The test used CL-YB-116B pressure sensor to test the support pressure change of the model floor. The sensors were arranged in two columns at different positions of the coal seam floor. The first column was No. 1 to No. 48, and the second column was No. 4 9 to No. 96. The numbers were arranged in turn. The total length of the sensor was the same as that of the coal seam excavation distance. The size of each sensor was 20 cm × 5 cm × 5 cm, and the range was 0–200 kg. Due to the physical dimensions and deployment configurations of sensors, potential limitations in spatial resolution may arise, accompanied by measurement inaccuracies induced by stress concentration phenomena or dispersion effects. Furthermore, the measurement fidelity could be affected by stability issues associated with prolonged loading conditions and progressive deformation of model materials. In order to avoid the interference of the mining process to the sensor, a 2 cm thick model material was covered above the sensor. And this shielding layer serves dual functions: preserving sensor integrity during dynamic stress redistribution while maintaining measurement continuity in the excavation simulation phase. The arrangement was shown in Figure 12.

In the unexcavated coal seam floor, three horizontal sensing optical fibers were arranged along the inclination of the coal seam to test the distribution of support pressure. The fiber numbers were D1, D2 and D3, which were 0.6 m, 1.2 m and 1.8 m away from the open off-cut, respectively. We chose a tight-sleeve single-mode fiber with a cross-section diameter of 2 mm. During the laying process of the test, it was necessary to ensure that the optical fiber was closely bonded to the model material and appropriate tensile prestress was applied.

In order to reduce errors, we could fix the optical fiber with a positioning fixture. It could enable its position in three-dimensional space. However, the uneven shrinkage of the model material during the curing process might affect strain transfer due to a slight separation occurring between the optical fiber and the medium. Meanwhile, uneven prestressing or the device being affected by the coal seam inclination might cause some errors. Thus, researchers should check the fiber continuity and signal strength before and after the test to eliminate interference factors. Some instruments and equipment were shown in Figure 13. From left to right, they represented optical fiber, the NBX-6055 Photonic Nanoanalyzer, and the connecting line. The arrangement was shown in Figure 14.

This experiment used the Pulse-PrePump Brillouin Optical Time Domain Analysis (PPP-BOTDA) (Kishida and Li, 2006), which had more advantages than traditional technology. PPP-BOTDA was different from BOTDA (Horiguchi and Tateda, 1989; Fang et al., 2023; Yin et al., 2023) distributed optical fiber sensing technology in that it had higher frequency resolution, high-accuracy Brillouin frequency shift and good Brillouin spectrum morphology. When the strain and temperature of the measured object change, the principle of Brillouin frequency shift change was shown in Equation 5. ∆ v B = C ϵ ∆ ε + C T ∆ T (5)

In the formula, ∆ v B was the Brillouin frequency shift amount, MHz; C ϵ was the Brillouin frequency shift strain coefficient; C T was the Brillouin frequency shift temperature coefficient; Δε was the strain change amount; ∆ T was the temperature change amount. Ordinary single-mode fiber, such as Corning SMF28 Brillouin line width temperature. The degree coefficient C T was about −0.93MHz/°c, and the strain coefficient C ϵ was about 0.0003MHz/με (Brown et al., 2005).

The so-called “strain” referred to the strain change in the axial direction of the optical fiber, which was actually produced by the joint action of axial strain and radial strain. Within a small deformation range, the two were directly proportional and related to Poisson’s ratio (Gu, 2013; Zuo et al., 2024). By monitoring changes in optical fiber transmission power, the Brillouin frequency shift of each section could be obtained, and then the strain or temperature information of the optical fiber could be obtained. This method had high measurement accuracy and small spatial resolution. The principles diagram of PPP-BOTDA distributed optical fiber sensing technology and optical fiber stimulated Brillouin scattering were shown in Figures 15, 16.

Although PPP-BOTDA technology has high spatial resolution and stability, there may still be systematic deviations in long-term or complex environments. For example, temperature fluctuations may cause frequency shift errors, and the device itself may also have signal drift and connection port reflection loss and optical path bending, which will also affect the signal-to-noise ratio. In order to reduce the impact of errors, prestress control and signal verification measures were taken. In the future, it would still be necessary to strengthen error compensation and stability evaluation to improve data reliability.

