With the booming development of the new industrial park project of fruit canning processing enterprises in Pingyi County,
it is of great practical significance to accurately predict the sales volume of canned goods for enterprises to make correct decisions to achieve the expected revenue target. The new hybrid intelligent algorithm proposed in this paper uses Newton interpolation formula and 3/8 Simpson equation to improve the adjacency matrix in the grey prediction model, changes the coefficient of the whitening equation, and combines BP neural network to predict the sales volume of canned fruit. The algorithm is obtained by comparison with the actual sales data analysis of the root mean square error (RMSE), mean absolute percentage error (MAPE) and mean absolute error (MAE), which are 2.4%, 2.0 and 1.3 respectively. The algorithm accuracy is higher compared with the original hybrid intelligent algorithm, effectively improving the reference value, and can help enterprises more reasonable arrange production.