A three neuron controller (TNC). IV. Shifting to biological problems

A TNC is a two-layered artificial neural network (ANN) with two neurons on the input or lower level and one neuron on the output or upper level. The difference between the actual input value and its neurons' resting value is mapped onto the range of [-1, 1] by normalizing this value by the difference between the resting value and the neurons' maximum (or minimum) value. The activation of the output neuron is calculated by integrating the two normalized input values onto the range (0, 1). The appropriate control signal is calculated from the value obtained by the integration by first subtracting from this value the output neuron's resting value, and then multiplying this difference by the maximum allowed control value. The weights connecting the neurons may be positive or negative. We demonstrate the robustness of TNCs by applying them to control the water level of a tank, backing of a truck and trailer, and the inverted pendulum problem.