Aiming at the problem that the fault diagnosis of photovoltaic array is interfered by harsh environments, and the single model is not effective in extracting effective feature information, which leads to low diagnosis...
详细信息
Aiming at the problem that the fault diagnosis of photovoltaic array is interfered by harsh environments, and the single model is not effective in extracting effective feature information, which leads to low diagnosis accuracy, a fault diagnosis method for photovoltaic arrays based on model fusion is proposed. In this method, a multi-channel one-dimensional convolutional neural network is designed to capture multi-scale information, thereby improving feature representation. LSTM, AdaBoost, PNN, and logistic regression methods were used to construct the stacking model to predict and classify the fault types of photovoltaic arrays. Experimental results show that the accuracy of fault classification of the proposed method reaches 96.4%, which verifies the effectiveness of the proposed method.
This paper takes the time deposit order data of banking institutions as the research object. Firstly, the data of banking institutions are analyzed and cleaned to convert the categorical data into numeric data. Then, ...
详细信息
ISBN:
(纸本)9781450376631
This paper takes the time deposit order data of banking institutions as the research object. Firstly, the data of banking institutions are analyzed and cleaned to convert the categorical data into numeric data. Then, to "split" preprocessing the data after processing, divided into training set and test set. In the training set, LR, Naive Bayes, K-NN, Decision Tree, Random Forests, Extra - trees, AdaBoost, GBDT, XGBoost, LightGBM and CatBoost algorithm (via the grid search method GridSearchCV) are established to study the customer's deposit ordering tendency respectively. Finally, the stacking fusion algorithm in Ensemble Learning(The meta-learner adopts LR algorithm) is used to fuse each single model to build the best bank customer time deposit ordering classification model. The research results show that using python, based on the different dimensions of bank customer information, the stacking fusion algorithm based on Ensemble algorithm has better prediction effect than the single model, and the stacking fusion algorithm has better robustness. Therefore, this paper takes the stackingfusion model as the final prediction model to effectively identify bank customers and help banking institutions understand customer subscription tendency.
暂无评论