| By the analysis of the monitoring data about the pollutant in surface water, this research discovered that the movement of the pollutant concentration in surface water has a rather strong non-linear characteristic. After comparing different forecast methods, we selected the BP Artificial Neural Network forecast method, which is based on the ameliorations and can catch the non-linear movement law completely.This method is different from other routine methods. It uses experiment data as the swatch, and studies on them by the use of the network's self-study and association memory abilities. It has a quite good function approach ability, so it can accord with the historic swatches commendably, and can achieve the aim of identifying the complicated non-linear mapping relationship among each influence factor. Besides, it can improve the precision of the water quality predict model. Furthermore, it provides an effective means for the environmental decision-making department to program the water environmental protection and treatment.On this basis, this research has set up a series of forecast model, which suits the permanganate exponent and ammonia nitrogen concentration of Shengmi monitoring section of water resource of Ganjiang Nanchang area. As a result, the forecast precision of permanganate exponent can achieve upwards of 70% on every month of the year, except for December, on which the precision is below 70%. So does the ammonia nitrogen concentration on every month of the year, except for August, on which the precision is below 70%. All of these indicates that the forecast precision is quite high, and the results are in relatively good agreement.This research created a water quality forecast system, compiled with VG language. The system possesses a friendly interface, and suits for Windows platform. It has the advantages of convenient-operating, easy-data import and intuitional-output, so it can serve for the decision-making and analysis intuitively.
|
PDF Full Text Request