Search

PurposeThe purpose of this paper is to use artificial intelligence to evaluate the risks of urban power network planning.Design/methodology/approachA fuzzy Bayesian least squares support vector machine (LS_SVM) model is established in this paper, which can learn the risk information of urban power network planning through artificial intelligence and acquire expert knowledge for its risk evaluation. With the advantage of possessing learning analog simulation precision and speed, the proposed model can be effectively applied in conducting a risk evaluation of an urban network planning system. First, fuzzy theory is applied to quantify qualitative risk factors of the planning to determine the fuzzy comprehensive evaluation value of the risk factors. Then, Bayesian evidence framework is utilized in LS_SVM model parameter optimization to automatically adjust the LS_SVM regularization parameters and nuclear parameters to obtain the best parameter values. Based on this, a risk comprehensive evaluation of urban network planning based on artificial intelligence is established.FindingsThe fuzzy Bayesian LS_SVM model established in this paper is an effective artificial intelligence method for risk comprehensive evaluation in urban network planning through empirical study.Originality/valueThe paper breaks new ground in using artificial intelligence to evaluate urban power network planning risks.

Following the fast urban economic development and the improvement of people living standard, urban residential electricity consumption is increasing quickly. Although the influencing factors of urban residential power demand are complex, objective analysis of risk factors of urban residential load will offer a scientific decision base for the urban power planning and Power Demand-side management. In this paper, related factors with resident living power utility of nine typical cites is chosen based on data mining technology, and association rule mining is finished based on R program language. Then, frequent itemsets of dataset associated with resident living power utility is excavated based on backward algorithm programming. Finally, risk factors of urban residential load can be concluded according to frequent itemsets. This method can avoid subjectivity of determining the risk factors of urban residential load, and also can provide a decision foundation for scientific forecasting of urban residential electricity consumption.