Approaches On Social Attribute Network Based Evaluation System For Cable Engineering Design

With the development of national economy and improvement of urbanization process,the cable engineering construction in our country has been gradually improved.As the source of cable construction,operation and maintenance,the quality of cable engineering design will affect the follow-ups directly or indirectly.The current evaluation problems of cable engineering design at home and abroad presented the following characteristics,such like high redundancy of evaluation index,promiscuous mode of qualitative and quantitative evaluation index,significant expert subjectivity in evaluation conclusion,low utilization rate of current cable engineering data information and etc.,which have greatly affected cable engineering design quality and urban power supply reliability.Therefore,it is of great practical significance to propose a reasonable evaluation index system,together with accurate comprehensive evaluation methods for cable engineering design,which can meet the requirements of current cable engineering design in China.It can effectively promote the healthy,sustainable and harmonious development of power engineering industry,and promote the popularization and application of advanced information processing technology in electric power engineering.On account of the correlation between cable engineering design quality and operation continuity of cable engineering equipment,the definition and calculation method of equipment selection value index for cable engineering are proposed.The equipment selection value index for cable engineering design is established according to the value engineering principle to present the continuous operation ability of cable engineering equipment in actual projects.Based on low rank matrix completion,an equipment selection value index calculation method is proposed,in which eigenvalue decomposition and operator shrinkage are introduced to fill the incomplete information of relative aging age and incomplete test data of cable engineering equipments of the same type in operation.Consequently,the problems of inconsistent cable engineering equipment testing conditions,incompletion of ind icators and the difficulties in the obtaining of equipment selection value index are solved.It is proved that the proposed equipment selection value index can effectively describe its influence on the cable engineering operation continuity,which turns out to be a useful supplement to the evaluation index system for cable engineering design.Aiming at the lack of optimization methods of the evaluation index system for cable engineering design,the social attribute network theory is proposed to achieve the optimization tasks of evaluation index and the evaluation index system structure for cable engineering design.The attribute measurement space and ordered segmentation class are introduced for attribute describing and correlation mining in the evaluation index system for cable engineering design.Finally,an evaluation index system,which covers the technicality,economy and continuity of cable engineering design,is built.The scientificity and stability of the evaluation index and evaluation index system structure for cable engineering design were verified through the reliability and validity analysis.And in consequence,the evaluation index system can support for the subsequent studies of the qualitative and quantitative comprehensive evaluation model for cable engineering design.In order to remove the significant expert subjectivity in traditional qualitative evaluation index for cable engineering design,a self-adaptive attribute preference learning method is proposed to improve the objectivity in evaluation index weights.This method employs the attribute measure matrix and comprehensive measure matrix to describe the attribute measurement space of evaluation object,and evaluates the targeted cable engineering design by learning historical sample data with random weight network.In addition,to solve the confidence selection problem in attribute preference learning,the confidence dispersion parameter is deduced by generalization error.And finally the adaptive optimization of comprehensive evaluation model is realized by using confidence interval method.The numerical simulation results demonstrated that the self-adaptive attribute preference learning algorithm can achieve comprehensive evaluation tasks of cable engineering designs,and the distribution of evaluation results turns out to be more discrete than those of traditional comprehensive evaluation methods,which provides great help to the decision makers for making reasonable judgments.Effective information extraction in the quantitative evaluation index data for the cable engineering design is quite difficult.In order to solve this problem,a feature extraction method based on random weight deep neural learning is proposed.In this method,a multi-hidden layer neural network model with gradually reduced neurons from outside to inside is established,then with the output weights and thresholds of the hidden layers randomly selected,the effective data characteristics of evaluation index for cable engineering design is deeply studied.Based on th ese data characteristics for cable engineering design extracted,a sparse random forest based comprehensive evaluation model is established by constructing strong decision tree with redundant decision tree constructed via multiple weak decision trees,and introducing sparse coding technique.The numerical simulation experiments demonstrated that the random weight deep neural learning algorithm can realize the multivariate statistics of the data information,and be superior to the principal component analysis and independent component analysis in feature extraction prerformances.It is also demonstrated that the sparse random forest algorithm is effective for the realization of the comprehensive evaluation tasks for cable engineering design,and the accuracy and robustness are satisfying for engineering requirement.