Photovoltaic power generation prediction method and system

Abstract

The present invention relates to the field of data processing technology, and in particular to a photovoltaic power generation prediction method and system. The content includes: collecting real-time power generation data and real-time environmental data of photovoltaic power generation, and performing first data processing and second data processing on the real-time power generation data and real-time environmental data to obtain a power generation data set and an environmental data set; extracting features and optimizing the power generation data set to obtain an optimized feature set; analyzing and quantifying the environmental data set to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation; predicting photovoltaic power generation based on the optimized feature set, and then optimizing the prediction of photovoltaic power generation based on the quantitative feedback to obtain an optimized prediction result. The technical problems of low accuracy and reliability of photovoltaic power generation data processing and poor accuracy of photovoltaic power generation prediction are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a photovoltaic power generation power prediction method and system. Background Art

[0002] In photovoltaic power generation systems, accurate prediction of power generation is crucial because it directly affects the efficiency of energy management and grid dispatch. However, accurate power generation prediction faces multiple challenges due to changing environmental conditions and complex equipment dynamics. Traditional prediction methods usually rely on linear analysis of historical data. Although these methods perform well when the data is stable and the change amplitude is not large, they often do not work well when dealing with nonlinear problems such as sudden environmental changes or equipment failures.

[0003] That is, the above technology has at least the following technical problems: low accuracy and reliability of photovoltaic power generation data processing, and poor accuracy of photovoltaic power generation power prediction. Summary of the invention

[0004] The present invention provides a photovoltaic power prediction method and system to solve the technical problems of low accuracy and reliability of photovoltaic power data processing and poor accuracy of photovoltaic power prediction.

[0005] A photovoltaic power generation power prediction method and system of the present invention specifically includes the following technical solutions:

[0006] A photovoltaic power generation power prediction method comprises the following steps:

[0007] S1. Collecting real-time power generation data and real-time environmental data of photovoltaic power generation, and performing first data processing and second data processing on the real-time power generation data and real-time environmental data to obtain a power generation data set and an environmental data set;

[0008] S2. Extract features and optimize the power generation data set to obtain an optimized feature set; analyze and quantify the environmental data set to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation; predict photovoltaic power generation based on the optimized feature set, and then optimize the prediction of photovoltaic power generation based on the quantitative feedback to obtain an optimized prediction result.

[0009] Preferably, the S1 specifically includes:

[0010] The first data processing is performed on the real-time power generation data and real-time environmental data of photovoltaic power generation, specifically including: first removing the random noise of the real-time environmental data and the real-time power generation data; then filling the missing values; then identifying the outliers, and correcting or deleting the detected outliers; further standardizing and formatting are performed to obtain the cleaned and formatted power generation data set and environmental data set.

[0011] Preferably, the S1 specifically includes:

[0012] In the process of implementing the first data processing, a continuous volatility detection algorithm is introduced to perform abnormal data detection.

[0013] Preferably, the S1 specifically includes:

[0014] In the implementation process of the first data processing, for the real-time power generation data, a time response interpolation algorithm is introduced. Based on the concept of relaxation time in physics, the missing values ​​are calculated using the known values ​​before and after the missing data.

[0015] Preferably, the S1 specifically includes:

[0016] The second data processing is performed on the cleaned and formatted power generation data set and environmental data set, specifically including: expanding the cleaned and formatted power generation data set and environmental data set; and integrating the power generation data and environmental data from different sources to obtain the power generation data set and environmental data set.

[0017] Preferably, the S2 specifically includes:

[0018] The environmental data set is analyzed and quantified using the environmental factor photovoltaic impact quantification algorithm to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation.

[0019] Preferably, the S2 specifically includes:

[0020] In the quantification algorithm of photovoltaic impact of environmental factors, the category correlation matrix is ​​introduced.

