Optimization of Extraction Process Based on Neural Network

Abstract

Liquid-liquid extraction is a chemical unit operation that utilizes the difference in solubility or distribution ratio of target components in two immiscible solvents to achieve separation, extraction or purification. There are many factors that affect the extraction efficiency, and it is difficult to quickly optimize the process using traditional methods. Artificial neural network is a system structure composed of multiple artificial neuron models, with functions such as self-learning, associative storage and fault tolerance. It can be used for optimization or control of multi-variable complex systems, and has been successfully applied to the extraction process of various products. optimization. This paper discusses the basic situation of artificial neural network, and analyzes the research progress of extraction process optimization based on neural network.

Advances in Neural Networks for Pharmaceutical Applications

“…In 2016, the advent of AlphaGo brought research enthusiasm for deep learning to a new height. Now ANN has been widely applied in various fields, such as image recognition [6][7][8], wireless signal processing [9][10][11][12], chemical process control and optimization [13][14][15][16], forecasting [17,18], security risk assessment [19], traditional Chinese medicine processing [20][21][22], aquatic products [23], intelligent driving [24,25], and so on.…”

Application of Artificial Neural Networks in Chemical Process Control

“…Information is passed as indicated by the arrows in the directed diagram, so the data flows in one direction. The most important feature of this neural network is that it can learn and store a large number of highly nonlinear mappings without building a mathematical equation to describe the mapping relationships in advance, exhibiting excellent nonlinear matching and generalization capabilities [47,48]. The structure of MLP consists of input layers, hidden layers, and output layers.…”

Comparison of some artificial neural networks for graduate students

“…The Back Propagation neural network is one of the most widely used neural networks at present, the BP used in finance, accounting, management science, e-commerce and computer science, as it is also called multilayer. was proposed by Rumelhart, Hinton and Williams [12] BP neural network is a typical multi-layer forward network, the work of the network is to enter information from the input layer, then process it in the hidden layer, and then the output layer. If the final output is not reached, the network weight is modified, it uses forward propagation of information and backward propagation of errors to adjust communication weights between cells to reduce error [13] It can be explained by the following steps:…”