Study on the Influence of LED Spectrum on the Visual and Nonvisual Effects in the Interior Zone of Tunnels

Judicious use of lamps is of profound signiﬁcance to improve the internal traﬃc safety of tunnels. This study evaluated the biological eﬀects of spectrum on human visual and nonvisual under mesopic vision category. According to the diﬀerence of human eyes’ response to the visual and nonvisual biological eﬀects of diﬀerent wavelengths of light radiation, a method of applying the mesopic visual spectral luminous eﬃciency curve to the assessment of nonvisual biological eﬀects and blue light hazards is proposed. The results show that, in theory, under mesopic vision category, the spectrum with high correlation color temperature (CCT) and high color rendering index (CRI) has higher blue light content Z -value and has greater impact on human nonvisual eﬀects and blue light radiation damage, but it can also provide higher human perception luminance. The visual eﬀect detection experiment of driving in the interior zones of the tunnel is simulated in the laboratory, and the critical ﬂicker frequency and reaction time are taken as physiological indexes. The results show that the spectrum with high CCTand high CRI corresponds to lower critical ﬂicker frequency drop value and faster reaction time, which has a positive eﬀect on improving driving safety. Therefore, this paper suggests that the LED with high CCT and high CRI (CCT � 6500K, CRI2012 � 95) should be used in the interior zones of the tunnel with relatively short lighting section, and the CCTand CRI should be appropriately reduced in other road environments with long lighting section to reduce the impact of spectrum on nonvisual photobiology, so as to improve road


Introduction
Fatigue and drowsiness are the main causes of traffic accidents [1][2][3]. Due to the semienclosed space characteristics of highway tunnels, its visual pressure and monotonous environment are more likely to aggravate the fatigue and irritability of drivers, resulting in decreased driver alertness, inattention, and performance [4][5][6]. At the same time, the impact of tunnel internal accidents is usually greater than that of open sections owing to the closed characteristics [7,8]. e lighting environment directly affects the driver's information cognition and then affects the driving safety.
ere is no natural light to provide lighting in the tunnel. e driver can only obtain the visual capture information through the artificial light source [9]. erefore, it is imperative to research tunnel lighting from the perspective of traffic safety. e traditional tunnel lighting design only pays attention to whether the driver can identify the obstacles and does not consider the influence of the light source on human physiology and psychology. However, studies have shown that light not only provides us with visual information, but also produces nonvisual biological effects that affect many physiological and behavioral processes such as emotion and activity [10]. In particular, CIE "position statement on nonvisual effects of light" on Oct. 3, 2019, clearly stated that "lighting should consider the visual and nonvisual effects of light comprehensively." erefore, in the tunnel interior lighting settings, we should fully consider the impact of light on human visual and nonvisual biological effects, to ensure driving safety as far as possible. According to the current tunnel lighting standards, the recommended luminance values for different tunnel areas are given [11]. However, the lack of guidance on the selection of correlated color temperature (CCT) and color rendering index (CRI) of tunnel lamps has led to random settings of CCT and CRI of tunnel lighting, which has a negative impact on the safety of tunnel operations. erefore, from the perspective of spectrum, this paper studied the effects of CCT and CRI on human visual and nonvisual biological effects under mesopic vision category.
After decades of in-depth research, the influence of spectrum on visual and nonvisual biological effects has been proved to exist [12,13]. In the current research, people think that the nonvisual effect is mainly related to retinal ganglion cells (ipRGCs) containing melanin protease [14]. However, existing studies have shown that rods and cones also affect ipRGCs. e proportion of these three kinds of cells in biological effects is not clear, but it is certain that their proportion will change with the change of light intensity [15]. At the same time, their spectral sensitivity is different, which means that under different brightness, the influence of spectrum on visual and nonvisual effects is different. Christian et al. found that the alert response to light is related to the wavelength. Shortwavelength light (460 nm) is more efficient than long-wavelength light (550 nm) to reduce sleepiness at night [16]. Phipps-Nelson et al. found that blue light exposure at night can reduce drowsiness [17]. Sahin and Figueiro have shown that red light can improve human alertness more than blue light in the afternoon [18]. Lisdiani and Yassierli studied the effect of blue light on human alertness at night and found that adding the blue light to the light environment can improve the alertness of office workers at night [19]. Figueiro et al. found that exposure to red light not only increased objective alertness but also significantly reduced the reaction time (i.e., improved performance) when performing GO/NOGO tasks at night [20].
At present, the study of the influence of the spectrum on human photobiological effect is usually conducted under photopic vision category. However, existing research shows that for human, even if exposed to low-intensity light, many physiological processes of human will be affected. Vision is a complex process of perception and thinking. Human eyes change with the change of light brightness in the visual environment. erefore, it is unreasonable to apply the conclusion under photopic vision category to the mesopic vision category such as road lighting. Decision makers and road lighting industry need to evaluate the impact of light source on the biological effects of visual and nonvisual under mesopic vision category. e purpose of this study is to study the influence of spectral components on people from the perspective of visual effect and nonvisual effect under mesopic vision category and to find more suitable CCT and CRI parameters for tunnel lighting environment, improve the quality of tunnel lighting, and improve the safety of road lighting. ese conclusions are of great significance to improve road safety and provide reference for tunnel traffic safety lighting design.