[0021] A photovoltaic power generation power prediction system, applied to the above photovoltaic power generation power prediction method, comprises the following parts:

[0022] Data acquisition module, environmental monitoring module, first data processing module, second data processing module, feature engineering module, environmental quantitative feedback module, intelligent prediction module;

[0023] A data acquisition module collects real-time power generation data of photovoltaic power generation in real time and sends the real-time power generation data to the first data processing module;

[0024] An environmental monitoring module, which captures real-time environmental data of photovoltaic power generation and sends the real-time environmental data to the first data processing module;

[0025] The first data processing module performs data cleaning, outlier detection, standardization, and formatting on the real-time power generation data and the real-time environmental data to obtain cleaned and formatted power generation data sets and environmental data sets, and sends the cleaned and formatted power generation data sets and environmental data sets to the second data processing module;

[0026] The second data processing module processes the cleaned and formatted power generation data set and environmental data set to obtain a power generation data set and an environmental data set, sends the power generation data set to the feature engineering module, and sends the environmental data set to the environmental quantitative feedback module;

[0027] The feature engineering module extracts and optimizes the features of the power generation data set to obtain an optimized feature set, and sends the optimized feature set to the intelligent prediction module;

[0028] The environmental quantitative feedback module analyzes and quantifies the environmental data set to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation, and sends the quantitative feedback to the intelligent prediction module;

[0029] The intelligent prediction module predicts the photovoltaic power generation based on the optimized feature set to obtain the prediction result, and optimizes the photovoltaic power generation prediction based on the quantitative feedback to obtain the optimized prediction result.

[0030] The beneficial effects of the technical solution of the present invention are:

[0031] 1. The present invention can significantly reduce the errors in the data and improve the accuracy of the data by applying a filtering algorithm to remove random noise in real-time environmental data and real-time power generation data. In addition, the continuous volatility detection algorithm can effectively identify and process abnormal points in the data, thereby ensuring the reliability and stability of the data set; the time response interpolation algorithm can effectively solve the problem of missing data, fill in the missing data points, and maintain the time continuity and consistency of the data, especially in the case of continuous missing values ​​caused by equipment failure or data transmission problems; through the application of the continuous volatility detection algorithm and the time response interpolation algorithm, especially when processing real-time data, the ability to capture the timing characteristics and time dependence of the data is enhanced, and the accuracy of photovoltaic power generation prediction is increased.

[0032] 2. The present invention can capture more subtle data patterns and associations through feature engineering techniques, such as feature derivation and the creation of interactive features, thereby improving the model's explanatory power and prediction accuracy for power generation data; using quantitative algorithms to clearly evaluate the specific impact of various environmental factors on power generation efficiency, providing a basis for real-time and dynamic adjustment of the model, and further improving the accuracy of predictions; through real-time environmental quantitative feedback, the model can dynamically adjust the weights of relevant features, making the prediction model more adaptable to environmental changes and improving the stability and accuracy of the model under different environmental conditions; by calculating and utilizing the category correlation matrix, the model can consider the interactions between different environmental factors, providing a more comprehensive perspective to understand and predict the impact of environmental variables on power generation performance; through the comprehensive use of optimized feature sets and quantitative feedback, the optimized prediction results obtained provide more accurate and reliable data support for photovoltaic power generation management. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a structural diagram of the photovoltaic power generation prediction system of the present invention;

[0034] Figure 2 This is a flow chart of the photovoltaic power prediction method described in the present invention. DETAILED DESCRIPTION

[0035] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0037] The specific scheme of a photovoltaic power prediction method and system provided by the present invention is described in detail below with reference to the accompanying drawings.

