A Fitting Method of Characteristic Curve of Pump Group

Abstract

The invention discloses a pump set characteristic curve fitting method which comprises the following steps: establishing a single-pump neural network model for each single pump in a pump set; connecting the single-pump neural network models in parallel, outputting output results of the single-pump neural network models in parallel, constructing an m-dimensional vector according to a parallel connection sequence, and carrying out point multiplication on the vector and a pump set state vector representing the operation state of each pump to obtain a pump set neural network model; Taking the monitored variable samples as training data, and repeatedly training the pump set neural network model to obtain a trained pump set neural network model; Inputting variables of a to-be-fitted characteristic curve of the pump set into the trained pump set neural network model to obtain dependent variables output by the pump set neural network model, and establishing a corresponding relation between thevariables and the dependent variables to obtain the characteristic curve of the pump set. The nonlinear relation between the flow and the pressure of the water pump can be well fitted, and the problems that a power frequency pump and a variable frequency pump are combined, and the operation condition is changeable and complex can be solved.

Description

technical field [0001] The invention relates to the field of monitoring the operating state of pump sets in a water supply system, in particular to a method for fitting characteristic curves of pump sets. Background technique [0002] The water supply system is the lifeline project of the city. With the rapid development of the urban economy and the improvement of people's living standards, the scale of water supply continues to grow, and the energy consumption of water supply is increasing. The traditional water supply system operation management scheduling mode is facing severe challenges. [0003] my country's water supply energy consumption occupies a relatively large proportion of the total electricity consumption, accounting for 0.45% of the total electricity consumption, reaching 11.607 billion kwh (2014) throughout the year, becoming one of the most important electricity consumers in cities . For the water supply enterprises themselves, the water supply energy consum...

AI Technical Summary

Problems solved by technology

[0006] The method based on experimental measurement needs to install monitoring instruments for each pump in the pumping station, and needs to adjust the working conditions to conduct a comprehensive measurement of the pump. This type of method has disadvantages such as complicated operation and may affect the safety of normal water supply.

Since my country generally only monitors the (total) flow of the pumping station at present, the method based on historical monitoring data is mainly to fit the head flow characteristic curve of the entire pumping station. However, the establishment of the hydraulic model and the calculation of energy consumption require The characteristic curve of each single pump of the group, so the above method cannot solve the problem

[0007] Based on the monitoring data of the pumping station, the model of reverse deduction and fitting of each single pump characteristic curve mainly adopts the method of parameter search. This method needs to determine the form and parameters of the characteristic curve equation, so the expressive ability of the equation curve is limited to a certain extent, and it cannot reflect the characteristics of the pump. Uncertainty caused by loss of use, which may have a certain error

Images