Materials and Methods
Based on the complexity of nonvisual biological effects and the particularity of tunnel lighting, this paper first selects the color coordinate Z-value of 1931 CIE-XYZ standard chroma system, circadian rhythm factor (CAF), perception luminance, and blue light hazard factor to evaluate the influence of spectrum on visual and nonvisual biological effects. e theoretical results show that the spectra with high CCT and CRI correspond to higher Z-value, CAF-value, perceived brightness, and blue light hazard factor.

LED Spectrum Used in Road
Lighting. At present, tunnel lighting mainly uses the blue light to excite the yellow phosphor to make the light-emitting diode (LED), whose spectrum presents the double peak characteristic. e CCT and CRI of the light source are determined by the spectral content at different wavelengths. In the experiment, seven kinds of CCT are selected, which are 3000 K, 3500 K, 4000 K, 4500 K, 5000 K, 5700 K, and 6500 K. Our previous research shows that CRI2012 has a good evaluation effect on tunnel lighting environment from several current color rendering evaluation indexes [21]. erefore, this paper uses CRI2012 to study the CRI of tunnel led. e brightness of the light source used in the tunnel is generally between 60 and 90. In order to analyze the spectra of different CRIs, four CRIs (65, 75, 85, and 95, resp.) were selected. SPDs of different CCTs and CRIs are shown in Figure 1 and measured using a I1Pro spectroradiometer. Figure 1(a) shows the spectra of seven CCTs with a CRI of 85, and Figure 1(b) shows the spectra of four CRIs with a CCT of 3000 K. Other spectra have the same trend. In the theoretical calculation and practical experiment, the spectrum value measured under the same tunnel illumination brightness is used.

Blue Light Content.
e blue wavelength part of the light source is not the main sensitive band of human eye sensitivity, which has little effect on improving the efficiency of the light source, but plays a certain role in the CCT and CRI, so it is an indispensable wavelength for the light source with good light color performance. In recent years, relevant theoretical studies have shown that the intermediate vision theory, nonvisual light biological effect and blue light damage are directly related to the blue light wavelength content, or even play a major role. CIE-XYZ system was defined by CIE in 1931 to determine the three primary colors (red, green, and blue) in the spectrum and then to represent all colors in the spectrum on a two-dimensional plane. Y and Z represent the relative values related to red, green, and blue, respectively, and Z � 1 − (X + Y). In this paper, the chromaticity coordinate Z-value is used to evaluate the blue light content in the spectrum. Figure 2 shows the chromaticity coordinate Z-value of the spectrum shown in Figure 1.
e Z-value in Figure 2 represents the content of blue light components corresponding to different spectra. Fitted is to use the fit linear function of Origin software to obtain the relationship between Z and CCT by linear fitting with least square method. R 2 represents the square of correlation coefficient. It can be seen from Figure 2 that for the LED spectrum now used, the blue light content of the spectrum shows an obvious linear increasing trend with the increase of CCT and CRI.

