A new method to identifying optimal adjustment strategy when the car cockpit is uncomfortable: optimal state distance method

Establishment of cockpit comfort evaluation model

Cockpit comfort evaluation experiment

In this article, we conducted a cockpit comfort evaluation experiment and a cockpit comfort adjustment experiment. All experiments involved in this article have obtained the ethical certification of the Ethics Committee of the School of Automotive and Transportation of Xihua University (2021LL(01)).

In the cockpit comfort evaluation experiment, the Audi A6L (2021) was selected as the cockpit environment for experimental evaluation, as shown in Fig. 2. The experiment was carried out on the campus of Xihua University, and a road section was simulated as a real traffic line. The roadmap is shown in the red line in Fig. 3.

This experiment was conducted in sunny and windless weather. The outdoor temperature was approximately 33 degrees Celsius. Since constant speed is the most common driving condition in real life, the experimental conditions for this experiment were all set to constant speed, with the speed controlled at about 40 km/h. Data were collected as previously described in Yang et al. (2022a) and Yang et al. (2022a). The noise measurement instrument is a precision noise level meter, a model AWA6291 handheld real-time signal analyzer. The cockpit light is measured with a digital illuminance meter and the cockpit temperature is measured with a resistance thermometer.

In the experiment, five experts in the field of the car were invited to rate the comfort of the interior environment of the car cockpit. They are all able to provide written consent to an informed form. Table 1 contains brief information about them. Combined with expert opinions, the comfort evaluation scale is shown in Table 2.

Table 1:

Brief information of experts.

Expert number	Name	Gender	Position	Field of work
Expert I	Qiping Chen	Male	Professor, East China Jiaotong University, Ph.D.	Intelligent Vehicles
Expert II	Yanli Yin	Female	Associate Professor, Chongqing Jiaotong University, PhD	Intelligent Vehicles
Expert III	Xiaoliang Pan	Male	Senior Engineer, Changan Automobile Company	Vehicle Engineering
Expert IV	Peilong Cheng	Male	Engineer, Tesla Motors	Intelligent Vehicle
Expert V	Yanhong Yang	Female	Senior Engineer of Auto Electric Control Company	Auto Electric Control

DOI: 10.7717/peerjcs.1324/table-1

Table 2:

Comfort evaluation grading standard.

Comfort evaluation level	Scoring range
Intolerable	[0,2]
Very uncomfortable	(2,4]
Uncomfortable	(4,6]
Slightly uncomfortable	(6,8]
Comfortable	(8,10]

DOI: 10.7717/peerjcs.1324/table-2

With the principle of the single factor variable, the sound decibels, light illumination, and temperature in the car were changed respectively to explore the changing rules of comfort. The corresponding mathematical model is obtained by fitting the experimental data as shown in Table 3.

Table 3:

Single indicator mathematical model.

•  y1 represents the evaluation value of cockpit noise and vibration comfort; N represents the evaluation value of noise obtained by using sound level A as the evaluation method, in dB(A).
•  y2 represents the evaluation value of cockpit optical environment comfort; C stands for illuminance in the cockpit, in lx.
•  y3 represents the evaluation value of cockpit thermal environment comfort; T stands for cockpit temperature in °C
Indicator	Mathematical model
Sound decibels	y1 =  − 0.150N + 16.795
Light illumination	y2 =  − 0.000031C2 + 0.033C + 0.597
Temperature	y3 =  − 0.054T2 + 2.649T − 22.856

DOI: 10.7717/peerjcs.1324/table-3

According to the fitting curves of the acoustic environment, optical environment, and thermal environment, the maximum value of comfort evaluation corresponding to the single environment of the cockpit can be obtained as follows.

• Acoustic environment: (50dB(A), 9.295)

• Optical environment: (532lx, 9.379)

• Thermal environment: (24.5 °C, 9.631)

Identify the optimal strategy for cockpit comfort adjustment

After evaluating the comfort of the car cockpit, if the comfort score Y<8 at a certain time, the comfort of the car cockpit should be adjusted. If the influence degree of three indicators on cockpit comfort at a certain time can be determined, the optimal strategy of cockpit comfort adjustment can be obtained.

