Table 3. Step-by-Step Pseudocode for the Proposed PSO-ANN Algorithm for Thermal Maturity Parameter Prediction.

steps	working
1	start
2	set input variables
3	initialize parameters of ANN such as learning rate, activation functions, etc.
4	vary the number of hidden layers (sensitivity of hidden layers, 1–3)
5	vary the number of neurons in the hidden layer (sensitivity of neurons, 5–30)
5	select the learning algorithm of ANN
6	select the learning rate [0, 1] for the selected learning algorithm
7	train and test the ANN model and
8	evaluate the objective function for a minimum convergence value
9	extract weights and biases from the trained model
10	initialize parameters of PSO algorithm such as the number of iterations, population of particles, cognitive and social accelerations, and initial and final inertia weights
11	set range for sample search space of each extracted weights and biases
12	feed extracted weights and biases in a PSO algorithm as the initial population
13	evaluate the objective function for a minimum convergence value
14	run the iterative process until the stopping criteriona is achieved
15	pick the global best solution
16	set optimum weights and biases from the globally best model in the network for the prediction of thermal maturity parameters
17	end

^astopping criterion = a maximum number of iterations are attained or a maximum level of inactivity is reached.