the photovoltaic output sequence shows certain similarities to historical data from the same period. Considering the weather conditions at different times of the day, we first identify similar days by analyzing the highest and lowest temperatures and weather conditions; then, we select the similar day with the closest characteristics for power prediction.

[0160] Because the high dimensionality of the input data results in a large number of correlation coefficients, it is difficult to obtain the overall similarity between curves. Therefore, a comprehensive similarity index is defined. It can be represented as:

[0161]

[0162] In the formula, N0 is the number of data items in the sequence.

[0163] Then, based on similarity index and comprehensive similarity, the data are sorted, and historical curves with high similarity to the input data at the time to be predicted are selected as similar day datasets to train the predictor and verify the prediction accuracy.

[0164] 6. Power prediction method based on BP neural network in steps S6 and S7

[0165] The neural network is a BP neural network. After normalizing the collected data, the BP neural network predicts the photovoltaic power of the fishery-solar system, and uses the normalized absolute mean error and the normalized root mean square error to evaluate the normalization error.

[0166] Specifically: such as Figure 5 As shown, the BP neural network is a type of multilayer feedforward neural network with error backpropagation correction capability. It has strong nonlinear mapping capability, can handle complex mapping relationships between input and output, and has self-learning and adaptive capabilities. It can automatically adjust the weights and constraints of the network to improve the analysis accuracy and is used to analyze the impact of complex environments on photovoltaic output power.

[0167] Step 1: Create a sample set for BP neural network analysis. Data sets are created for each of the distinguished weather types and preprocessed. Correlation analysis is performed on multi-factor meteorological indicators to identify the parameters that have a significant impact on power, resulting in a meteorological feature set. The input and output vectors of the correlated factors and power are then merged to form the training samples.

[0168] Step 2, training the BP neural network based on the sample set. The collected data is normalized, and the normalized absolute mean error e is calculated. NMAE and normalized root mean square error e NRMSE The expression is:

[0169]

[0170] In the formula, x′(i) is the predicted output value of the neural network; x(i) is the normalized measured power value; and N is the number of predicted samples.

[0171] The processed data is divided proportionally into training, validation, and test sets. The first two sets are used for training the BP neural network, while the test set is used to test the generalization ability of the trained network. The neural network structure, such as the number of layers, the number of neurons per layer, the types of activation functions, the training method, and parameters (such as the learning rate), are further determined. Based on the selected training method and samples, the neural network is trained. Finally, the test set is used for testing, and the network output is compared with the actual output to evaluate the accuracy. If the accuracy does not meet the requirements, the structure, learning method, or parameters of the BP neural network are changed, and it is retrained until the requirements are met.

[0172] The accuracy of power prediction can be evaluated by comparing the network output after verification with the actual output. Evaluation indicators include:

[0173]

[0174] In the formula, MAPE is the mean absolute percentage error; MAE is the mean absolute error; MBE is the mean error; and RMSE is the root mean square error.

[0175] Step 3: Use the trained BP neural network to predict photovoltaic power. Using the meteorological indicators for the target day as input, input the trained BP neural network to obtain the corresponding photovoltaic power data, and then perform inverse normalization to obtain the predicted actual value.

[0176] When a BP neural network is trained using historical data, the meteorological information for the predicted day is input into the network to obtain the output power at each time of the predicted day. After inverse normalization, the output power is compared with the measured photovoltaic power. The prediction error of the network can be calculated according to different error evaluation methods. If it meets the requirements (generally the prediction error is between 15% and 35%), it means that the network can be used to predict the output power of the photovoltaic power generation module.

[0177] Establish a regression model between predicted and measured power. Calculate the goodness of fit R0. 2 The indicator is calculated as follows:

[0178]

[0179] In the formula, SSR is the regression sum of squares; SSE is the residual sum of squares; and SST is the total sum of squares.

[0180] Based on the model validation results, the goodness of fit Rfit is... 2 A value close to 1 indicates that the actual photovoltaic power and the predicted photovoltaic output have a good fit.

