Multi-objective-decision-based pumping unit parameter optimization method

The invention provides a multi-objective-decision-based pumping unit parameter optimization method, comprising the steps: 1, determining production efficiency influence factors and performance variables of a pumping unit; 2, requiring sample data; 3, performing dimensionality reduction on load data; 4, building a net input variable set by non-load variables and new principal elements of load; 5, building sample values of input variables; 6, performing normalization processing; 7, selecting training and test samples; 8, building a generalized regression neural network and training; 9, building a liquid production capacity preference function; 10, building a fitness function for evaluating decision variable individual dominance relation; 11, calculating a mean value of system environment variables; 12, building an initial population by decision variables and generating an elite population; 13, performing genetic iterative calculation to obtain a second generation of elite population and a parent population; 14, iteratively circulating to finally acquire optimized production efficiency influence factor values. After optimization, lowest power consumption can be ensured in case that the liquid production capacity is basically fixed.