[0038] See attached Figure 1 , which shows a structure diagram of a photovoltaic power generation prediction system provided by an embodiment of the present invention, the system includes the following parts:

[0039] Data acquisition module, environmental monitoring module, first data processing module, second data processing module, feature engineering module, environmental quantitative feedback module, intelligent prediction module;

[0040] A data acquisition module collects real-time power generation data of photovoltaic power generation in real time and sends the real-time power generation data to the first data processing module;

[0041] The real-time power generation data includes equipment operation status data, such as equipment operation data, fault records and maintenance logs;

[0042] An environmental monitoring module, which captures real-time environmental data of photovoltaic power generation and sends the real-time environmental data to the first data processing module;

[0043] The real-time environmental data include temperature, humidity, wind speed, sunshine duration, rainfall, and cloud cover;

[0044] The first data processing module performs data cleaning (such as removing noise and filling missing values), outlier detection, standardization, and formatting on the real-time power generation data and the real-time environmental data to obtain cleaned and formatted power generation data sets and environmental data sets, and sends the cleaned and formatted power generation data sets and environmental data sets to the second data processing module;

[0045] The second data processing module accurately processes the cleaned and formatted power generation data set and environmental data set to obtain the final power generation data set and environmental data set, and sends the power generation data set to the feature engineering module and the environmental data set to the environmental quantitative feedback module;

[0046] The feature engineering module extracts and optimizes the features of the power generation data set to obtain an optimized feature set, and sends the optimized feature set to the intelligent prediction module;

[0047] The environmental quantitative feedback module analyzes and quantifies the environmental data set to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation, and sends the quantitative feedback to the intelligent prediction module;

[0048] The intelligent prediction module predicts the short-term or long-term photovoltaic power generation based on the optimized feature set constructed by the feature engineering module to obtain the prediction results. At the same time, it optimizes the photovoltaic power generation prediction based on quantitative feedback to obtain the optimized prediction results and realize the photovoltaic power generation prediction.

[0049] See attached Figure 2 , which shows a flow chart of a photovoltaic power prediction method provided by an embodiment of the present invention, the method comprising the following steps:

[0050] S1. Collecting real-time power generation data and real-time environmental data of photovoltaic power generation, and performing first data processing and second data processing on the real-time power generation data and real-time environmental data to obtain a power generation data set and an environmental data set;

[0051] According to the expert experience method and the demand of photovoltaic power generation, data acquisition equipment is selected to collect the real-time power generation data and real-time environmental data of photovoltaic power generation, so as to obtain the real-time power generation data and real-time environmental data of photovoltaic power generation; the real-time power generation data and real-time environmental data of photovoltaic power generation are subjected to the first data processing, and the first data processing specifically includes: applying the filtering algorithm to remove the random noise of the real-time environmental data and the real-time power generation data, and then selecting the filling strategy according to the type and importance of the data to achieve the missing value filling, such as using the method of filling the previous and next values, the mean, and the median; then applying the statistical test (such as the box plot analysis or the Z-score method) to identify the outliers, and correcting or deleting the detected outliers based on the degree of anomaly and the business impact; further using the standardization processing method to perform standardization processing, and finally formatting the standardized real-time environmental data and real-time power generation data to obtain the cleaned and formatted power generation data set and environmental data set.

[0052] In order to prevent the real-time environment data from being affected by external factors, causing sudden jumps or data loss, which would lead to reduced data accuracy, a continuous volatility detection algorithm is introduced to detect abnormal data; the continuous volatility detection algorithm identifies anomalies by finely processing and identifying the periodic and time-dependent volatility features in the data. The specific implementation process is as follows:

[0053] Assume x(t) is the real-time environmental data at time t, W is the sliding window size, calculate the mean μ and standard deviation σ in the window, and based on the mean and standard deviation, introduce the autocorrelation and weighted moving average of the time series to improve the ability to capture time dependence. The specific implementation formula is as follows:

[0054]