[0181] Example 2

[0182] This embodiment provides a photovoltaic power prediction device for a fishery-solar system that takes into account environmental changes and photovoltaic panel aging. It is characterized by including a memory and a processor. The memory stores a computer program, and the processor calls the computer program to execute the steps of the photovoltaic power prediction method for a fishery-solar system that takes into account environmental changes and photovoltaic panel aging as described in Embodiment 1.

[0183] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for predicting photovoltaic power in a fishery-solar system considering environmental changes and photovoltaic panel aging, characterized in that, Includes the following steps: Data for predicting photovoltaic power in fishery solar power systems is acquired, power index analysis is performed, and weather types are classified to obtain datasets for various weather types. Based on the dataset, the correlation between various meteorological factors affecting photovoltaic power generation and photovoltaic power generation is analyzed, thereby identifying the main influencing factors that have a significant impact on power. Based on the photovoltaic array mechanism model, theoretical power values ​​are calculated for the irradiance and temperature parameters in the dataset, and compared with the corresponding measured power values ​​to identify and optimize power anomalies. The IV curve of the photovoltaic module is tested by an IV tester, and the relative aging degree of the photovoltaic module is calculated, thereby correcting the photovoltaic output power in the data. Based on the measured data of the predicted day, analyze the power index, classify the weather type, and find historical data of similar days from the dataset to construct a sample set; The neural network is iteratively trained using a sample set until a well-trained neural network is obtained; Based on the measured data of the predicted date, the features corresponding to the main influencing factors are extracted and input into the trained neural network to obtain the photovoltaic power prediction results of the fishery-solar system. The calculation process for the relative aging degree of the photovoltaic module includes: Based on the tested IV curve of the photovoltaic module, the measured open-circuit voltage, short-circuit current, and maximum power point are obtained, and the fill factor is calculated. The expression for the fill factor is as follows: In the formula, The fill factor; This is the open-circuit voltage; This is the short-circuit current; This is the voltage corresponding to the maximum power point; This is the current corresponding to the maximum power point; Based on the characteristic that the fill factor of photovoltaic modules is generally greater than 0.7 during normal operation, the relative aging degree is calculated, and the corresponding calculation expression is as follows: In the formula, This refers to the relative degree of aging. The calculation expression for correcting the photovoltaic output power in the dataset is as follows: In the formula, P t This refers to the photovoltaic output power without considering the effects of aging before correction. P s To account for the effects of aging on photovoltaic output power; K loss This is the power loss factor.

2. The photovoltaic power prediction method for a fishery-solar system considering environmental changes and photovoltaic panel aging as described in claim 1, characterized in that, The process of classifying weather types is as follows: Instantaneous power output and daily cumulative power generation are used as power indicators. The similarity between different weather types is calculated by Euclidean distance, thereby classifying the weather types. The Euclidean distance is calculated as follows: In the formula, and Two different weather types in i The average power generation at any given time; E p and E q These represent the cumulative daily power generation for two different weather conditions. n This is the number of time periods to be selected from the day.

3. The photovoltaic power prediction method for a fishery-solar system considering environmental changes and photovoltaic panel aging as described in claim 1, characterized in that, The method uses a grey relational analysis method to analyze the correlation between various meteorological factors and photovoltaic power generation. The method for analyzing factors affecting photovoltaic power generation based on grey relational analysis includes: Normalize the data in the dataset; For the normalized data, construct a feature vector with each meteorological factor as a similarity factor; Based on the feature vectors, a reference sequence and a comparison sequence are constructed, and the difference is calculated to obtain the difference sequence; The minimum value of each difference sequence is defined as the minimum range, and the maximum value of each difference sequence is defined as the maximum range; thus, the correlation coefficient between the comparison sequence and the reference sequence is calculated to reflect the shape differences of each output time series curve. The correlation coefficients of each comparison sequence are normalized to obtain the corresponding factor weight coefficients, which describe the relative effects of each meteorological factor on the system, thereby screening out the main influencing factors that have a greater impact on power.

