Hong Jun Wang

The application of BP Neural Network based on improved PSO in BF temperature forecast

The BP network has the disadvantages such as low learning efficiency, low speed of convergence, easily falling into the local minimum state, poor ability to adapt, ect. For PSO algorithm, it is fast for convergence, especially at the initial stage, simple for the computing, and is easy to implement. Compared with the genetic algorithms, it does have not the complex operations of hybrid codecs, mutation, so it is a good optimization algorithm. However, PSO algorithm also has some shortcomings it is more and more slow for convergence rate at the late evolution of the algorithm. In this paper, a new BP Neural Network based on improved Particle Swarm Optimization (PSO) is proposed. The convergence speed of this algorithm and the capacity of searching global extremum is increased through adjusting the adaptive capacity of learning factor. The simulation results illustrate that the improved PSO is superior to the standard BP algorithm and particle swarm optimization.

Research and Realization on Image Preprocessing Technology Based on Improved Median Filter

Aiming at the problem that traditional median filter algorithm cannot process collected images quickly and efficiently, this paper adopts improved median filter and makes use of advantages, such as fast running speed, parallel running of inner program, to design an image preprocessing system with high real time ability and high flexibility. At last, compared with MATLAB median filter simulation figure and multilevel median filter, it has shown that using FPGA to process and improve median filter can not only conduct median filter to images successfully, but also has the ability of fast operation speed and low energy consumption.

Power Consumption Prediction Modeling of Cement Manufacturing Based on the Improved Multiple Non-Linear Regression Algorithm

The principal component analysis (PCA) is applied in this paper, since the existing power consumption prediction models of cement manufacturing influenced by many factors are quite complex and have low accuracy. In this way, four new key factors affecting the power consumption of cement manufacturing are obtained instead of the eleven original ones, with the complexity of the computing model simplified. Built upon this is the power consumption prediction model of cement manufacturing based on an improved multiple non-linear regression algorithm. Then the efficiency of the model, obviously improved the forecasting precision, is verified in Pingyi Zhonglian Cement Plant. In other words, a theoretical basis for cement plants power consumption forecasting management is provided in this paper.

Application of Improved Fuzzy PID Control Algorithm in Heating Furnace

e furnace of steel plant is a complex controlled object and it has the properties of nonlinear, Time-varying and delay. Its modeling and control are very difficult. The temperature control of the furnace mainly depends on the control of gas flow. Therefore, the study of a reasonable gas flow control program is the key to increase the level of heating control. In this paper, an improved fuzzy PID control algorithm is proposed, in which, PID control algorithm and fuzzy control algorithm are integrated together, and its characteristics are improved according to feature of furnace. This made the algorithm to have good adaptability and Interference capability. The simulation results show that the improved control algorithm is better than traditional algorithm in overcoming the non-liner, delay of the object and the performance is excellent.

Research of night vision image denoising method based on the improved FastICA

When apple harvesting robot operates at night, there are lots of noise in the apples night vision image captured by image processing system. And the noise is mainly Gaussian noise, and mixed with some Salt and Pepper noise. In order to improve the harvesting efficiency and precision, the Independent Component Analysis (ICA) is introduced into the denoising method for night vision image. Mixture ICA model with noise based on the Maximum Likelihood Estimation regards likelihood of the observed signals as the objective function, which leads to the estimation to the mixed matrix poor. Aiming at the problem, this paper improves the objective function by bias removal technique, which can reduce the bias caused by the noise. And ICA is optimized by Fixed-point algorithm, so that the denoising method can operate efficiently. Through ICA transform, the robot image processing system gets the images independent components, and then these components with noise-free are estimated by using certain denoising method. Finally, this algorithm realizes the night vision image denoising. In order to verify the effectiveness of the improved FastICA, both Median-Average Filtering denoising and FastICA algorithm based on Maximum Likelihood estimation are simulated by MATLAB to compare the denoising effect. From the Relative Peak Signal-to-Noise Ratio (RPSNR), the improved FastICA denoising method in this paper relative to the other two denoising methods, respectively increases by 19.32%, 4.65%. In conclusion, the improved FastICA algorithm has unique advantage for night vision image denoising, which provides a solid foundation for the night operation of apple harvesting robot.

