Research On Prediction Method Of Optimum Drop Height In Drop Jump Based On BP Neural Network

Objective:The aim of this research is to evaluate the optimal drop height of individuals by multi-stage drop jump test based on reactive strength index.In anthropometric indexes(height,weight,etc.)and the strength quality indexes(Back Squat 1 RM,CMJ height,etc.),the indicators which are correlated with the drop height were used to construct BP neural network model to predict individual optimal drop height.Finally,the model’s fitting precision and prediction precision are evaluated.So that the prediction method of OPT can be used by the athletes who need drop jump training.Methods:In this study,49 adult high-level male athletes aged 18-25 in Shandong training center were selected.The strength quality indexes and anthropometric indexes were collected by testing method.Among them,the optimal drop height was identified by the jumping boxes(30-75cm,increment of 5cm)and the maximum reaction strength index were used as the standard to screen the optimal drop heights of individuals.And the Pearson correlation test was conducted for each index by mathematical statistics method.Then the BPNN model was established to predict the optimal drop height of drop jump training by mathematical modeling method.After the model was successfully built,using the Paired Sample T-test to analyse the deviation between the actual value of optimal drop height and the value predicted by the model.Meanwhile,the effect size between these two variables was also calculated.Results:(1)The relative strength of the back squat(r=0.696,P<0.001)and RSR(r=0.605,P<0.001)were highly positively correlated with the optimal drop height.Body weight(r=-0.455,P<0.01),CMJ(r=0.417,P<0.01)and body height(r=-0.413,P<0.01)had a moderately correlation with the optimal drop height.(2)The R value,R2 value,MAE value,MRE value and MSE value of the training set of model were 0.983,0.8,2.384,6.01%and 11.051 respectively.The R value of the test set was 0.974,R2 value was 0.974,MAE value was 2.468,MRE value was 6.44%,and MSE value was 11.685.The overall R value was 0.943,R2 value was 0.889,MAE value was 2.650,MRE value was 6.22%,and MSE value was 13.186.(3)The paired sample t-test indicated that the difference between the actual value of the optimal drop height(M=46.12,SD=11.01)and the predicted value(M=45.86,SD=10.79)are of nonsignificance(P>0.05).The Cohen’s d value of effect size was 0.04,showed that the actual value and the predicted value of the optimal drop height was almost identical in practice.Conclusions:(1)During dropping jump,with the increase of the drop height,the value of RSI increases at first and then drops.As a result,the individual optimal drop height can be obtained according to maximum reactive strength index.(2)Among the selected indicators,such as body height,weight,CMJ height,relative strength of back squat and RSR were all significantly correlated with the optimal drop height.When drop jump exercise was selected as the training method of plyometric training,the influence of the above factors on the optimal drop height should be taken into consideration.(3)The model constructed in this study was using body height,relative strength of back squat,CMJ height and RSR as the output layer.And the individual optimal drop height was used as the output layer.As a result,the BPNN model can accurately predict the individual optimal drop height relatively.And the fitting and forecast effects of the model are relatively well.