Circadian Action Factor.
In recent years, CAF is regularly used to quantitatively reflect the influence of the spectrum on human nonvisual biological effects [15,[22][23][24]. In workplaces, lights with high CAF values can enhance workers' level of vigilance and excitation, while in bedrooms, the light with low CAF values are required to help people relax [23,25,26].
where L ed (λ) denotes the SPD of the light source. E (λ) denotes the visual spectral luminous efficiency function. C (λ) denotes the spectral circadian efficiency function. K c denotes the maximum spectral luminous efficiency. E (λ) refers to the spectral luminous efficiency function of the human eye. Photoreceptor cells in the retina can be divided into two types: cones and rods. e luminance of tunnel lighting belongs to the range of mesopic vision. At this time, the rod cells and cone cells work at the same time and the specific gravity changes with the luminance. At the same time, due to the different spectral sensitivity curves of the two kinds of cells, the spectral luminous efficiency function also changes with the luminance. Our previous research shows that it is reasonable to use the MES-2 model to calculate the mesopic visual curve in the tunnel lighting area [27]. erefore, in order to better analyze the nonvisual biological effects of different spectra under tunnel illumination, we use the MES-2 model to calculate the mesopic visual curve corresponding to different spectra and take it as E (λ). e form of the MES-2 model is as follows: L mes,n � m 2,(n−1) L p K m ′ + 1 − m 2,(n−1) L s K m m 2,(n−1) K m ′ + 1 − m 2,(n−1) K m , m 2,n � 0.3334 log L mes,n + 0.767, 0 ≤ m 2,n ≤ 1, where V ′ (λ) is the scotopic spectral luminous efficiency function, and K m ′ is its peak value. V (λ) is the photopic spectral luminous efficiency function, and K m is its peak value. L ed (λ) is the SPD of the light source. L s is the scotopic luminance. L p is the photopic luminance. Equations (4) and (5) are iterative processes, n is an iteration step, and m 2,end is the final iteration output. V mes (λ) is the mesopic spectral luminous efficiency function. Substituting the above L ed (λ) spectral curve into the formula calculation, the mesopic spectral luminous efficiency curve corresponding to each spectrum is shown in Figure 3. Figure 3 shows the mesopic spectral luminous efficiency curve corresponding to the spectrum. It can be seen from Figure 3 that the difference is slight between mesopic  e low CCT and low CRI spectrum correspond to a higher mesopic spectral luminous efficiency curve.
At present, C (λ) is mainly drawn by the influence of different wavelengths of light on a certain physiological parameter of the human body. ere are four main C (λ) models: Brainard model based on melatonin inhibition, apan model based on melatonin inhibition, Jiaqi Ju model based on heart rate change, and Berman model based on pupil size change [22,[28][29][30]. Recent studies have shown that the nonvisual biological effect is mainly caused by the action of ipRGCs, which can transmit the induced light signal to the supraoptic nucleus and then control the secretion of melatonin in pineal gland [21,[31][32][33][34]. Melatonin can increase fatigue and accelerate the state of sleepiness, which leads to poor reaction ability and judgment ability and affects automatic function [35]. e nonvisual bioeffect evaluation curve based on melatonin inhibition is shown in Figure 4. According to the experimental results of Brainard and apan on the spectrum of photoinduced melatonin inhibiting several wavelengths, Gall defines a circadian rhythm action function C (λ) [36].
e CAF values corresponding to different spectra are obtained by substituting E (λ) and C (λ) into the circadian rhythm factor calculation formula, as shown in Figure 5.
CAF values in Figure 5 represent the values of circadian rhythm factors corresponding to different spectra. It can be seen from Figure  rendering. at is, with the increase of color temperature and color rendering, the circadian rhythm factor corresponding to the spectrum increases, and the nonvisual biological effect increases.

