Skip to Main Content
In this paper artificial immune recognition system (AIRS) is employed as an emerging technique of data mining to extract the water reservoir operating rules with a case of water supply reservoir, and aiming to explore the impact of learning performances of the AIRS on the operating rule derivation. Based on the performances, the recognition abilities of the AIRS subject to different gene representatives of antibody or antigen in the ARIS are analyzed firstly, and 83.3% classification accuracy confirms the AIRS is capable of extracting the operating rules. Secondly the subjectivity and uncertainty of annual hydrological condition (AHC), one of the attributes of the operating data, determined merely by the frequency of the annual runoff, are discussed and therefore information entropy theory is adopted to take traditional reservoir operating decision-making into account to determine the AHC in order to improve the AIRS capability of extracting the operating rules, and the identification result of 86.1% shows AIRS based on information entropy (IEAIRS) can tackle the subjectivity and uncertainty of the AHC . Finally so as to further illuminate the AIRS learning capabilities, the rules extracted by the AIRS and IE ARIS are compared with those by the RBF networks, which indicates AIRS can be good for mining the reservoir operating rules which are of more transparent and interpretive, and dynamically update the operating rules in the memory set.