The reason for this is the difference between the simulation software and the actual situation. The state of many self-excited oscillation circuits at time 0 determines whether they can start to oscillate. The reason why this circuit can't start vibration is because the software thinks that the voltage on the capacitor at time 0 is 4.5V following the negative terminal of the operational amplifier, and it doesn't start vibration. In fact, the voltage of the capacitor is 0.

The solution is also very simple, and a switch is connected in parallel with the capacitor C (not needed in the actual circuit). After the simulation starts, press this switch and then release it, and the circuit can start to vibrate. This is a software problem, not a circuit problem.

Look at the following figure, which is the process waveform of starting vibration when the button is pressed before starting vibration. Also pay attention to adding a ground wire to the negative position of the battery.