[0055] Where D(t) represents the anomaly detection metric; It is a weighted moving average function, which considers the time-dependent weighted average and uses different weights to reflect the importance of different historical data points. It is calculated based on the data of the past W time points. The weight distribution can be determined according to the expert experience method. Represents a vector containing a data sequence from time tW to time t; It is a weighted function based on standard deviation, which is used to measure the variability of data within the time window W. It is calculated from the standard deviation of the past W time points to adjust the sensitivity to the fluctuation of historical data. The weight distribution can be determined according to the expert experience method; represents a vector containing the standard deviation sequence from time tW to time t; α is the adjustment parameter, which adjusts the sensitivity of deviation from the mean; β is the attenuation coefficient, which controls the exponential decay rate and reflects the reaction speed to abnormal fluctuations; ∈ is a small constant to ensure the stability of the calculation and avoid the denominator being zero; δ is the adjustment coefficient, which adjusts the amplitude of the sine wave part;

[0056] If the value of D(t) is greater than the threshold λ preset based on expert experience, x(t) is considered to be an outlier. For the detected outlier, it is decided based on expert experience whether to delete the outlier data point or replace it with the average of the previous and next data points.

[0057] For real-time power generation data, continuous missing values ​​may occur due to equipment failure or data transmission problems. In order to accurately fill the missing values, a time response interpolation algorithm is introduced; the time response interpolation algorithm is based on the concept of relaxation time in physics and uses the known values ​​before and after the missing data to calculate the missing values. The specific calculation formula is as follows:

[0058]

[0059] in, represents the interpolated power generation data at time t; and Represent the known power generation data at time t-1 and t+1 respectively; is the order of the fractional derivative, between 0 and 1, describing the strength of memory, that is, the degree of influence of the past state on the current state; is the time constant, describing the rate of response, A smaller value means a faster response to external changes, while A larger value means a slower response to external changes; is the Mittag-Leffler function used to describe non-integer time dynamic behavior, and in this application it is used to model the time dependence of power generation data, especially in missing data interpolation.

[0060] The second data processing is performed on the cleaned and formatted power generation data set and environmental data set. The specific implementation process is as follows:

[0061] The cleaned and formatted power generation data set and environmental data set are expanded and processed using existing data expansion technology based on the periodicity and trend characteristics of the data to obtain the expanded power generation data set and environmental data set. The existing data fusion technology is then used to integrate the power generation data and environmental data from different sources, and the complete power generation data set and environmental data set are obtained. For example, the weather station data is combined with the local sensor data to provide a more comprehensive environmental view, and finally the power generation data set and environmental data set are obtained.

[0062] S2. Extract features and optimize the power generation data set to obtain an optimized feature set; analyze and quantify the environmental data set to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation; predict photovoltaic power generation based on the optimized feature set, and then optimize the prediction of photovoltaic power generation based on the quantitative feedback to obtain an optimized prediction result.

[0063] The power generation data set is feature extracted and optimized to obtain an optimized feature set. The specific implementation process is as follows: first, statistical analysis and visualization tools are used to identify key variables and potential data characteristics in the power generation data set, including data distribution, correlation analysis, and potential information gain. Feature engineering technology is then used to extract features to obtain a preliminary feature set. The existing feature derivation technology is used to design composite features and interactive features for the preliminary feature set to obtain a newly constructed feature set. Feature conversion is performed on the newly constructed feature set to obtain a converted feature set. Finally, automatic feature selection technology is used to perform feature screening optimization to obtain an optimized feature set.

[0064] The environmental data set is analyzed and quantified using the environmental factor photovoltaic impact quantification algorithm to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation; the specific implementation process includes: first, a weight vector is assigned to each environmental category according to the expert experience method, and the weight reflects the importance of each category on power generation; the environmental categories are such as temperature, humidity, wind speed, sunshine duration, rainfall and cloud cover; at the same time, a category correlation matrix is ​​introduced to describe the interaction between environmental categories; the environmental impact score is further calculated, and each environmental category data is nonlinearly transformed according to the expert experience method to obtain an environmental impact score vector, and then the category correlation matrix is ​​applied to adjust the environmental impact score vector by multiplying it with the environmental impact score vector; finally, the weight vector of each environmental category is multiplied and added with the adjusted environmental impact score vector to obtain the final environmental impact score as quantitative feedback on the impact of environmental data on photovoltaic power generation.

