A neruocontrol methodology is presented for optimal temperature control of
a thermal plant with multiple chambers. The plant, a household refrigerator
, uses a simple cost-effective temperature control approach that provides c
hamber-dependent temperature control with minimized energy consumption at o
nly a single control point. Technological advances in control hardware, inc
luding airflow control devices such as automatic thermal dampen, can add co
ntrol degrees of freedom to achieve optimal control of temperature and ener
gy consumption at all control points. The neurocontrol methodology presente
d uses the generalized learning approach for mapping the plant's inverse dy
namics to the desired control signals. Two unconventional control strategie
s are examined: variable temperature bandwidths, and uncoupled compressor a
nd evaporator fan operation. A plant model, representing the behavior of a
conventional, dual chamber, top mount style refrigerator, was used to gener
ate results for both strategies with manual and automatic thermal damper co
nfigurations. The neural net was trained using plant outputs from various c
ombinations of these plant control configurations and strategies. An optima
l temperature control model was defined by the combined use of variable tem
perature bandwidths and uncoupled compressor and evaporator fan operation w
ith an automatic damper.