Modeling Preparation Condition and Composition–Activity Relationship of Perovskite-Type La_<i>x</i>Sr_1–<i>x</i>Fe_<i>y</i>Co_1–<i>y</i>O₃ Nano Catalyst

In this paper, an artificial neural network (ANN) is first applied to perovskite catalyst design. A series of perovskite-type oxides with the La_<i>x</i>Sr_1–<i>x</i>Fe_<i>y</i>Co_1–<i>y</i>O₃ general formula were prepared with a sol–gel autocombustion method under different preparation conditions. A three-layer perceptron neural network was used for modeling and optimization of the catalytic combustion of toluene. A high <i>R</i>² value was obtained for training and test sets of data: 0.99 and 0.976, respectively. Due to the presence of full active catalysts, there was no necessity to use an optimizer algorithm. The optimum catalysts were La_0.9Sr_0.1Fe_0.5Co_0.5O₃ (<i>T</i>_c = 700 and 800 °C and [citric acid/nitrate] = 0.750), La_0.9Sr_0.1Fe_0.82Co_0.18O₃ (<i>T</i>_c = 700 °C, [citric acid/nitrate] = 0.750), and La_0.8Sr_0.2Fe_0.66Co_0.34O₃ (<i>T</i>_c = 650 °C, [citric acid/nitrate] = 0.525) exhibiting 100% conversion for toluene. More evaluation of the obtained model revealed the relative importance and criticality of preparation parameters of optimum catalysts. The structure, morphology, reducibility, and specific surface area of catalysts were investigated with XRD, SEM, TPR, and BET, respectively