Skip to main content

Hierarchical Optimization of Landing Performance for Lander with Adaptive Landing Gear


A parameterized dynamics analysis model of legged lander with adaptive landing gear was established. Based on the analysis model, the landing performances under various landing conditions were analyzed by the optimized Latin hypercube experimental design method. In order to improve the landing performances, a hierarchical optimization method was proposed considering the uncertainty of landing conditions. The optimization problem was divided into a higher level (hereafter the “leader”) and several lower levels (hereafter the “follower”). The followers took conditioning factors as design variables to find out the worst landing conditions, while the leader took buffer parameters as design variables to better the landing performance under worst conditions. First of all, sensitivity analysis of landing conditioning factors was carried out according to the results of experimental design. After the sensitive factors were screened out, the response surface models were established to reflect the complicated relationships between sensitive conditioning factors, buffer parameters and landing performance indexes. Finally, the response surface model was used for hierarchical optimization iteration to improve the computational efficiency. After selecting the optimum buffer parameters from the solution set, the dynamic model with the optimum parameters was simulated again under the same landing conditions as the simulation before. After optimization, nozzle performance against damage is improved by 5.24%, the acceleration overload is reduced by 5.74%, and the primary strut improves its performance by 21.10%.

1 Introduction

Legged lander has been used for deep space exploration because of its high landing stability and terrain adaptability [1]. In order to isolate vibration and reduce load during soft landing, the legged lander generally uses the plastic material such as honeycomb as the main absorber to design the landing gear. However, the performances of these landing gears are unable to be adjusted during soft landing. In order to cope with complex landing terrain, larger design margin should be reserved, resulting in the heavier soft landing system [2]. With the continuous progress of deep space exploration, the terrain environment of interesting regions will be more complex and harsh, and landing in multiple regions to accomplish different detection missions may be needed. So it is required that the lander has better terrain adaptability and its landing gears are reusable.

Considering those requirements, Adaptive landing gear was proposed as a possible solution. Refs. [3,4,5] introduced hydraulic system, intelligent materials and pyrotechnics devices into the design of landing gear to realize adaptive control. Among them, magnetorheological damper (MR damper) is widely studied because of its cheerful prospect. In Refs. [2, 6,7,8,9], the single MR damper was designed and analyzed in detail, and the equivalent mathematical model of its characteristics was obtained. Refs. [10,11,12,13], which proposed a variety of control strategies for the lander with adaptive landing gears, proved the effectiveness of adaptive gears in enhancing soft landing performances. Previous studies mostly discussed the implementation of adaptive lander, and few concerned the soft landing performance optimization for the adaptive lander. But performance optimization is of great importance for the weight reduction of lander and it benefits the improvement of terrain adaptability. Existing researches about landers optimization mostly focus on conventional passive control lander [14,15,16]. Furthermore, the worst condition uncertainty caused by the change of design variables was ignored in the existing researches. And the selection or optimization of the parameters was just based on the typical condition, which leads to the instability of soft landing safety.

Aiming at the uncertainty of the worst condition, a hierarchical optimization method was proposed to update the worst condition dynamically during the progress optimizing the adaptive buffer parameters. First, a dynamic analysis model of adaptive lander was established, and its soft landing performance was analyzed and evaluated. Then the response surface was adopted to participate the iterative computation of hierarchical optimization. The lander with the optimized adaptive buffer was simulated. The results show that the optimization effectively improves the soft landing performance, which verifies the feasibility of the hierarchical optimization method.

2 Dynamic Model of the Lander

2.1 Configuration and Coordinate System Definition of Lander

Figure 1 shows the overall configuration of the lander studied in this paper, which consists of a main body and four symmetrically distributed landing gears (Figure 2). The main body is a mounting platform for various detecting instruments and control subsystem. All of the landing gears, with the same configuration and size, are composed of one primary strut, two secondary struts and one footpad. The connection between struts is realized through a universal joint, so as between struts and main body, while the footpad is connected with the primary strut by the ball joint. The primary struts are adaptive buffers, and the secondary ones use aluminum honeycomb core as buffer component. The relationship of secondary strut between the cushioning force fa and the buffering stroke da is shown in Figure 3, while the characteristic of primary strut will be discussed later. Take Refs. [17, 18] as reference, the structure parameters of the lander at touchdown is selected (Table 1).

Figure 1
figure 1

Configuration of the lander

Figure 2
figure 2

Main structures of the lander

Figure 3
figure 3

Force–stroke relationship of secondary strut

Table 1 Parameters of the lander at touchdown

The dynamic analysis model of soft landing was established by ADAMS, and gravity environment was set in the moon. The lander footpad numbering and the coordinates definition is shown in Figure 4, where Os-XsYsZs is global coordinate system, Oc-XcYcZc is centroid control coordinate system, αe is the equivalent slope of landing surface, the speed along Xs is vertical velocity vx and the speed along Xs is horizontal velocity vz. The rotation angles from Os-XsYsZs to Oc-XcYcZc in order of Z-X-Y are defined as: φ (rotation about the Xs), θ (rotation about the Ys) and ψ (rotation about Zs). The contact force between footpad and landing surface was simulated by nonlinear spring damping model and Coulomb friction model [15].