Perception Luminance.
With the development of lighting technology, the composition of light environment is becoming more and more diversified. It is very meaningful to study the perception of human eyes to the light environment, especially for the environment with low brightness such as road lighting. Traditional luminance meter is only used to measure the brightness of the object under test, and it is an objective physical quantity, not the brightness that the human eye can actually perceive. In order to better analyze the applicability of different spectra in road lighting environment, the human eye perception luminance corresponding to different spectra in the mesopic vision is calculated. e calculation formula of human eye perceived luminance under mesopic vision is as follows: where L pl is the perceived luminance of the human eye under mesopic vision, L b is the brightness of the light environment,   Journal of Advanced Transportation and V mes (λ) is the mesopic visual curve shown in Figure 3. e results of normalization are shown in Figure 6. e results show that the white LED spectrum with high CCT and CRI can provide higher human eye perception luminance.

Blue Light Hazard.
Due to the production process, production cost, and other factors, most of the road lighting uses blue light to excite yellow phosphor to produce white light. As blue light is the main light-emitting device, the blue spectrum component of white LED spectrum accounts for a large proportion, thus forming the blue light problem. And more and more electrical designers and ordinary consumers know and concern. ere is no requirement for blue light hazard index in current main road lighting design specifications and standards. e blue light harm of LED lamps is controllable. Only by correctly understanding and fully considering the blue light harm of LED lamps can the blue light harm be avoided correctly and reasonably. In the current research, the calculation formula of the blue light damage efficiency K b of light radiation is as follows: where L ed (λ) represents the SPD of the light source, B (λ) represents the spectral weighting function of retinal blue light damage, E (λ) represents the spectral luminous efficiency function of human vision, and K e is its maximum spectral luminous efficiency. By analogy with the above circadian rhythm factor calculation process, we still use the mesopic visual spectral luminous efficiency curve in Figure 3 as E (λ). Figure 7 shows the spectral weighting function of retinal blue light damage B (λ).
Substituting the LED spectrum into the above blue light hazard efficiency calculation formula, the blue light hazard factors corresponding to different spectra are obtained, as shown in Figure 8.
From the fitting results in Figure 8, it can be seen that the blue light hazard factors of LED with different spectra show an obvious linear increasing trend with the increase of CCT and CRI. It means that with the increase of CCTand CRI, the damage of blue light caused by the spectrum is more serious.

Selection of Experimental
Indicators. According to the above theoretical analysis, we come to the conclusion: in the intermediate vision, high CCT and high CRI spectra with high blue light components can provide higher perceived luminance for human eyes, but will have a greater impact on human nonvisual photobiological effects, and the potential blue light harm is also higher. In the actual tunnel driving environment, the influence of different spectra on human visual efficacy needs to be further studied. erefore, we select the critical flicker frequency (CFF) [37][38][39][40] and reaction time as the characterization parameters of visual experiment and analyze the changes of CFF and reaction time under different light environments through relevant experiments, so as to realize the evaluation of driver's visual performance under different spectra.
As an organ that directly feels light stimulation, the change of eye activity can most directly reflect visual fatigue. e research of Iwasaki and Akiya confirmed that the decline of CFF reflects the weakening of retinal function and measures the fatigue degree of human eyes. As a reflection of high-level neural activity and physiological state, it has been widely used to detect the activity level and arousal level of cerebral cortex in medicine in recent years [41][42][43]. With the increase of flash frequency, the phenomenon that human eyes regard flash light as continuous light is called flash fusion, and the minimum stimulus frequency causing fusion is called CFF. e reason for this phenomenon is that after the light stimulation stopped, the nerve excitation caused by it did not disappear immediately, but left a certain trace in the cerebral cortex, so the visual image did not disappear immediately, but stayed for a period of time, resulting in visual residue. Visual traces give us the illusion that light is continuous. e lower the CFF value, the greater the fatigue degree [44]. Compared with other indexes, CFF is simple and accurate in measuring visual fatigue.
Detection methods based on stimulus response ability are often used to evaluate driving behavior [45][46][47][48]. e process of human visual response includes five stages: investigation, discovery, recognition, recognition, and decision-making. Reaction time is the time from the beginning of stimulation to the beginning of observer reaction. At the same time, reaction time reflects the comprehensive influence of visual psychology and physiology on driving visual behavior in the process of road lighting, and is an important evaluation index for early recognition of obstacles and dangerous situations [49]. e main purpose of this experiment is to test the effect of CCT and CRI on human eye performance in tunnel driving environment.