Cockpit comfort adjustment experiment

In this experiment, 90 groups of common environmental conditions in the cockpit were randomly established. After evaluation by the comfort evaluation model, 80 groups of conditions with comfort score Y<8 were obtained. The 80 groups of cockpit environment conditions were used for the experiment. After five experts experienced the 80 groups of cockpit environment conditions, the optimal cockpit adjustment strategy was given for the corresponding conditions. Some experimental results are shown in Table 4.

Optimal state distance

To measure the degree of influence of the three indicators of the acoustic environment, optical environment, and thermal environment on the comfort of the cockpit at a certain time, the distance of an indicator from its optimal state value is used to reflect the necessity of adjustment of the indicator, to propose the optimal adjustment strategy of cockpit comfort. In this article, an optimal state distance (OSD) theory is defined as follows.

Definition: Assume that S = (s₁, s₂, …, s_n) is a feature vector representing an index, where s₁, s₂, …, s_n are n attribute vectors constituting the working condition environment. P = (p₁, p₂, …, p_n) is the feature vector representing the optimal state of the index, where p₁, p₂, …, p_n represents the n attribute vectors constituting the optimal state of the index. OS distance is used to measure the distance between an index and its optimal state, as shown in Eq. (3). The larger the OS distance value, the farther an indicator deviates from its optimal state value. (3)${\displaystyle D\left(S,P\right)=\sqrt{{\left(\frac{s_1}{p_1}-1\right)}^2+{\left(\frac{s_2}{p_2}-1\right)}^2+\cdots +{\left(\frac{s_n}{p_n}-1\right)}^2}}$

To make the distance Eq. more universal in engineering applications, the modified parameter K is introduced, and Eq. (4) is obtained. (4)${\displaystyle D\left(S,P\right)=K\sqrt{{\left(\frac{s_1}{p_1}-1\right)}^2+{\left(\frac{s_2}{p_2}-1\right)}^2+\cdots +{\left(\frac{s_n}{p_n}-1\right)}^2}}$

In Eq. (4), the correction parameter K ∈ (0, 1).

The optimal cockpit adjustment strategy is identified by OSD

To propose the optimal cockpit adjustment strategy, it is necessary to identify the indexes that most affect cockpit comfort in the acoustic environment, optical environment, and thermal environment. In this article, the optimal state distance is used to identify. There are three steps to establish the OSD model. The identification results of the established model are compared with the adjustment target suggested by experts. If the correct rate is greater than 90%, it means that the model is successfully established. The specific flow of this model is shown in Fig. 4.

Step 1. Set the feature vector and calculate the OS distance of each indicator

The physical parameter value and the single-factor comfort evaluation value of each index are taken as the two attribute vectors that constitute the characteristic vector of the index. For example, the characteristic vector of the acoustic environment: S_A = (s_A1, s_A2), where s_A1 is the physical parameter of noise, the unit is dB(A), and the s_A2 is the comfort evaluation value of the acoustic environment; the optimal state characteristic vector of the acoustic environment: P_A = (p_A1, p_A2), where p_A1 is the optimal state value of noise and its value is 50 dB(A). p_A2 is the optimal comfort evaluation value of the acoustic environment and its value is 9.295.

Therefore, the OS distance of the acoustic environment under a certain cockpit operating condition is shown in Eq. (5). (5)${\displaystyle D_A\left(S_A,P_A\right)=K_A\sqrt{{\left(\frac{s_{A1}}{p_{A1}}-1\right)}^2+{\left(\frac{s_{A2}}{p_{A2}}-1\right)}^2}}$