The optical fiber sensor monitored the axial strain based on the Brillouin scattering principle. When the fiber was subjected to axial stress, its Brillouin frequency shift changed. It was assumed that the Brillouin frequency shift was v B 0 when the initial stress was not applied. When the strain ϵ was applied, the Brillouin frequency shift became v B ϵ , and the frequency shift variation ∆ v B was expressed as Equation 6: ∆ v B = v B ϵ − v B 0 (6)

The variation ∆ v B reflected the frequency shift difference of the fiber under stress. According to the Brillouin scattering theory, there was a linear relationship between Brillouin frequency shift and strain ε, and the proportional coefficient C ϵ represents the relationship between the two, which was: ∆ v B = C ϵ · ϵ (7)

Among them, C ϵ was the sensitivity coefficient of strain, which indicated the sensitivity of Brillouin frequency shift to strain. According to the experimental and theoretical derivation, the linear relationship between the Brillouin frequency shift change. ∆ v B and the strain ε was obtained as follows Equation 8: ∆ v B = 4.88 v B 0 ϵ (8)

Among them, the change of the Brillouin frequency shift was proportional to the strain, and its proportional coefficient was 4.88 times the initial Brillouin frequency shift. Which ϵ can be obtained from Equation 9 ϵ = σ / E (9)

Among them, σ was the local stress; E was the elastic modulus of the optical fiber. In general, the Brillouin frequency shift was mainly caused by the axial strain of the optical fiber, which in turn comes from the externally applied stress or support pressure. When the optical fiber was tightly bonded to the medium, the local pressure was transmitted to the optical fiber through the medium, generating axial strain, which leads to the change of the Brillouin frequency shift.

In the test system, the sensing fiber D1 corresponded to the pressure sensor 2#60, the sensing fiber D2 corresponds to the pressure sensor 2#72, and the sensing fiber D3 corresponds to the pressure sensor 2#84. The average Brillouin frequency shift of the length (200 mm) of the pressure sensor corresponding to the sensing fiber and the change of the inclined support pressure measured by the pressure sensor are plotted as shown in Figure 19. It indicated that the Brillouin frequency shift measured by the D1, D2 and D3 sensing fibers was consistent with the change trend of the support pressure measured by the pressure sensor.

When the working face was advanced to 182 m, the Brillouin frequency shift value of the D1 sensing fiber began to increase. The working face was advanced to 363 m and the frequency shift value of the D1 sensing fiber reached the maximum value. When the working face was advanced to 693 m, the frequency shift value of the fiber was basically stable. Along the advancing direction of the working face, the advanced influence range of the support pressure measured by the D1 sensing fiber was 74 m, the peak point was 133 m behind the working face, and the lag influence distance reached 443 m. The corresponding No. 60 pressure sensor had tendency to advance the influence range of 60 m, the peak point was 177 m away from the rear of the working face, and the lag influence distance reached 450 m.

The working face advanced to 400 m, the Brillouin frequency shift value of the D2 sensing fiber began to increase. The working face advances to 551 m, the Brillouin frequency shift value of the D2 sensing fiber reached the maximum. When advancing to 858 m, the Brillouin frequency shift value of the D2 sensing fiber was basically stable. Along the advancing direction of the working face, the advance influence range of the support pressure measured by the D2 sensing fiber was 77 m, the peak point was 126 m away from the rear of the working face, and the lag influence range reached 347 m. The advanced influence range of the tendency measured by the No. 72 pressure sensor was 49 m, the peak point was 114 m behind the working face, and the lag influence distance was 258 m.

The working face advanced to 595 m, the frequency shift value of the D3 sensing fiber began to increase. At 748 m, the frequency shift value of the fiber reached the maximum value. When the working face advanced to 873 m, the frequency shift value of the fiber was basically stable, and excavation was completed at 920 m. The advance influence range of support pressure measured by the D3 sensing fiber was 93 m, the peak point was 85 m behind the working face, and the lag influence distance was 113 m. Corresponding to the No. 84 pressure sensor, the advanced influence range was 41 m, the peak point was 41 m away from the rear of the working face, and the lag influence distance was 150 m. The mentioned discrepancy suggests that fiber optic sensors might be more sensitive to micro-strain accumulation, capable of capturing earlier and more subtle stress variations in the surrounding rock. In contrast, pressure sensors could detect stress concentration at localized contact points and might underrepresent broader deformation fields. These differences might contribute to knowing the importance of support pressure distribution and rock behavior.

Fiber optic monitoring was based on the principles of light interference and scattering and had the advantages of high precision and strong anti-interference, while traditional pressure sensors relied on changes in electrical signals such as resistance and capacitance to achieve measurement. The fundamental difference in the measurement principles between the two made fiber optic monitoring more suitable for continuous monitoring over long distances and in plasma environments, while traditional sensors were more suitable for local monitoring.

