The invention discloses a quasi-distributed steel electric furnace fsFBG temperature intelligent optimization method based on an expert model, and the method comprises the steps: firstly installing a femtosecond laser point-by-point direct writing fiber bragg grating sensor at a key position of an electric furnace to collect data, and carrying out the demodulation through different methods, so as to guarantee the accurate obtaining of temperature information; establishing an expert model, and fusing a plurality of independent gradient boosting tree models; then, deep data analysis is performed, and temperature data identification key elements are mined; an optimization strategy is generated based on an expert model and data analysis, and an optimal temperature control strategy under various working conditions is determined through high-performance calculation; an optimization strategy is fed back to an electric furnace control system through a real-time data system, the actual temperature is continuously monitored, and deviation is reduced through a PID controller; converting the adjusted control strategy into a precise instruction to regulate and control the power of a heating element and the flow of a cooling system so as to realize precise temperature control; and finally, the operation effect of the electric furnace is evaluated regularly, temperature control and energy consumption conditions are compared, and the expert model and the control algorithm are continuously optimized.