Similarly, the OS distance of the optical environment and the thermal environment under a certain cockpit working condition D_O, D_T as shown in Eqs. (6) and (7). (6)${\displaystyle D_O\left(S_O,P_O\right)=K_O\sqrt{{\left(\frac{s_{O1}}{p_{O1}}-1\right)}^2+{\left(\frac{s_{O2}}{p_{O2}}-1\right)}^2}}$ (7)${\displaystyle D_T\left(S_T,P_T\right)=K_T\sqrt{{\left(\frac{s_{T1}}{p_{T1}}-1\right)}^2+{\left(\frac{s_{T2}}{p_{T2}}-1\right)}^2}}$

Step 2. Set the correction factor to optimize the OS distance

For the OS distance to reflect the expert’s adjustment recommendations more comprehensively, the correction parameters need to be set. The weight vector ω = (0.42, 0.23, 0.35) representing the effects of the acoustic environment, optical environment, and thermal environment on cockpit comfort is introduced above, that is, the weights ω_A, ω_O and ω_T of the influence of the acoustic environment, optical environment and thermal environment on cockpit comfort are 0.42, 0.23, and 0.35, respectively. However, considering that the current car in the acoustic environment, optical environment, and thermal environment adjustment difficulty is different, and experts have different adjustment tendencies for the three indicators; therefore, it is necessary to introduce a coefficient θ that can represent the expert adjustment tendency, combined with expert opinions, set the expert adjustment tendency coefficients θ_A, θ_O and θ_T corresponding to the acoustic environment, optical environment, and thermal environment to 0.27, 0.35, and 0.38 respectively.

Set the correction factor K of the OS distance of a certain indicator of the cockpit as shown in Eq. (8). (8)${\displaystyle K=\omega \times \theta }$

The correction coefficients corresponding to the acoustic environment, optical environment, and thermal environment are calculated: ${\displaystyle K_A={\omega }_A\times {\theta }_A=0.1134}$ ${\displaystyle K_O={\omega }_O\times {\theta }_O=0.0805}$ ${\displaystyle K_T={\omega }_T\times {\theta }_T=0.1330}$ Define the set U = D₁, D₂, D₃, where D₁, D₂, D₃ can be obtained by Eq. (9). ${\displaystyle D_1=max\left\{D_A,D_O,D_T\right\}}$ (9)${\displaystyle D_3=min\left\{D_A,D_O,D_T\right\}}$ ${\displaystyle D_2=C_U\left(D_1\cup D_3\right)}$

By calculating the OS distance of the three indicators, the index with the largest output distance is the goal that urgently needs to be adjusted at a certain time in the cockpit, that is, it is j₁ for the optimal adjustment strategy of cockpit comfort. j₁ is expressed by Eq. (10). (10)${\displaystyle j_1=\left\{\begin{array}{c}\hfill A\hfill \\ \hfill O\hfill \\ \hfill T\hfill \end{array}\begin{array}{c}\hfill D_1=D_A\hfill \\ \hfill D_1=D_O\hfill \\ \hfill D_1=D_T\hfill \end{array}\right.}$ where A, O, and T represent that the cockpit should adjust the acoustic environment, optical environment, and thermal environment, respectively.

Step 3. Model accuracy optimization.

The defined ɛ is the OS distance difference, which is calculated as shown in Eq. (11). Set the accuracy of the distance difference ɛ_o = 0.01. (11)${\displaystyle \varepsilon =D_1-D_2}$

When calculating the OS distance, if it is ɛ ≤ ɛ_o, it is considered that the D₁ and D₂ are almost equal, and based on expert suggestions, the party with the largest expert adjustment tendency coefficient θ is preferred as the optimal cockpit adjustment strategy j₂. Since the expert adjustment tendency coefficients θ_A, θ_O and θ_T set above are 0.27, 0.35, and 0.38, respectively, θ_T > θ_O > θ_A can be obtained. j₂ is expressed by Eq. (12). (12)${\displaystyle j_2=\left\{\begin{array}{c}O{D}_3=D_T\hfill \\ T\left\{\begin{array}{c}\begin{array}{c}\hfill D_3=D_O\hfill \\ \hfill D_3=D_A\hfill \end{array}\hfill \end{array}\right.\hfill \end{array}\right.}$

In Eq., O and T represent that the cockpit should be adjusted to the optical environment and thermal environment, respectively.

