An enhanced self-adaptive differential evolution based on simulated annealing for rule extraction and its application in recognizing oil reservoir

This study proposes an enhanced self-adaptive differential evolution based on simulated annealing for rule extraction (ESADE-Rule). ESADE-Rule adopts IF-THEN as the rule’s form, AND as the connection word for the rule’s antecedent, class label as the rule’s consequence. Rules are encoded as individuals in population of ESADE, and each individual represents a rule set consisting of three parts: the rule’s parameters (including the controls of the rule, antecedents and class labels), the control parameters (including scaling factors and crossover rates), and the fitness value. Mutation and crossover operations of ESADE are further carried out on the population. Then, selection operation is conducted by comparing the fitness values, through which the best individual would be selected out and be decoded to obtain the optimal rule set. Besides, ten benchmark datasets and three logging datasets are adopted to test ESADE-Rule’s performance. The three logging datasets contain oilsk81, oilsk83 and oilsk84 that come from Jianghan oilfield for testing recognition accuracy rate of reservoir. ESADE-Rule is compared with four rule extraction methods from the perspectives of recognition accuracy rate, rules’ number, antecedents’ number and samples’ number that are not covered by the rule set. The results prove that ESADE-Rule performs better at recognition accuracy rate and interpretability. With oilsk81 as training data set, oilsk83 and oilsk84 as testing data set, the testing results of recognition accuracy rate of oil reservoirs illustrate that compared with other four rule extraction methods, ESADE-Rule can obtain more general rules set when the attributes of datasets are similar.