Figure 4
figure 4

Definition of coordinate systems

2.2 Adaptive Buffer and Its Control Strategy

Unlike the conventional landing gears, such as honeycomb core and air bag, buffer characteristics of adaptive buffer are able to be controlled by adopting some structures or intelligent material. So the adaptability of lander equipped with this kind of buffer can be improved. Even if the adaptive control system fails, the adaptive landing gear will degenerate to the conventional passive landing gear but not palsy, which ensures the safety and reliability of the landing system [11].

Considering the maturity of techniques, MR damper was chosen as carrier for adaptive control strategy. The main components of adaptive buffer are spring and MR damper, in which the spring provides restoring force, and MR damper provides damping force. The structure of adaptive buffer is shown in Figure 5. The magnetic field strength of the coil is changed by controlling the energizing voltage, so as to dynamically adjust the damping coefficient of MR damper.

Figure 5
figure 5

Sketch of adaptive buffer

According to the existing research results, the MR damper produces a large damping force at a relatively small velocity (about 0.1 m/s) [19, 20]. Considering that the touchdown velocity is above 3 m/s, the lander will slow down with a large acceleration overload, which will affect the stable operation of the instruments on board and is undesired. To preserve the transportability of the strategy, the conventional linear damping force model is adopted. The damping force of MR damper is controlled to keep a linear relation with the buffer velocity by adjusting the applied current. The equivalent force of adaptive buffer can be simplified as shown in Eq. (1) [21,22,23]:

$$ f = - c\dot{s} + ks, $$

where f is the equivalent force, c is the equivalent damping coefficient, k is the equivalent stiffness coefficient and s is the cushioning stroke of the buffer.

To ensure the controlling flexibility and promptness, a jump control strategy based on the minimum energy principle was adopted to realize the adaptive adjustment of the damping coefficient [13]. Considering the symmetry of the land model, the damping coefficient control function is shown as Eq. (2):

$$ c_{i} = \frac{{c_{\text{max} } - c_{\text{min} } }}{2}\text{sgn} \left[ {\left( { - \dot{\theta } + \dot{\psi }} \right)s_{i} } \right] + \frac{{c_{\text{max} } + c_{\text{min} } }}{2}, $$

where ci is the equivalent damping coefficient of damper i, and cmin, cmax are the lower and upper limit to be controlled.

In the whole simulation analysis of the soft landing, the angular velocity \( \dot{\theta } \), \( \dot{\psi } \) of the lander and the buffer speed of the main strut \( \dot{s} \) were monitored in real time through measurements. According to the Eq. (1), four cushioning forces are applied to primary struts, where the damping coefficient model is shown as Eq. (2). Finally, the independent feedback adjustment of the damping coefficients is realized.

In order to determine the initial buffer characteristic parameters of the buffer, the soft landing process is simplified as spring damping model. Under the ideal condition that four pads simultaneously touch the ground, it was required that there was no vibration during soft landing [24]. The damping ratio under above condition is chosen as 2.0 and the buffer is designed with a dynamic range ratio of 10 [25]. When the lander hits ground at the vertical speed of 3.5 m/s, the nominal buffer stroke is 0.075 m. Based on the above chosen parameters, the nominal stiffness coefficient and the maximum damping coefficient are determined according to the method described in Ref. [13]. The initial buffer characteristic parameters are listed in Table 2.

Table 2 Initial buffer characteristic parameters

3 Simulation and Analysis for Adaptive Lander

3.1 Indicators for Soft Landing

Considering the implementation requirements of landing exploration missions, the main concerns about soft landing performance are as follows.

  1. 1)

    Nozzle performance against damage. Landing on regions with rough terrain may damage the nozzle due to rugged landing surface, which affects the performance of the main engine. The minimum distance between the bottom of the nozzle and the landing surface is chosen as the evaluation index. The larger the index is, the better.

  2. 2)

    Acceleration overload. Considering the acceleration tolerance of astronauts and the instruments equipped on the lander, the overload during the soft landing should not exceed 15g to ensure the progress of the detection mission. The maximum acceleration during soft landing is selected as an index to access the overload characteristic. The smaller the value is, the better.

  3. 3)

    Buffer performance. Considering the uncertainty of the environment of the target landing regions, the buffer performance should meet the demand of the worst condition. The maximum buffer stroke during soft landing is selected as one of the indexes. And a smaller value means that the volume and weight of the landing gear can be reduced correspondingly, which is beneficial to soft landing.

  4. 4)

    Landing stability. A vertical plane passing through the center of two adjacent footpads, which is parallel to the gravity vector, is defined as an “stability wall” [26]. Since the lander has four legs, there are four such walls. If the centroid of the lander exceeded the enclosure formed by the four stability walls, the landing was considered to be unstable. Here stability distance T is introduced as a parameter measuring the minimum distance between the centroid of the lander and four stability walls during each soft landing. If T remained positive, the landing was declared to be stable.