Step 4. The final cockpit optimal adjustment strategy identification model

From the above derivation, combined with the analysis of the optimal cockpit adjustment strategy j₁, j₂, the final cockpit optimal adjustment strategy J is obtained as shown in Eq. (13). (13)${\displaystyle J=\left\{\begin{array}{c}\hfill j_1\hfill \\ \hfill j_2\hfill \end{array}\begin{array}{c}\hfill \varepsilon >{\varepsilon }_o\hfill \\ \hfill \varepsilon \le {\varepsilon }_o\hfill \end{array}\right.}$

The model was used to identify 80 groups of test conditions that needed to be adjusted for cockpit comfort, and the optimal adjustment strategy of cockpit comfort based on OS distance was obtained, which was compared with the adjustment strategy suggested by experts. The results are shown in Fig. 5.

Comparing the optimal comfort adjustment strategy J obtained by OSD with the expert-suggested adjustment strategy, it can be seen that the number of samples consistent with the expert’s recommendation according to the comfort adjustment strategy obtained by OSD is 76, and the total sample is 80, with a consensus rate of 95%. It can be seen that the model is successfully established and has a good identification effect, which can obtain the optimal strategy for cockpit comfort adjustment.

Cockpit comfort evaluation

In this article, starting from the three indicators of the acoustic environment, optical environment, and thermal environment that affect the comfort of the car cockpit, the car cockpit comfort evaluation experiment is carried out. The cockpit evaluation experiment created in this article simulates the real occupant riding state and fully considers the riding experience of experts. Through the fitting analysis of the experimental data, a single index comfort evaluation model is obtained. Then, by combining the comfort evaluation model of a single index, the cockpit comfort evaluation model was established. The model can calculate the comfort evaluation value Y of the current cockpit environment through the physical parameter values of sound decibels, light illuminance, and temperature in a certain cockpit working environment. When the cockpit comfort evaluation value is Y > 8, it is considered that the comfort of the cockpit comfort is good; Conversely, it is believed that the comfort of the cockpit still needs to be improved.

This article has a comprehensive evaluation of car cockpit comfort, is a comprehensive evaluation of comfort, and has certain innovative value in engineering applications. With the innovation and development of science and technology, people’s cockpit comfort requirements will become higher and higher. In addition to the acoustic environment, optical environment, and thermal environment, more indicators will be added, and this article provides a reference direction for future cockpit comfort evaluation.

Identify optimal adjustment strategies for cockpit comfort

Model validation

To verify the accuracy and usability of the model, 30 different cockpit environmental conditions were randomly set, and the comfort score of each working condition was obtained from the cockpit comfort evaluation model, and the cockpit working conditions that needed to be adjusted were compared with OSD identification and machine learning identification. The identification results are shown in Fig. 6. Table 5 has the specific experimental information for each group of experiments.

Combined with Fig. 6 and Table 5, it can be seen that the accuracy rate of the cockpit environmental adjustment strategy identified by OSD identification is 93.3%, and the accuracy rate is higher than all machine learning methods. The effect of OSD identification is more prominent (as shown in Fig. 7).

Model analysis

After the evaluation of cockpit comfort is completed, it is necessary to adjust the cockpit where comfort will be poor. In a certain cockpit working environment, it is often necessary to adjust the indicators that most affect the comfort of the cockpit to have a better adjustment effect. In this article, an OSD method is introduced to measure the distance that each indicator deviates from its optimal state value. Based on the OSD, the correction parameters are introduced in combination with the specific engineering application environment, and the optimal cockpit adjustment strategy identification model is established. Combined with the cockpit comfort adjustment experiment, it is known that the accuracy of the model is greater than 90%, and it has a good effect. Through this model, the indicators that have the greatest impact on cockpit comfort in a certain cockpit working environment can be identified, and the corresponding actuator can be adjusted to improve cockpit comfort.

Table 5:

The model validates the data information.

The ‘1, 2, and 3′in columns 6 to 10 of the table respectively represent optimal adjustment strategies.

