基于属性偏好学习的配电网综合评价方法

为了摆脱在传统地区配电网评价方法中对参评人员个人评价偏好的过度依赖,实现合理、精准的属性权重确定,提出了一种基于属性偏好学习的配电网多指标智能综合评价方法。依据属性测度理论,在置信度准则与评分准则下完成对配电网综合评价模型的构造。进而,提出数值绝对偏移率指标以实现对中间值指标的数据预处理。最后,应用随机权神经学习,通过对配电网历史训练样本进行有监督学习,计算得到指标属性偏好权重,并依据配电网综合评价模型以及计算所得属性偏好权重完成对配电网待测样本的智能综合评价。与传统的AHP、PSO-SVM以及RWN算法的对比仿真实验验证了该方法的精确性与稳定性,表明该方法实现了合理、客观的配电网综合评价,在地区配电网评价方面具有一定的应用价值。

Attribute Preference Learning Based Comprehensive Evaluation Method for Distribution Network

To avoid the over reliance of personal preferences on traditional regional distribution network evaluation and achieve reasonable, accurate attribute weights, a multi-index attribute preference learning based intelligent comprehensive evaluating method for distribution network is proposed in this paper. In the proposed method, the distribution network comprehensive evaluation model is established under confidence criterion and score criterion, according to the attribute measure theory. Then, the intermediate value indexes are pre-processed by introducing numerical absolute deviation rate. Finally, based on the historical training samples of distribution network, the preference weights of indexes are calculated by employing supervised random weighted neural network learning model. The intelligent evaluation of test samples is performed by using the distribution network comprehensive evaluation model and well-trained attribute preference weights. Compared with traditional AHP, PSO-SVM and RWN algorithms, the experimental result analysis demonstrates that the proposed method is feasible, effective and robust, which can achieve a reasonable and objective comprehensive evaluation of target distribution network, and has certain application value on regional distribution network evaluation.