4. The photovoltaic power prediction method for a fishery-solar system considering environmental changes and photovoltaic panel aging as described in claim 3, characterized in that, The calculation expression for the normalization process is as follows: In the formula, For the first A sequence; For sequence i The first in One data point; For the first The normalized values ​​of the data points; , The first The maximum and minimum data points in the sequence; m is the total number of data points in the sequence; n is the total number of data points in the sequence; The expression for calculating the difference sequence is: In the formula, ; As a reference sequence, when At that time, it is called For comparison sequences; and and All are normalized sequences; The formula for calculating the correlation coefficient is as follows: In the formula, The correlation coefficient; The resolution coefficient; The minimum range; The maximum range; The formula for calculating the factor weight coefficient is as follows: In the formula, For the first Factor weight coefficients of each comparison sequence.

5. The photovoltaic power prediction method for a fishery-solar system considering environmental changes and photovoltaic panel aging according to claim 1, characterized in that, The photovoltaic array mechanism model includes a current source, a diode, an equivalent parallel resistance reflecting the non-ideal characteristics of the PN junction and the influence of nearby impurities, a contact resistance composed of electrodes and semiconductors, and a series resistance composed of semiconductor and electrode resistances. The diode and the equivalent parallel resistor are connected in parallel across the current source, and the contact resistor and the series resistor are connected in series and then connected in parallel across the current source.

6. The method for predicting photovoltaic power in a fishery-solar system considering environmental changes and photovoltaic panel aging, as described in claim 5, is characterized in that... Based on the photovoltaic array mechanism model IV The relational expression is used to calculate the theoretical power value. IV The relational expression is: In the formula, I This is the load current; I p This is the current flowing through the parallel resistor; I ph Photocurrent; I d It is dark current; I sd This is the reverse saturation current of the diode; U This is the load voltage; R s and R p These are the series resistance and the equivalent parallel resistance, respectively; a This is the diode ideality factor; T emp The temperature of the photovoltaic cell; k 0 is the Boltzmann constant. k 0 = 1.381 × 10 -23 J / K; q For electron charge, q =1.6021×10 -19 C; in, In the formula, G This represents the actual irradiance. G ref For reference irradiance; T ref For reference temperature; I sc_ref The short-circuit current is based on reference irradiance and reference temperature; The short-circuit current temperature coefficient; In the formula, E g It is the band gap energy; I os The reverse saturation current of the diode under reference temperature conditions; In the formula, I sc and U oc These represent the short-circuit current and open-circuit voltage of the photovoltaic cell at the reference temperature, respectively.

7. The method for predicting photovoltaic power in a fishery-solar system considering environmental changes and photovoltaic panel aging, as described in claim 1, is characterized in that... The process of finding historical data with similar dates from the dataset is as follows: First, similar days to the predicted date are found using the highest and lowest temperatures and weather conditions. Then, the day with the closest characteristics is selected as the similar day, and a comprehensive similarity index is used as the evaluation criterion. The formula for calculating the comprehensive similarity index is as follows: In the formula, This is the calculated value of the comprehensive similarity index; For the current sequence i Correlation coefficient with reference sequence; N 0 represents the number of data items in the sequence.

8. The method for predicting photovoltaic power in a fishery-solar system considering environmental changes and photovoltaic panel aging, as described in claim 1, is characterized in that... The neural network is a BP neural network. After normalizing the collected data, the BP neural network predicts the photovoltaic power of the fishery-solar system and uses the normalized absolute mean error and normalized root mean square error to evaluate the normalization error.

9. The photovoltaic power prediction method for a fishery-solar system considering environmental changes and photovoltaic panel aging according to claim 8, characterized in that, The evaluation metrics used in the training process of the BP neural network include: In the formula, Mean absolute percentage error; Mean absolute error; This represents the average error. This is the root mean square error; This represents the predicted output value of the BP neural network. x i This is the actual value; N The number of data items in the sequence.

Images