Fish meal is rich in protein and fat and is the main animal-derived feed material. In the process of transportation and storage, the fish meal can be easily to deteriorate, resulting in the reduction of its nutritional components. This is not only endangers the normal reproduction and health of animals, but also poses certain potential safety hazards to the safety of animal-derived food and human health. It is necessary to carry out an effective detection for quality of animal-derived feed. In recent years, bionic olfaction technology has developed rapidly, it can quickly detect the quality of products, compared with near infrared spectroscopy and traditional physical and chemical index detection methods, it is faster and more accurate, but it is rarely used in the quality detection of fish meal. Therefore, in order to solve the above problems, a portable detection device on quality of fish meal has been developed in this paper. The hardware part of the device is mainly composed of a gas acquisition and transmission module, a control processing storage module, a data acquisition module, and a sensor array module, which the gas acquisition and transmission module includes a sample supplement gas cylinder, a sample gas generation chamber, an activated carbon purification bottle, two two-position two-way electromagnetic valves, a micro air pump, a gas flowmeter, a gas sampling chamber, and a one-way valve. In the software part, the strawberry pie is the core, qt creator graphical programming software across platforms was selected, and the real-time data acquisition display and storage part was mainly designed. The key component was the 10-bit ARPI600 data acquisition module. The device could basically realize the quality detection of fish meal, and the detection result was relatively accurate. In order to obtain the detection performance of the device, the test parameters of the device needed to be optimized. First, fresh fish meal was placed in a 35 ℃ thermostatic artificial climate box to make it decay gradually at the storage time. Fish meal of different levels was collected as the test sample during the storage process, and a total of six types of fish meal samples which was at different storage times were selected for the samples of the test. In this paper, it selected the factors that included gas flow, sampling time and cleaning time that had a great influence on the detection results, taking the dispersion ratio as an index, the optimal parameters were obtained through optimization analysis by using response surface method and design - expert software, and the feasibility and detection performance of the detection device for fish meal quality under the optimal parameters were verified. The test results showed that the gas flow rate, sampling time and cleaning time were all significant factors, and the primary and secondary order of factors was as follow: Cleaning time, sampling time and gas flow, and the interaction between them was significant. With the increase of gas flow and sampling time, the dispersion ratio tended to decrease first and then increase, and when the gas flow was in the range of 2-2.5 L/min and the sampling time was in the range of 35-45 s, the dispersion ratio was small. As the gas flow increased, the cleaning time decreased, while the dispersion ratio decreased first and then increased. When the gas flow rate was 2-2.5 L/ min and the cleaning time was in the range of 75-80 s, the dispersion ratio was small. As the sampling time increased, the cleaning time decreased, the dispersion ratio decreased first and then increased, and when the sampling time was in the range of 35-40 s and the cleaning time was in the range of 75-80 s, the dispersion ratio was small. In considering of all factors, the best parameters were 2.2 L/min of gas flow, 39 s of sampling time and 77 s of cleaning time. At this time, the dispersion ratio was the smallest, which was 0.579 9. Under the optimal parameters, fish meal samples with different storage time were tested, and the accuracy of different storage time discrimination was 89.4%, which could realize the detection of quality of fish meal and provide data reference and technical support for the subsequent research on the quality detection device. [ABSTRACT FROM AUTHOR]