Friends who missed the last live broadcast can use this article to learn more about AI to solve the problem of electricity stealing and leakage in the power industry and its solutions, and this article adds detailed data and model explanations in actual projects, which will definitely help you gain something.
The overall modeling process based on the scene is as follows:
01 Modeling goals
Through automated machine learning (AutoML), the easy-to-use and efficient modeling process of abnormal power scenarios is completed, and the artificial intelligence algorithm platform is used to realize model data preprocessing and model training.
02 Solutions
Through the technical matrix in the figure below, the modeling technology can be further clarified: that is, a regression model is established through structured data, supervised learning and deep neural network algorithms.
03 Modeling method
04 Data acquisition
04 Research on Impact Factors
According to the actual business judgment, the influencing factors of all stations include: daily electricity energy curve, daily power curve, daily current curve, daily voltage curve, and transformer capacity of the station area.
05 Algorithm use
Deep learning is a method in machine learning based on representational learning of data. By establishing and simulating a neural network for analysis and learning of the human brain, it imitates the mechanism of the human brain to interpret data and realizes an artificial neural network based on machine learning (Artificial Neural Network)
There are many types of neural networks, the most important of which is the multi-layer perceptron deep learning structure: by combining low-level features to form more abstract high-level representation attribute categories or features, to discover the distributed feature representation of data.
The benefit of deep learning is to replace handcrafted features with efficient algorithms for unsupervised or semi-supervised feature learning and hierarchical feature extraction.
06 Model training
Training times: number of samples / training package size * training period
07 AI Platform Architecture
