Desulfurization efficiency prediction method based on time delay neural network and support vector machine

Abstract

The invention belongs to the technical field of desulfurization, and particularly discloses a desulfurization efficiency prediction method based on a time delay neural network and a support vector machine. Comprising the steps of collecting operation data and desulfurization efficiency data of a desulfurization system, constructing a primary desulfurization efficiency prediction model based on a neural network structure, adding a hidden layer of the primary desulfurization efficiency prediction model to construct the desulfurization efficiency prediction model, and calculating an accumulated error of a back propagation error of the desulfurization efficiency prediction model. Transmitting the accumulative error to a multi-classification support vector machine, carrying out accumulative error training, taking the optimal accumulative error as a descending strategy of the desulfurization efficiency prediction model, carrying out convergence calculation, and timely adjusting the weight and the threshold value of the desulfurization efficiency prediction model, so that the prediction precision of the desulfurization efficiency prediction model meets the requirement. And obtaining an optimal desulfurization efficiency prediction model. According to the method, the prediction precision can be greatly improved, and the method has better robustness, memory ability, nonlinear mapping ability and strong self-learning ability.

Description

technical field [0001] The invention belongs to the technical field of desulfurization, and more specifically relates to a method for predicting desulfurization efficiency based on a time-delay neural network and a support vector machine. Background technique [0002] Although the limestone-gypsum wet flue gas desulfurization technology is mature, it has great advantages in desulfurization efficiency, operation stability and by-product treatment. However, its process is complicated, and it is easy to block pipelines and equipment, and the energy and material consumption in daily operation is relatively expensive. [0003] Existing wet desulfurization systems are difficult to accurately control energy consumption, which can easily lead to increased electricity consumption, limestone consumption and SO 2 The outlet concentration fluctuates violently, mainly due to the large delay, large inertia and nonlinearity of the desulfurization efficiency of the traditional wet desulfur...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the above defects or improvement needs of the prior art, the present invention provides a desulfurization efficiency prediction method based on time-delay neural network and support vector machine, in which the characteristics of the desulfurization system itself and the characteristics of the desulfurization efficiency prediction process are designed accordingly. Desulfurization efficiency prediction method based on time-delayed neural network and support vector machine, in which, the time-delayed neural network including hidden layer is used to construct the desulfurization efficiency prediction model, and the cumulative error of each neuron in the hidden layer is calculated, and the support vector machine is used to predict the desulfurization efficiency. The cumulative error is used for iterative training to seek the optimal solution, thus overcoming the problem of inaccurate results or difficult convergence caused by directly adjusting the parameters through its own threshold in the calculation process of the existing neural network, thus greatly improving the prediction accuracy and having a more Good robustness, memory ability, nonlinear mapping ability and powerful self-learning ability, the system prediction error is less than 1%

Images