The work aims to study the way of query in a distributed network real-time database for complex packaging product information with continuous state space. Combined with the enhanced learning (EL) and the adaptive response structure (ARS) neural network, the ELARS2 algorithm, ARS neural network based on enhanced learning was given. The way of selection and assessment methods for EL were introduced in ARS2 algorithm, so as to solve the query problems of sort mode in ARS2 algorithm. The simulation test in which the distributed network real-time database in the memory space was used to inquire targets was designed and it was achieved with two ELARS2 algorithms (TDARS2 and QARS2). Such algorithms were compared with the classical EL algorithm. The average time to complete target query in two ELARS2 algorithms was significantly less than the classic EL algorithm. In two ELARS2 algorithms, the effect of TDARS2 is better than QARS2.