•  1—Acoustic environment
• 2—Optical environment
• 3—Thermal environment
Number	Acoustic environmentdB(A)	Optical environmentlx	Thermal environment °C	Model comprehensive evaluation value	Experts recommend adjustment strategy	OSD identification adjustment strategy	SVM identification adjustment strategy	Random Forest identification adjustment strategy	XGBoost identification adjustment strategy
1	82.5	450.3	30	6.44	1	1	3	1	1
2	72.3	868.9	35.4	4.80	3	3	2	3	3
3	70.5	880.6	21.8	6.98	2	2	1	1	2
4	67.6	985.1	18.8	5.88	2	2	2	2	2
5	68.2	674.6	33	6.70	3	3	3	3	3
6	66.4	1011.7	27.6	5.86	2	2	2	2	2
7	63.5	698.5	12	3.96	3	3	1	2	3
8	77.2	582.6	36.4	4.27	3	3	3	3	3
9	79.6	654.3	30.8	6.50	1	1	3	1	1
10	59.6	939.9	26.8	7.24	2	2	2	2	2
11	77.6	518.2	34.3	5.63	3	3	3	3	3
12	64.8	495.4	37	4.09	3	3	3	3	3
13	69.4	599.6	33.9	6.33	3	3	3	3	3
14	79.5	999.3	30	5.03	1	2	2	2	2
15	84.2	422.7	22.6	6.62	1	1	3	1	1
16	78.2	860.3	33.8	5.25	2	2	1	3	3
17	69.1	949.4	23.5	6.62	2	2	2	2	2
18	76.5	877.5	26.4	6.60	1	1	1	1	1
19	71.5	211.7	27.8	7.01	2	2	3	2	2
20	67.9	985.1	18.8	5.87	2	2	2	2	2
21	62.6	939.9	26.8	7.06	2	2	2	2	2
22	74.5	799.3	30	6.73	1	1	1	1	1
23	64	927.9	31.2	6.48	2	2	2	2	2
24	64.2	931.1	34.5	5.35	2	3	2	2	3
25	78.3	856.8	12.4	3.60	3	3	1	1	3
26	76.2	1004.2	36.6	3.04	3	3	2	3	2
27	82.6	908.2	36.9	3.01	3	3	1	3	3
28	67.4	444.9	36.4	4.73	3	3	3	3	3
29	77.5	644.2	34.8	5.34	3	3	3	3	3
30	56	489.1	33.1	7.49	3	3	3	2	3

DOI: 10.7717/peerjcs.1324/table-5

The OSD method has excellent generalization performance and low computational complexity, which can also ensure high recognition accuracy when the number of target categories to be identified increases. Therefore, this method can also be applied to more complex engineering practices.

Dynamic adjustment for cockpit comfort

In the process of driving the car, the environment in the cockpit of the car is constantly changing, so the cockpit comfort is also constantly changing. After the evaluation of cockpit comfort and the identification of optimal adjustment strategies are realized, the cockpit comfort can be monitored and adjusted (as shown in Fig. 8).

In specific engineering practice, a cockpit comfort evaluation system can be installed in the ECU of a vehicle. The sensors collect the specific physical parameters of the acoustic, light and thermal environments and transmit them to the ECU for processing, and the ECU obtains the comfort evaluation value of the cockpit at a certain moment according to the cockpit comfort evaluation model. Then, according to the cockpit comfort optimal adjustment strategy model, the optimal control strategy is obtained, and the corresponding actuator is operated to adjust the comfort level. This facilitates the dynamic adjustment of the cockpit comfort to ensure occupant comfort.

Limitations and future work

Future work

In the future, our team will work to establish a more comprehensive and complete cockpit comfort evaluation system, establish more detailed and comprehensive experiments by increasing the number and types of car models, optimize the cockpit comfort evaluation model, improve the cockpit optimization adjustment strategy model based on optimal state theory, and strive to achieve dynamic adjustment of cockpit comfort and improve the occupant’s riding experience.