[0065] Any element in the category correlation matrix can be expressed as:

[0066]

[0067]

[0068]

[0069] Among them, a ij is any element in the category correlation matrix; τ is a tuning parameter used to adjust the smoothness of the exponential function; n is the number of environmental categories; r ij is the correlation between the i-th and j-th environmental categories; m is the total number of data in each category; x′ ik,x′ jk represents the normalized k-th environmental data value of the i-th and j-th environmental categories; x ik represents the kth environmental data value of the ith environmental category; max(x i ),min(x i ) represent the maximum and minimum values ​​of the ith environmental category, respectively.

[0070] The photovoltaic power generation is predicted based on the optimized feature set, and then the prediction of the photovoltaic power generation is optimized based on the quantitative feedback to obtain the optimized prediction result, thereby realizing the photovoltaic power generation prediction.

[0071] First, according to the expert experience method, combined with data characteristics and business needs, a pre-trained deep learning model suitable for time series prediction is selected, and model parameters such as learning rate, batch size, and training cycle are configured; the optimized feature set is processed by the pre-trained deep learning model to obtain the prediction result, and then the photovoltaic power generation prediction is optimized based on the quantitative feedback, specifically including: adjusting the weight of the corresponding environmental factors in the pre-trained deep learning model according to the degree of influence obtained from the environmental quantitative feedback. For example, if the analysis shows that the impact of temperature changes on power generation is greater than expected, the weight of temperature-related features can be increased; at the same time, based on the environmental quantitative feedback, the feature set is optimized, and the model parameters are adjusted using the grid search technology to optimize the pre-trained deep learning model. At the same time, based on the environmental quantitative feedback, the optimized feature set is upgraded using the existing data mining and statistical analysis technology to obtain a complete feature set containing environmental data, and the complete feature set is processed using the optimized pre-trained deep learning model to obtain the optimized prediction result, realizing the photovoltaic power generation prediction.

[0072] In summary, a photovoltaic power generation power prediction method and system are completed.

[0073] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0074] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0075] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A photovoltaic power generation power prediction method, characterized in that: The following steps are involved: S1. Collect real-time power generation data and real-time environmental data of photovoltaic power generation, and perform first data processing and second data processing on the real-time power generation data and real-time environmental data to obtain a power generation data set and an environmental data set; the real-time power generation data includes equipment operation data, fault records and maintenance logs; the real-time environmental data includes temperature, humidity, wind speed, sunshine duration, rainfall, and cloud cover; S2. Perform feature extraction and optimization processing on the power generation data set to obtain an optimized feature set; perform analysis and quantitative processing on the environmental data set to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation; select a pre-trained deep learning model based on data characteristics and business needs, configure model parameters, and process the optimized feature set through the pre-trained deep learning model to predict photovoltaic power generation; based on the degree of influence obtained from the quantitative feedback, adjust the weights of the corresponding environmental factors in the pre-trained deep learning model, optimize the feature set based on the quantitative feedback of the environment, adjust the model parameters using grid search technology, optimize the pre-trained deep learning model, and at the same time, use data mining and statistical analysis technology to upgrade the optimized feature set of photovoltaic power generation based on the quantitative feedback of the environment to obtain a complete feature set containing environmental data, and process the complete feature set using the optimized pre-trained deep learning model to obtain optimized prediction results.

2. A photovoltaic power generation prediction method according to claim 1, characterized in that: The S1 specifically includes: The first data processing is performed on the real-time power generation data and real-time environmental data of photovoltaic power generation, specifically including: first removing the random noise of the real-time environmental data and the real-time power generation data; then filling the missing values; then identifying the outliers, and correcting or deleting the detected outliers; further standardizing and formatting are performed to obtain the cleaned and formatted power generation data set and environmental data set.

