It is not completely clear for me what do you mean by "randomly specific abnormal values", "random normal values", and " another case", but I guess you want something like this:
Suppose the probabilities of the different cases are P1 and P2 (and P3=1-P1-P2).
x = rand(1,noNodes);
faulty = (x<P1);
normal = ((x>=P1) & (x<P1+P2));
fire = (x>=P1+P2);
reading(faulty) = .............. random values with some distribution
reading(normal) = .............. random values with another distribution
reading(fire) = ............ other values
There are other ways (perhaps more clear if you are not familiar with Matlab) to have the same result, but ths is rather compact.
Regards
Z