Abstract: Anaerobic fermentation is one of the important ways for the utilization of livestock and poultry manure resources. As an important intermediate product, volatile fatty acids can well reflect the degradation process of organic substances, and its real-time detection is of great significance for online monitoring of the fermentation process. In this study, a near-infrared real-time dynamic detection device and a digestate circulation loop system were designed based on an online bypass detection method on an existing laboratory-type anaerobic fermentation system, and the feasibility of online and real-time detection of total volatile fatty acids and individual volatile fatty acids such as acetic acid, propionic acid, isobutyric acid, butyric acid, isovaleric acid, valeric acid and TVFA were investigated. Specifically, the online bypass system includes an online detection device and a digestate circulation loop system. The core of the online detection device is to use the near infrared spectrometer (MicroNIR pat-u, 950-1 650 nm) to collect the optical information of digestate online in real time. The spectral acquisition mode is transflectance mode, which can obtain the spectral information of digestate to the greatest extent. The digestate circulation loop system is able to drain the digestate to the online detection site. After the detection, the loop system is cleaned by water and dried by hot air, so as to avoid the cross contamination between the two measurements and ensure the correct optical information collected by the spectrometer each time. This study takes pig manure anaerobic fermentation process as the research object. The experiment lasts for 60 days. The optical information of the anaerobic fermentation process is collected daily by the near-infrared spectrometer of the online bypass system, and the digestate samples are collected for chemical analysis at the same time. The volatile fatty acid content determined by gas chromatography and the collected optical information of digestate were used to establish a quantitative analysis model. Finally, the quantitative analysis model was used to realize the dynamic detection of volatile fatty acids in the anaerobic fermentation process, in which the chemometrics algorithms were used including the spectrum preprocessing algorithm (1st derivative, Autoscale, SNV) and partial least squares modeling algorithm.The results show that the designed online detection scheme and calibration model have a good effect on the dynamic real-time detection of volatile fatty acid content, and even if the concentration of isobutyric acid is relatively low (5/10 000), it can still achieve a more accurate detection. The results were that the R2p=0.811, RPD=2.055 and RSD=2.288% for the acetic acid, the R2p=0.849, RPD=2.116 and RSD=2.195% for the propionic acid, the R2p=0.968, RPD=5.555 and RSD=2.132% for the isobutyric acid, the R2p=0.889, RPD=2.701 and RSD=1.846% for the butyric acid, the R2p=0.94, RPD=3.843 and RSD=3.444% for the isovaleric acid, the R2p=0.889, RPD=2.751 and RSD=2.387% for the valeric acid, the R2p=0.734, RPD=1.846 and RSD=2.150% for the total volatile fatty acids. Based on the online detection scheme proposed in this study, near-infrared spectroscopy technology can achieve fast, online, and real-time dynamic detection of volatile fatty acids in the anaerobic fermentation process. In the next step, this research can be carried out in a plant-scale biogas project with online bypass system, so as to realize the long-term detection of biogas project by near-infrared technology, providing real-time data support for regulating anaerobic fermentation process, generating income for enterprises and reducing economic losses.