Parameter Settings.
In order to study the effect of CCT and CRI on reaction time and CFF, three experimental parameters were set up, including CCT, CRI, and target. Seven kinds of CCTs, four kinds of CRIs, and nine kinds of target sizes were selected in this paper.
Spectra corresponding to seven CCTs (3000 K, 3500 K, 4000 K, 4500 K, 5000 K, 5700 K, and 6500 K) and four CRIs (65, 75, 85, and 95) were selected to study the influence of spectra on visual efficiency. ree kinds of mesopic visual luminance are set: 1 cd/m 2 , 2 cd/m 2 , and 3 cd/m 2 , which is consistent with the actual luminance range of the tunnel.
In the experiment, the red Landolt loop is used as the target. Our previous experiments found that the red target can be detected quickly, but its gap direction is relatively difficult to distinguish. Before the experiment, we set up nine groups of Landolt loop of different sizes to make sure that everyone could see the orientation of the targets, and chose the size as small as possible. e target is shown in Figure 9.
We set the target luminance according to the current road lighting requirements. e target contrast can be calculated by (7): where C is the target-background contrast, L t is the target luminance, and L b is the background luminance. In general, the standard of road lighting has a contrast of 0.4. erefore, Journal of Advanced Transportation we set three groups of target luminance corresponding to three kinds of background luminance. Table 1 shows the target luminance. In practice, we use laptops to display target. e target luminance is determined by changing the screen luminance of the laptop and detecting whether the luminance value of the Landolt loop meets the requirements by Konica Minolta CS-2000.

Experimental Setup.
According to the principle of tunnel driving, a tunnel model with a width of 5 m, a height of 3 m, and a length of 10 m was built. In order to reduce the interference of external light as much as possible, an experimental platform was built to simulate the tunnel environment composed of two 2 m * 3 m and one 3 m * 3 m experimental plates, and a layer of wallpaper imitating the tunnel pavement and wallpaper were pasted on its surface. Figure 10 shows the real experimental situation corresponding to the seven CCTs. Figure 11 illustrates a schematic diagram of the experimental setup.
As shown in Figure 11, the observer is the experimental object. e luminaire's arrangement in the experiment adopts the mid-line symmetry method, and the angle is 36°, which conforms to the principle of luminaire arrangement and the requirement of the illumination angle in the tunnel. Flicker fusion apparatus is a bright spot scintillator named EP-403 produced by Beijing Lange Technology Company, which is used for CFF measurement. e adjustment range of flicker frequency is 8 Hz∼60 Hz, and the display accuracy is less than 0.1 Hz. Lenovo X395 is a laptop for displaying  Journal of Advanced Transportation targets. Bluetooth keyboard and laptop work together as a timing tool for subjects to respond to targets. PC is the computer that the experimenter uses to record the reaction time and CFF. e wallboard is a light barrier.

Procedure Specification.
Forty subjects with normal vision (or corrected vision) with driver's licenses participated in the study, including 20 women and 20 men. e subjects were healthy and drug free. ey all had normal color vision in terms of the Ishihara test. All had regular sleep-wake cycles with no complaints of sleep disorders (Pittsburgh sleep quality index <5). All subjects gave written informed consent. In order to avoid mutual interference, CFF and reaction time experiments were carried out, respectively. For at least three nights prior to the study, subjects were required to keep a regular sleep-wake cycle (sleeping from 23.00 to 07.00 h). For 24 h preceding each study session and for its duration, subjects refrained from caffeinated drinks, alcohol, excessive exercise, and bright light. e subjects participated in the experiment within a week, and they were asked to keep their daily life as consistent as possible for a week. e experimental process is as follows.
Firstly, the experiment adjusts the LEDCube to the light environment required. It should be noted that all light sequences in this experiment are random. Subjects were asked to remain seated and avoid head movement. ey were instructed to keep their eyes open and focus their eyes on the center of the screen.
Secondly, the subjects were asked to measure CFF at the beginning of the CFF experiment. e frequency of CFF measurement was from low to high once and from high to low again. e CFF value recorded by the experimenter was the average value of the two measurements. CFF values were recorded at 5, 10, and 15 minutes using the same method. ese times also correspond to the actual travel time in the tunnel.
irdly, the subjects were told to rest. After 5 minutes, the subjects repeated the experiment until all the light environment experiments were completed or the experiment time was over.
ere are some differences between the CFF experiment and the reaction time experiment. We have made 30 groups of videos in advance, and a visual target will appear at 5, 10, and 15 minutes of the video. In order to avoid the subjects recording the direction of the visual target, the direction of the visual target in any video and at any time is set to be different, and the screen is in the black state for the rest of the time to reduce the interference to the experiment. e subjects had 10 seconds to distinguish the direction of the visual target after the appearance of the visual target, and the visual target would disappear after 10 seconds. After the visual target appears, the PC will start timing until the subject presses the corresponding position of the Bluetooth keyboard to finish timing. e subjects were told to identify the correct direction of the target as much as possible and then press the keyboard, and the wrong data will be discarded.

