مشخصات پژوهش

Optimum Control by Artificial Intelligence (OCAI) is presented in this paper as a dynamic decision making algorithm for optimising pumpage schedule to remediate a contaminated aquifer using the Pump, Treat and Inject (PTI) method. OCAI integrates three modules to control contaminants, to reduce runtime and to meet water quality constraints and discharge capacity at the wells. There is no bespoke capability for the strategy presented by the paper, which formulates: (i) Module 1 comprises models of physics-based flow and transport for simulating heads and contamination; (ii) Module 2 serves as the “surrogate” of Module 1 by transforming the simulation outputs of Module 1 into two fast forecasting Sugeno Fuzzy Logic (SFL) models; and (iii) Module 3 is a user-defined unit to implement OCAI, to run Genetic Algorithm (GA) and to interrogate Module 2, where Modules 2 and 1 are pre-processed. The OCAI strategy resolves two barriers: (i) the information created in the past time step is passed on to new time step for an efficient control; and (ii) the ‘hunger’ of GA for function evaluation is met by the fast Module 2 but not by the slow Module 1. The novelty in OCAI includes: an optimum PTI schedule to control contaminants and to remediate contaminated plumes. The results show that the maximum Total Dissolved Matter is reduced from a range of 3500 to 8000 to a range from 1490 to 3450. The results provide “proof-of-concept” for OCAI.