SF₆ is applied as an insulating medium in gas-insulated equipment due to its excellent insulation and arc extinguishing properties. However, aging or overheating within the equipment can cause partial discharge, resulting in the decomposition of SF₆ to form low-fluoride compounds, which react with water and oxygen inside the equipment to produce the characteristic gas H₂S. As a decomposition product of SF₆, H₂S is one of the characteristic components of fault diagnosis of gas-insulated equipment, and has strong corrosivity and toxicity. The inferior nature of this gas can lead to serious damage inside the equipment. Therefore, high-sensitivity online monitoring of trace H₂S gas is of great significance for the stable operation of gas-insulated equipment. Detection techniques for SF₆ decomposition products envelop non-optical and optical methods. Beer-Lambert-based Photoacoustic Spectroscopy (PAS) sensing technology is an indirect absorption spectroscopy technique that has gained a wider range of attention than non-optical detection methods due to its advantages of no gas load, no maintenance, fast response time, and high sensitivity. When the concentration of the gas to be measured is unchanged, the excitation light efficiency of the photoacoustic signal can be enhanced by increasing the length of the absorption path. The excitation light injected into the multi-pass cell is collimated by the collimator and reflected back and forth between the concave mirrors installed on both sides of the buffer of the photoacoustic cell. H₂S has a good spectral line absorption peak in the near-infrared band. The near-infrared laser is incident on a multi-pass photoacoustic cell after power amplified, and the excitation efficiency of the photoacoustic signal is greatly improved through multiple optical reflections. Combined with acoustic resonance amplification, optical fiber amplification and wavelength modulation-second harmonic detection technology, a set of photoacoustic spectroscopy gas detection system is built to achieve high-sensitivity detection of trace H₂S gas in SF₆ background. A 1 574.56 nm near-infrared distributed feedback DFB laser cascaded a high-power Erbium-doped Fiber Amplifier (EDFA) is selected as the excitation light source of the system. A single-mode fiber-optic and a collimator are used to collimate the light source into the photoacoustic cell. The photoacoustic cell consists of a resonant tube and two buffer chambers flanked by concave mirrors. The photoacoustic signals are detected in real time by capacitive sensors, and then processed by a Field-programmable Gate Array (FPGA)-based lock-in amplifier. The second harmonic signals are obtained by the LabVIEW program, and the concentration of the target gas to be detected is calculated. SF₆ is a gas with special physical properties such as high density, high diffusion coefficient and low specific heat capacity, which is easily affected by the high-pressure environment inside the field work equipment. COMSOL simulation software is used to simulate the resonance frequency of the photoacoustic cell in SF₆ background and N₂ background, and the experimental results are consistent with the simulation results. In SF₆ background, the photoacoustic cell resonance frequency is reduced to 638 Hz. In order to effectively improve the amplitude of the photoacoustic signal of the fiber amplification laser photoacoustic spectroscopy detection system based on wavelength modulation-second harmonic detection technology, the amplitude of the modulation parameters is optimized. When the modulation current of the DFB laser is set to 3.5 mA, the amplitude of the photoacoustic signal detected by the system is the highest. To assess the response of the multi-pass photoacoustic cell to the measured gas concentration, the high-purity SF₆ gas is mixed with a H₂S/SF₆ mixture of 50 ppm in a certain proportion, and the system shows a good linear response of 9.59 μV/ppm for H₂S gas concentrations up to 50 ppm. A continuous charge of pure SF6 gas into the multi-pass cell is used to measure the noise level and evaluate the Minimum Detection Limit (MDL) of the system. The standard deviation (1σ) of the noise is 0.85 μV. The experimental results show that the Normalized Noise Equivalent Absorption (NNEA) coefficient is 2.23×10^-9 cm^-1·W·Hz^-1/2. Allan deviation analysis is performed on the noise over a time period of 1 000 s. The Allan deviation result shows that when the averaging time is 100 s, the MDL of the system is 27 ppb. The system gradually stabilized with the increase of the mean time. The detection limit on the order of ppb provide a favorable solution for diagnosing the failure of gas insulation equipment.