Multilayer Perceptron with Analysis of Variance for Detection of Rough Handling in the Small Household Appliances Express Service

In recent years, rough handling and improper behavior in express delivery services have brought losses and poor experience to enterprises and users, and caused huge material waste. Therefore, it is particularly important to propose an objective and quantitative monitoring and evaluation method. This paper proposes an ANOVA and MLP fusion method for detecting rough handling in small home appliances. Firstly, this paper uses a high-precision and low-power air pressure sensor to collect air pressure data of rough handling for small home appliances. Secondly, a data interception and pre-processing method based on air pressure response value are designed. Finally, an Analysis of Variance (ANOVA) and Multilayer Perceptron (MLP) fusion method for small home appliance is proposed. Experiments show that the proposed detcetion method has excellent detection results with an average recognition accuracy of 99.9%. Its detection accuracy is higher than that of Linear Discriminant Analysis (LDA), K-Nearest Neighbor(KNN), Support Vector Machine (SVM), and Multilayer Perceptron(MLP) by 3%, 1.8%, 1.5%, and 1%. The proposed method can effectively identify four types of rough handling in the express service: side extrusion, stacking extrusion, bilateral extrusion, and heavy trample, which can reduce the occurrence of small appliance rough handling and improve the express environment of small appliances.