3. A photovoltaic power generation prediction method according to claim 2, characterized in that: The S1 specifically includes: In the implementation process of the first data processing, a continuous volatility detection algorithm is introduced to detect abnormal data. The specific implementation formula is as follows: Where x(t) is the real-time environment data at time t; W is the sliding window size; D(t) represents the anomaly detection metric value; is the weighted moving average function; represents a vector containing the data sequence from time tW to time t; is a weighting function based on the standard deviation; represents a vector containing the standard deviation sequence from time tW to time t; α is a tuning parameter that adjusts the sensitivity to deviations from the mean; β is the attenuation coefficient; ∈ is a small constant to avoid the denominator being zero; δ is a tuning coefficient that adjusts the amplitude of the sinusoidal wave part.

4. A photovoltaic power generation prediction method according to claim 2, characterized in that: The S1 specifically includes: In the implementation process of the first data processing, for real-time power generation data, a time response interpolation algorithm is introduced. Based on the concept of relaxation time in physics, the missing values ​​are calculated using the known values ​​before and after the missing data. The specific calculation formula is as follows: in, represents the interpolated power generation data at time t; and Represent the known power generation data at time t-1 and t+1 respectively; is the order of the fractional derivative; is the time constant describing the response rate; is the Mittag-Leffler function.

5. A photovoltaic power generation prediction method according to claim 2, characterized in that: The S1 specifically includes: The second data processing is performed on the cleaned and formatted power generation data set and environmental data set, specifically including: expanding the cleaned and formatted power generation data set and environmental data set; and integrating the power generation data and environmental data from different sources to obtain the power generation data set and environmental data set.

6. A photovoltaic power generation power prediction method according to claim 1, characterized in that: The S2 specifically includes: The environmental data set is analyzed and quantified using an environmental factor photovoltaic impact quantification algorithm to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation. The specific implementation process includes: first, a weight vector is assigned to each environmental category, and the environmental categories include temperature, humidity, wind speed, sunshine duration, rainfall and cloud cover; at the same time, a category correlation matrix is ​​introduced to describe the interaction between environmental categories; the environmental impact score is further calculated, and a nonlinear transformation is performed on the data of each environmental category to obtain an environmental impact score vector, and then the category correlation matrix is ​​applied to adjust the environmental impact score vector by multiplying it with the environmental impact score vector; finally, the weight vector of each environmental category is multiplied and added with the adjusted environmental impact score vector to obtain the final environmental impact score, which serves as quantitative feedback on the impact of environmental data on photovoltaic power generation.

7. A photovoltaic power generation prediction system, applied to the photovoltaic power generation prediction method as claimed in claim 1, characterized in that: Includes the following parts: Data acquisition module, environmental monitoring module, first data processing module, second data processing module, feature engineering module, environmental quantitative feedback module, intelligent prediction module; A data acquisition module collects real-time power generation data of photovoltaic power generation in real time and sends the real-time power generation data to the first data processing module; An environmental monitoring module, which captures real-time environmental data of photovoltaic power generation and sends the real-time environmental data to the first data processing module; The first data processing module performs data cleaning, outlier detection, standardization, and formatting on the real-time power generation data and the real-time environmental data to obtain cleaned and formatted power generation data sets and environmental data sets, and sends the cleaned and formatted power generation data sets and environmental data sets to the second data processing module; The second data processing module processes the cleaned and formatted power generation data set and environmental data set to obtain a power generation data set and an environmental data set, sends the power generation data set to the feature engineering module, and sends the environmental data set to the environmental quantitative feedback module; The feature engineering module extracts and optimizes the features of the power generation data set to obtain an optimized feature set, and sends the optimized feature set to the intelligent prediction module; The environmental quantitative feedback module analyzes and quantifies the environmental data set to obtain quantitative feedback on the impact of environmental data on photovoltaic power generation, and sends the quantitative feedback to the intelligent prediction module; The intelligent prediction module predicts the photovoltaic power generation based on the optimized feature set to obtain the prediction result, and optimizes the photovoltaic power generation prediction based on the quantitative feedback to obtain the optimized prediction result.

Images