基于字典学习算法的调度运行信息稀疏编码方法

A Sparse Coding Method for Power System Operation Information Based on Dictionary Learning Algorithm

在电力系统中,随着电网规模的不断扩大,运行信息的数据规模与复杂程度不断增大,对其编码效率提出了更高的要求。字典学习算法是当前应用较为广泛的人工智能算法,其优势在于能够给出一个标准的基础信息库,使得依据该库所得到运行信息表示形式相对稀疏。首先介绍了字典学习算例原理与实施方法;结合电力系统调度运行控制实际,提出了基于字典学习的调度运行信息稀疏编码方法。最后基于某电网实际,利用近二年的运行记录信息构建了调度运行字典。算例表明利于该字典所编码表示的调度运行记录稀疏度处于0.35~0.80之间,符合大数据处理的稀疏性要求。

In power system, the data size and complexity of the operation information are increasing quickly with the continuous expansion of power grid, which puts forward higher requirements for sparse coding efficiency. Dictionary learning algorithm is a widely used artificial intelligence algorithm at present. Its advantage is that it can provide a standard basic information base to present a relatively sparse representation for operation information based on the standard basic information base. The principle and implementation of dictionary learning algorithm was firstly introduced. Then a specific sparse coding method for dispatching operation information based on dictionary learning was proposed. Finally, based on the actual situation of a power grid, the dispatching operation dictionary was constructed with the operation record information of the last two years. The case study shows that the sparsity of the scheduling operation record encoded by the dictionary is between 0.35 and 0.80, which meets the sparsity requirement of big data processing.