微型电脑应用

受到群体用电行为多样性的影响，不同属性行为之间的差异化严重，群体用电行为特征划分难度大，预测结果误差偏大。基于上述原因，构建群体用电行为双变量多层贝叶斯预测模型。通过引入双变量概念，分析群体用电行为多样性，进行群体用电行为特征划分。构建多层贝叶斯预测模型，利用双变量多层贝叶斯计算用电行为特征先验分布，推导输出模型化预测的分布参量。实验结果表明，在4种难度指标上，模型识别率均能保持在87%以上，预测响应时间变化曲线趋势平稳，预测精准度、响应速度、识别能力均满足群体用电行为的预测要求，具有较高的应用与推广价值。

Affected by the diversity of group electricity behaviors, the differentiation between different attribute behaviors is severe, and it is difficult to classify the diversity characteristics of group electricity consumption behaviors, resulting in a large prediction error. Based on the above reasons, this paper plans to construct a bivariate multi-layer Bayesian prediction model for group electricity consumption behaviors. By introducing the concept of bivariate, analyzing the diversity of group electricity consumption behaviors and dividing the characteristics of group electricity consumption behaviors, this paper constructs a multi-layer Bayesian prediction model, uses bivariate multi-layer Bayesian model to calculate the prior distribution of electricity consumption behaviors characteristics, and derives the distribution parameters for output modeled predictions. The experimental results show that the recognition rate can be maintained at more than 87% on the four difficulty indicators, the trend of the prediction response time change curve is stable, the prediction accuracy, response speed, and recognition ability meet the prediction requirements of group electricity consumption behaviors, which has high application and promotion value.