How to define the Values Ai and Bi of the coeficients for a cluster using the Matlab function newgrnn to model a system with changing characteristics?

I am using the matlab function newgrnn to generate a neural network, but in my work, I need to process a different type of input data that are collected from the same system. One input vector is a set of three vectors concatenated. the vectors are data obtained by processing signals recorded by three power network monitors. The difference comes from the fact that, The failure of one monitor results in a vector filled with zeros and the two others remain normal. If such a vector is processed, the network outputs efficiency drops very low since it does not fit any of the patterns used for training. As this is a situation of modelling a system with changing characteristics, my reference paper for GRNN work is "A general Regression Neural Networks" by Donald F. Specht. Therein, there is a possibility of using eq. (11) for this type of situation. Under Matlab, how can I implement a GRNN able to take into account the system changing characteristics? Thanks in advance