Preparation of Partially-saponified Humic Acid and Evaluation of Scale Removal and Scale Prevention Performances

2017-0

Pages 1-9
Author Yanchui WangYilei SunRongming Zhan
Abstract By controlling the concentration of NaOH in solution, the reaction mass ratio in the reaction between HA and NaOH, and the direct contact between the HA and NaOH solutions, various types of partially saponified HA were produced. Scale removal and scale prevention evaluations were performed according to the calcium scale precipitation method and the precipitation weight method. A scheme with optimal effects on scale removal and scale prevention integration was found, wherein the anti-calcium scale rate exceeded 80% and the calcium scale dissolution rate exceeded 70%. The influences brought on by the addition dosage of partially-saponified HA-U, duration of scale removal and scale prevention and temperature on the scale removal and scale prevention effects were investigated.

Single Threshold General Regression Neural Network Power Load Forecasting Accounting for Weather Factor

2017-0

Single Threshold General Regression Neural Network Power Load Forecasting Accounting for Weather Factor

Pages 10-18
Author Changhao XiaJin Ca
Abstract With the development of modern power systems, the accuracy requirement for load forecasting is getting higher and higher. Neural network forecasting method has the advantages of self-learning. While BP neural network is difficult to determine the structure and requires a plurality of training parameters, with the problems of slow convergence speed and local minimum. General Regression Neural Network (GRNN) has some strong advantages in approximation ability, classification ability and learning speed. All the network learning depends on the data sample. Once the learning samples are determined, the corresponding network structure is identified. In this way, the network can utmost avoid the effects of subjective assumptions on the result of prediction. In this paper, based on only one threshold parameter of GRNN, using the actual historical load data of a city and the weather data such as temperature and rainfall, a load forecasting model is established. The results show that the model can overcome the problems of BP neural network, not only the accuracy is very ideal, but also it can satisfy the requirement of load forecasting, also there are characteristics of simple process and good stability.