# Synchronous detection method for senescence quality of damaged Korla fragrant pears during storage

## TL;DR

This study develops a method to detect and predict the quality decline of damaged Korla pears during storage, using enzyme activity and machine learning models.

## Contribution

A novel multi-output model using SVR for synchronous detection of senescence quality in damaged pears is proposed and validated.

## Key findings

- Senescence quality of damaged pears changes rapidly with higher injury levels during storage.
- SVR model outperformed PLSR and LSTM in predicting senescence indicators with R2 values above 0.95.
- Enzyme activities like SOD, CAT, and POD increase with storage time and injury levels.

## Abstract

Korla fragrant pears are highly susceptible to mechanical damage during harvest, storage, and transport, which accelerates fruit browning and senescence, leading to fruit degradation and even a complete loss of commercial value.

To enhance the utilization value of damaged pears, this study used superoxide dismutase (SOD) activity, catalase (CAT) activity, peroxidase (POD) activity, superoxide anion (
O2-.) generation rate, and hydrogen peroxide (H2O2) content—factors directly related to pear browning—as evaluation indicators of senescence quality, and investigated the changes in the senescence quality of damaged pears with varying injury levels under impact load during storage. Furthermore, a multi-output model for predicting the senescence quality of damaged pears during storage was constructed using partial least squares regression (PLSR), support vector regression (SVR), and long short-term memory (LSTM). The optimal prediction model was subsequently selected from these.

The results indicated that as storage time increased, the average SOD activity, CAT activity, POD activity, O2-. generation rate, and H2O2 content in pears with different injury levels gradually increased. Higher damage levels resulted in a more rapid change rate of senescence quality. The constructed SVR multi-output model was the optimal model for predicting the senescence quality of damaged pears during storage, achieving R2 values above 0.95 for the prediction of SOD activity, CAT activity, POD activity, O2-. generation rate, and H2O2 content.

These findings provide a theoretical reference for investigating fruit senescence mechanisms and the synchronous detection of senescence quality.