MPPT control method of PV system based on active disturbance rejection control

This paper presents a maximum power point tracking method based on the active disturbance rejection controller. The tracking-differentiator is used to obtain the differential signal of current and voltage, and the system is controlled by the active disturbance rejection controller to ensure the system works at the maximum power point. The proposed strategy is validated in Matlab/Simulink, and the feasibility and superiority of the algorithm are verified by comparing with the traditional incremental conductance method based on PI controller.

Study on DV-HOP node location algorithm for Wireless Sensor Networks

The traditional DV-Hop localization algorithm uses the number of hops multiplied by the average jump distance to estimate the distance and uses the maximum likelihood estimation method to locate. In view of this positioning method leads to greater error, in this paper, an improved DV-Hop algorithm is proposed based on hop number correction and improved particle swarm optimization. On the one hand, the minimum number of hops of the unknown node to the anchor node is corrected by introducing limit the number of hops mechanism, thus obtaining the optimal hops value; On the other hand, the improved particle swarm optimization algorithm is used to calculate the node coordinates instead of the maximum likelihood estimation method. Verified by matlab simulation software, compare with traditional DV-HOP and PSO-DVhop, the results of the improved algorithm are superior and feasible.

Research on data processing for condition monitoring of wind turbine based on Hadoop platform

The status monitoring data of wind turbines have large, multi-source, heterogeneous, complex and rapid growth of large data characteristics. The existing data processing methods are difficult to guarantee efficiency when handling massive amounts of data, and may miss the best time to troubleshoot. How to deal with the monitoring data more efficiently is of great significance to the accurate judgment of the fault. This paper proposes the use of cloud platform to deal with massive data to improve efficiency. Firstly, the state monitoring model of wind turbine is put forward. Then, the fuzzy C means clustering algorithm is introduced, and the algorithm process is realized by MapReduce model. Finally, the experiment is carried out with Hadoop platform, using distributed database HBase to store data, and using distributed programming framework MapReduce to calculate data. It is found that with the increase of the data volume and the number of nodes, the cloud platform is able to store and calculate data at a faster speed.

Study on prediction method of hot metal temperature in blast furnace

In order to improve the accuracy of the blast furnace hot metal temperature prediction, the Least Squares Support Vector Machine optimized by an improved Particle Swarm Optimization is proposed to solve this problem. By using the improved optimization algorithm to optimize the support vector machine parameters, the prediction model can obtain well prediction effect. Whats more, to avoid particle swarm easily falling into local minim problem during the process of the optimization, the chaotic method is introduced to help the particle swarm to jump out of the local optimal solution in the process of particle swarm optimization. Besides, the varying learning factors and the quadratic decreasing of the inertia weight are introduced to enhance the local search ability of the algorithm and pick up the convergence speed. The prediction model is established by the least squares support vector machine based on the chaos particle swarm optimization, then this method is compared with other prediction methods. The simulation result indicates that the prediction model has higher prediction accuracy. Therefore, a more accurate prediction method of the blast furnace hot metal temperature is proposed.

Research on LEACH algorithm based on double cluster head cluster clustering and data fusion

For the problems of uneven cluster distribution and excessive route energy consumption, which are caused by the clustering of low energy adaptive clustering algorithm (LEACH) in wireless sensor networks, an algorithm based on the clustering of double cluster heads and the data fusion mechanism of information entropy are proposed: In the cluster head selection algorithm, Select two levels of cluster heads in a well-divided cluster, two cluster heads perform different duties, this can be better to share the energy consumption, extend the network life cycle. Cluster head nodes use information entropy for classification and fusion, making the fusion results more accurate and data transmission more efficient. The algorithm is a good way to balance the energy consumption of nodes of a wireless sensor network composed of static location nodes, and extend the survival time of the network.