Results and Discussion
In order to familiarize the subjects with the whole experimental process, three preliminary experiments were being conducted in a random lighting environment before the formal experiment begins, and those results are not be considered in the final data.
rough the experimental program, the CFF and reaction time values of 40 subjects under different light conditions were obtained.

Experimental Results and Discussion of CFF.
In the experiment, 0 min of each experiment was used as the initial value, and the initial value minus the CFF value at 5 min, 10 min, and 15 min as the CFF drop value. Higher CFF drop value suggests higher fatigue. e average CFF drop value with time in 40 subjects under different light environments is shown in Figure 12. Figure 12 shows the average CFF drop value over time for all observers in different light environments. It can be seen from Figure 12 that the CFF drop value increases with time, which indicates that the fatigue of the experimenter is increasing with time. When the luminance and CRI were fixed, the CFF drop value decreases with the increase of CCT. When the luminance and CCT were fixed, the CFF drop value decreases with the increase of CRI. When the CCT and CRI were fixed, the decrease of CFF decreases with the increase of luminance. is shows that improving the CCT and CRI of tunnel lighting has a positive effect on improving the driver's visual performance.
In order to study the correlation between characteristic variables (CCT, CRI, luminance, and experimental duration) and target variables (CFF drop value and reaction time). e correlation coefficient and P value were obtained by Pearson's correlation analysis of experimental data by SPSS software.  Table 2 shows the results of correlation and significance analysis with CFF drop value as target variables and CCT, CRI, luminance, and experimental duration as characteristic variables. In terms of correlation, CCT, CRI, and luminance are positively correlated with CFF drop value, while experimental duration is negatively correlated with CFF drop value. On the extent, experimental duration > luminance > CCT > CRI, in which the experimental duration corresponds to high correlation, CCT and luminance correspond to low correlation, and CRI corresponds to weak correlation.
is shows that the experimental duration has the highest impact on visual fatigue. Without changing the brightness of the tunnel light source, we can use the light source with high CCT and CRI characteristics    to alleviate the driver's visual fatigue to a certain extent. P value showed that CCT, CRI, brightness, and experimental duration were significantly correlated at the level of 0.01, which showed that the results were statistically significant.

Experimental Results and Discussion of Reaction Time.
e average reaction time of 40 subjects in different light environments with time is shown in Figure 13. Figure 13 shows the average reaction time of all observers in different light environments. It can be seen from Figure 13 that the reaction time increases with time, which indicates that the fatigue of the experimenter is increasing with time. When the luminance and CRI were fixed, the reaction time decreases with the increase of CCT. When the luminance and CCT were fixed, the reaction time decreases with the increase of CRI. When the CCT and CRI property were fixed, the reaction time decreases with the increase of luminance. is shows that when the tunnel lighting environment adopts high luminance, high CCT, and high CRI light source, the driver has a faster response when encountering dangerous situations, and the probability of traffic accidents will also be reduced. e experimental results are consistent with the above theoretical results. Table 3 shows the results of correlation and significance analysis of reaction time as target variables and CCT, CRI, luminance, and experimental duration as characteristic variables. Table 3 shows the results of correlation and significance analysis of reaction time as target variables and CCT, CRI, luminance, and experimental duration as characteristic variables. In terms of correlation, CCT, CRI, and luminance are negatively correlated with reaction time, while experimental duration is positively correlated with reaction time.
On the extent, luminance > experimental duration > CCT > CRI, in which the luminance corresponds to high correlation, experimental duration, CRI, and CCT correspond to low correlation. is shows that the experimental duration has the highest impact on fatigue. is means that to create a safe road environment, improving luminance is still the best choice. However, considering the tunnel energy saving, it is also a desirable method to improve the CCT and CRI of lamps.

