Intellectual control of the electrolysis process in the production of caustic soda

3Mathematical modeling of the electrolyzer


Input parameters for modeling the operating mode of the electrolyzer: pH value of salt water in the electrolyzer is 2 (pH); current to electrolyzer 40 A/m², temperature of salt water to electrolyzers, 65-70°C (T_ras); NaCl concentration in salt water is 29-29.5%; the pressure between the anode and cathode compartments is 0.003MPa.
Output parameters for modeling the operating mode of the electrolyzer: anolyte temperature 90°C, pressure of moist chlorine gas coming out of the anode compartment of electrolyzers (0.018-0.02) MPa, hydrogen pressure coming out of the cathode compartment (0.021-0.023) MPa.
The required number of neurons in the hidden layers of the perceptron can be determined by the formula that is a result of the Arnold Kolmogorov-Hecht-Nielsen theorems:


(2)

Here, - output signal size, =4, - the required number of synaptic connections; - size of the input signal ( =8); Q - examples of the number of elements of the educational complex (Q=1676)

Using this formula, we calculate the required number of synaptic connections , we can calculate the required number of neurons in the hidden layers. For example, the number of neurons in the hidden layers of a two-layer sensor is:
(3)
Minimum number of neurons:

Maximum number of neurons:

The minimum value of neurons in one hidden layer is selected to train the first neural network - 40. Next, it is necessary to determine the maximum permissible error. The output parameter is continuous and ranges from 0.709 to 1.279. The accuracy of measuring the volume fraction of hydrogen in chlorine gas in the production of caustic soda is 0.01%. It is with this accuracy that the voltage is given in the original data. So there should be no minimum error.

(4)

The error value is set to 0.005, and the number of cycles is 1000.