By comparing the fiber frequency shift with the numerical simulation, it can be seen that the consistency of the two was mainly reflected in the evolution law of support pressure and the dynamic process of rock failure. First of all, both show that with the advancement of the working face, the pressure at different stages shows a peak change: the support pressure was not obvious at the initial stage, then gradually concentrated and reached the peak value, and finally the regression was stable. This process reflects the significant influence of working face advance on stress distribution. Secondly, both of them emphasize the destructive control of conglomerate layers. In the process of rock failure, the failure of the lower and upper conglomerate layers not only affects the concentration of support pressure but also leads to the complex evolution of stress transfer, which ultimately affects the surface deformation and the stability of the goaf. The above commonality shows that the optical fiber frequency shift monitoring method has good consistency with the numerical simulation, which can more comprehensively grasp the stress change and rock failure law in the process of coal seam mining and provide a scientific basis for support design and mine safety management. These monitoring data not only verify the reliability of the theoretical model but also provide more detailed pressure distribution information, which has important reference value for predicting rock damage and pressure concentration areas in the process of coal seam mining and helps to improve the accuracy and safety of support design.

In this section, we adopted a complementary analytical approach combined with theoretical calculation and testing. Fiber optic monitoring was mainly used to capture internal responses, such as internal strain or deformation. However, pressure sensors were used to record pressure changes outside the structure. This combination of internal and external monitoring methods helps to verify and supplement the theoretical analysis results from different angles. It helps to enhance the reliability of the conclusions. The influence distance of advance mining support pressure and the influence range of lag tendency support pressure/MPa Brillouin frequency shift/MHz 64 measured by D1, D2 and D3 sensing optical fibers and pressure sensors were drawn in Table 4. The comparative analysis of the data in the table showed that the average mining influence distance measured by the optical fiber was 81.3 m, while the result measured by the pressure sensor was 50 m. The average lag influence distance measured by the optical fiber was 301 m, while the result measured by the pressure sensor was 286 cm. The resulted measured by the two are basically the same. Among them, the mining influence distance measured by optical fiber D3 was 14 m different from that of pressure sensor, and the lag influence distance measured by optical fiber D2 was 89 m different from that of pressure sensor. The reason was that after the working face advances to 516 m, the pressure on the working face was more severe, the load of overlying strata was transferred in a large range, and the overlying strata behind the goaf are rebalanced many times.

Tables 4, 5 are the comparative analysis of the optical fiber tendency support pressure data and the theoretical calculation values. It can be seen that there was a high consistency between the optical fiber monitoring and the theoretical calculation results. Among them, the data in Table 4 prove the stability of optical fiber monitoring on different monitoring indicators. In the monitoring of the mining influence range and lag influence range, there was a certain gap between the results of optical fiber monitoring and the pressure sensor. But the trend of the two was consistent. This indicated that the relationship between theoretical calculations and on-site monitoring was not antagonistic but complementary: the theoretical model provides reliable expectations for monitoring, while on-site monitoring captures more dynamic changes due to the complexity of geological conditions.

Accordingly, the optical fiber monitoring results of D1 and D3 points in Table 5 were basically consistent with the influence range of the theoretical calculation of the peak value of the inclined support pressure, which indicated that the theoretical model can accurately predict the distribution of pressure during mining in these areas. Although the monitoring value of the D2 point (218 m) was higher than the theoretical calculation value (200 m), which was about 1.09 times, this deviation also had a reasonable explanation. Since the working face was mined to the middle of the model, the movement range of the rock stratum in the process of continuous rotary sinking was expanded, and the transfer of the load to the inclined side of the coal body was enhanced, so the influence range of the support pressure was increased. At the same time, this result showed that in the actual working conditions, the influence of the complexity of geological conditions and the non-uniformity of stress transfer on the pressure range may exceed the prediction of the theoretical model, that was, the theoretical calculation will have certain limitations, but the basic trends of the two were still consistent. This provided a strong verification for the reliability of optical fiber monitoring in practical engineering applications.

In addition, combined with the comprehensive analysis of Tables 4, 5, it can be seen that the optical fiber monitoring technology had high reliability and sensitivity in the monitoring of mining, hysteresis effect and support pressure range. Optical fiber monitoring technology had obvious advantages in long-distance monitoring and was more suitable for long-distance monitoring in complex geological environments.

Optical fiber monitoring showed significant advantages over traditional pressure sensors and theoretical calculations in the monitoring of mining influence and support pressure peak influence. In terms of mining influence range, the average value of optical fiber monitoring (81.3 m) was higher than that of pressure sensor (50 m). At D3 point, the optical fiber monitoring range (93 m) was significantly higher than that of pressure sensor (41 m). This showed that optical fiber monitoring technology had obvious advantages in long-distance monitoring, and was more suitable for long-distance monitoring in complex geological environment. In addition, the performance of optical fiber monitoring in the lag influence range was better than that of the pressure sensor, with an average value of 301 m, while the pressure sensor was 286 m, indicating that optical fiber monitoring was more sensitive and stable in capturing the lag response.