Discussion on the Influence of the Spectrum on Visual and
Nonvisual Effects. In order to study the correlation and significance between the above variables (CCT, CRI, Zvalue, perceived luminance, biological rhythm, blue light hazard, CFF decline value, and reaction time), in this paper, we use IBM SPSS software to do Pearson's correlation analysis on the experimental data, and the results are shown in Table 4.
From the table, we can see that compared with CCT and CRI, Z-value has a higher correlation with perceived brightness, biological rhythm, blue light hazard, and other parameters, reaching 0.939, 0.974, and 0.976, respectively.
is means that the content of blue light can better reflect the influence of the spectrum on biological effect. is means that future lighting standard design department can consider Z-value as one of the standards. From the perspective of blue light hazards, high CCT and CRI light sources have negative effects on visual health. However, it can improve the perceived brightness of human eyes, which are conducive to driving behavior under low illumination. At the same time, this study shows that the high CCT (6500 K) and high CRI (95) spectrum can improve the driver's visual performance  in the tunnel lighting environment with short lighting period. Current studies have shown that the damage and nonvisual biological effects of blue light are closely related to light intensity and lighting duration. Long time exposure to blue light can promote the visual center to make the ciliary muscle movement into a tight state, make people visual fatigue and blurred vision, and also affect melatonin secretion and circadian rhythm. is means that we can use high CCT (6500 K) and high CRI (95) light sources in the road lighting environment with short lighting period, and consider using low CCT and CRI light sources in other lighting environment with long lighting period.

Conclusions
In this paper, the influence of spectrum on driver's visual and nonvisual biological effects under intermediate vision is studied. Firstly, combined with the influence of the spectrum on human visual and nonvisual biological effects, it is analyzed that the existing photometric test methods are not suitable for road traffic lighting and other intermediate visual brightness environment. en, the intermediate visual spectral luminous efficiency curve is applied to the evaluation methods of nonvisual biological effects, perceived brightness, and blue light hazards. e results show that high CCT and CRI spectra with high blue light content can improve the perceived brightness, have stronger impact on human nonvisual effects, and cause more serious damage to blue light radiation. At the same time, the research shows that Z-value can better reflect the biological effect of the spectrum, and the lighting standard design department can consider Z-value as one of the standards in the future. en, a driving simulation experiment was carried out to test the driver's visual performance under different spectral illumination. Changes of CFF and RTS reflect the visual performance of drivers in different light environments. e results show that the light source with high CCT and high CRI corresponds to lower CFF drop value and RTS; that is to say, the driver's visual fatigue is lower, the visual effect is higher, and the driving behavior is more favorable under this light condition.
rough the study of the driving behavior of tunnel, it is found that the higher blue light content spectrum has a more positive effect on the visual efficiency when exposed to low light for a short time. erefore, from the perspective of driving safety, this paper proposes to use LED with high CCT (6500 K) and high CRI (95) in the inner area of the tunnel to improve the driver's visual effect and driving safety. is paper provides a reasonable reference for tunnel lighting specification and tunnel lighting design department, which help to ensure personal safety and avoid traffic accidents inside the tunnel.
Although the current results are promising, they also have limitations. In the road environment with long lighting period, whether the LED light source with high CCT and CRI is harmful to driving behavior needs to be further studied. At the same time, for LED lighting, further experiments are needed to evaluate whether the circadian light value and blue light harm defined by the simplified model are consistent with the real situation. e current and previous studies have not fully clarified the physiological mechanism related to these results, and further research is needed.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.