Neural network prediction method based on boiler combustion characteristics

A boiler combustion and neural network technology, applied in the field of neural network prediction based on boiler combustion characteristics, can solve problems such as too large initial population range, long convergence time of genetic algorithm, etc., to overcome the defects of network initial connection weight and threshold, and guarantee Group Diversity, Guaranteed Convergence Effects

Active Publication Date: 2022-05-10
武汉博赛环保能源科技有限公司
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to the use of a genetic algorithm for optimization in the above method, there is a problem that the initial population range is too large, resulting in a long time for the convergence of the genetic algorithm.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Neural network prediction method based on boiler combustion characteristics
  • Neural network prediction method based on boiler combustion characteristics

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0039] At present, the BP neural network is mostly used to establish the boiler combustion prediction model, and the output of the BP neural network is comprehensively optimized based on the genetic algorithm to find the best boiler efficiency setting value and the best NO x The emission set point, by adjusting the input of BP neural network, makes the boiler efficiency and NO x Emissions are optimized. However, in the above method, due to the use of a genetic algorithm for optimization, there is a problem that the initial population range is too large, resulting in a long time for the convergence of the genetic algorithm. Therefore, in order to solve the above problems, such as figure 1 and figure 2 As shown, the present embodiment provides a neural network prediction method based on boiler combustion characteristics, which includes the following steps:

[0040] S1. The gas volume and flue gas oxygen content are used as the input of the BP neural network, and the boiler e...

Embodiment 2

[0062] Since the BP neural network algorithm is a method of adjusting connection weights and thresholds based on gradient descent, the initial connection weights and thresholds of the network structure are randomly set before training. Once the initial parameter setting is unreasonable, it will easily fall into In the local minimum defect. Genetic Algorithm (GA) has the ability of global search, and is not easily restricted by local optimum in the search process. Combining the large-scale nonlinear mapping capability of the BP neural network algorithm with the global optimization characteristics of the GA algorithm can overcome the defects of the traditional BP neural network algorithm that randomly generates network initial connection weights and thresholds, and improve the learning rate of the BP neural network algorithm. and linear approximation capabilities. Therefore, on the basis of Embodiment 1, this embodiment provides a genetic algorithm-based optimization of the ini...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

According to the neural network prediction method based on the boiler combustion characteristics, the boiler combustion prediction model is established based on the two-layer genetic algorithm and the BP neural network, the group elite is reserved through the first-layer genetic algorithm, convergence can be ensured, the second-layer genetic algorithm adds a new population to the elite group reserved by the first-layer genetic algorithm, and the prediction accuracy is improved. The population diversity of the genetic algorithm can be ensured, and the population is prevented from being converged on individual minority solutions; in the first-layer genetic algorithm, a global optimal point can be quickly found by taking a larger value of the crossover probability in the earlier stage of iteration; the crossover probability is small in the later stage of iteration, and the situation that convergence cannot be achieved due to the fact that the structure of the found optimal point is damaged can be avoided; the variation probability in the first-layer genetic algorithm adopts a variable probability method, so that the genetic algorithm can be prevented from falling into a local optimal point; the second-layer genetic algorithm selects an individual with the highest fitness and directly carries out decoding operation, crossover and mutation operation is not needed, and it can be guaranteed that the genetic algorithm converges to an extreme point as soon as possible.

Description

technical field [0001] The invention relates to the technical field of automatic control, in particular to a neural network prediction method based on boiler combustion characteristics. Background technique [0002] In order to reduce NO x The low-oxygen combustion method adopted for the emissions, although it can effectively suppress NO x However, it will also cause a sharp increase in CO concentration and fly ash content, and at the same time reduce the combustion efficiency of the boiler; when increasing the oxygen content of the flue gas, although the boiler operating efficiency and furnace temperature are improved, it will also cause NO x At the same time, due to the increase of air volume, the heat loss of exhaust smoke will also increase. Theoretically, there is an optimal point that can balance the contradiction between the two. At present, the BP neural network is mostly used to establish the boiler combustion prediction model, and the output of the BP neural ne...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06F30/27G06N3/04G06N3/08G06N3/12
CPCG06F30/27G06N3/04G06N3/084G06N3/086G06N3/126
Inventor 李超邱丹姚睿龙格姝婷刘炘坤廖先刘莉杨勇徐建局
Owner 武汉博赛环保能源科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products