Fault tolerant liquid measurement system using multiple-model state estimators

In liquid measurement apparatus comprising a plurality of sensors for measuring parameters of a liquid in a container;

• each said sensor generating a time varying measurement signal representative of the liquid parameter measured thereby, a system for detecting a faulty measurement signal comprising:
• means for grouping said sensors into a number of sets, each set including some sensors of another set;
• means for providing current and past measurement samples of each measurement signal;
• and
• processing means for processing said current and past measurement samples of the measurement signals of each set of sensors to determine for each set of sensors a current signal representative of a likelihood that at least one of said measurement signals of the corresponding set is a faulty measurement signal, said processing means for detecting a faulty measurement signal of liquid quantity in said container based on a function of said likelihood signals.

A fault tolerant liquid measurement system comprising:

• a number M of sensors for measuring parameters of a liquid in a container;
• each said sensor generating a time varying measurement signal representative of the liquid parameter measured thereby;
• means for grouping said M sensors into M+1 subsets, one subset including all M sensors and said other M subsets each including all of said M sensors except one and said missing sensor in each of said other M subsets is a different sensor of said M sensors;
• means for providing current and past measurement samples of each measurement signal;
• and
• processing means for processing said current and past samples of the measurement signals of each subset of sensors with a respectively corresponding algorithm based on a state estimation model to determine for each subset of sensors a first estimate signal representative of a likelihood of measurement signal validity for said measurement signals of the corresponding subset, and a current second estimate signal representative of liquid quantity in said container based on said and past samples of the measurement signals of the corresponding subset;
• said processing means for determining a third estimate signal of liquid quantity in said container based on a function of said first and second estimate signals.