However, stability is not the only requirement for a successful soft landing. Here, the other indexes except the landing stability were grouped as performance indexes to access the soft landing systematically. In summary, the four selected indicators are listed in Table 3.

Table 3 Indicators for soft landing

3.2 Analysis of Landing Performance

Based on theory of probability and statistics, experimental design is a scientific and reasonable arrangement of experiments. It extracts a number of sample points within the design interval to better reflect the characteristics of the whole space. To analyze the soft landing performance of adaptive lander under various conditions, optimal Latin hypercube method was adopted. As a result, 36 conditions were screened out as the inputs of experimental design. Taking returned terrain data and current hovering control ability as references, the selected soft landing conditioning factors and its ranges are listed in Table 4 [15], where the horizontal velocity vz and the control errors of rolling angle ψ and pitching angle θ are ignored.

Table 4 Soft landing conditioning factors

The dynamic simulations of the 36 soft landing conditions were carried out, and corresponding soft landing performance data were obtained, which will be mentioned later. In order to display the soft landing process more intuitively, typical 2-2 condition (Table 5) was selected to be analyzed in detail [27]. Based on the simulation results of 2-2 condition, the relationships between damping coefficient of primary buffer, landing stability index and three performance indexes against time increment are shown respectively in Figures 6, 7, 8, 9 and 10.

Table 5 Parameters of 2-2 condition
Figure 6
figure 6

Damping coefficient of adaptive buffer

Figure 7
figure 7

Buffer stroke of primary strut

Figure 8
figure 8

Landing stability index

Figure 9
figure 9

Nozzle’s performance against damage

According to the above figures, during the whole soft landing process, the damping coefficients of the four adaptive buffers are dynamically adjusted as the attitude of the lander changes, which leads to a stable soft landing. However, buffer parameters selected above based on only an ideal working condition where four legs hit the ground at the same time. While for deep space exploration mission, there is great uncertainty about the terrain and condition of interesting regions. So it is necessary to optimize the buffer parameters based on the uncertainty of landing conditions, so as to enhance the landing adaptability facing different complex conditions.

Figure 10
figure 10

Acceleration overload

4 Hierarchical Optimization

In this section, hierarchical optimization was adopted to improve the soft landing performance of the lander. The optimization took buffer parameters k, cmax and r as design variables, while the performance indexes U, L and S as three primary objectives. Since the change of the buffer parameters is followed with the change of the worst condition, the worst condition and corresponding performance indexes should be updated duly during the optimization process. Aiming at this complex optimization problem, a hierarchical optimization method was proposed decomposing the problem into a leader and several followers. After receiving the buffer parameters from the leader and modifying the model correspondingly, the followers took the conditioning factors as design variables to find the worst conditions respectively. Then the results of this “reverse optimizations” were delivered to the leader. The leader then optimized the buffer parameters trying to better the worst performance indexes. And the new buffer parameters determined by leader are transmitted to the followers to start the next iteration. The cycle continues until the terminating conditions are satisfied. Finally, the Pareto optimal set of the buffer parameters is obtained after hierarchical optimization.

But because this optimization method needs much more iterations, time cost will be much higher if the dynamic model is used to computed. Therefore, this paper uses the response surface model instead of time consuming dynamic analysis model to iterate, shortening the actual solution time and improve the optimization efficiency. In addition, the influence of the conditioning factors on the performance of soft landing is complex, so sensitivity analysis was carried out to determine the influence degree of each conditioning factor on the soft landing performance before establishing response surfaces. To sum up, the flowchart of hierarchical optimization is shown in Figure 11.

Figure 11
figure 11

Flowchart of hierarchical optimization

4.1 Sensitivity Analysis of Conditioning Factors

The 36 results obtained from experimental design analyzing the landing performance in Section 3.2 is used as the sample points for sensitivity analysis. The Pareto diagram, which displays the sensitivity of 4 conditioning factors to landing indicators is shown in Figures 12, 13, 14 and 15. Here, the solid bars indicate positive effects while hollow bars mean negative effects. As the figures show, there are both positive and negative effects to 4 landing indicators, which means that four indicators are conflict with each other and one performance improvement often leads to the other performances lowered. In addition, the sensitivity analysis shows that the initial vertical velocity has less influence on the stability distance. So vx is neglected while establishing the response surface model of stability distance.

Figure 12
figure 12

Sensitivity of conditioning factors to U

Figure 13
figure 13

Sensitivity of conditioning factors to L

Figure 14
figure 14

Sensitivity of conditioning factors to S

Figure 15
figure 15

Sensitivity of conditioning factors to T

4.2 Response Surface Model

According to the results of sensitivity analysis, there was a complex coupling relation between the conditioning factors influencing the landing indicators. Thus, the incomplete three order polynomial function was selected to establish the response surface models, whose basic structure is shown as Eq. (3):

$$ \begin{aligned} f\left( x \right) = \,& \beta_{0} + \sum\limits_{i = 1}^{n} {\beta_{i} x_{i} \,+ } \sum\limits_{{ij\left( {i < j} \right)}}^{n} {\beta_{ij} x_{i} x_{j} } \\ +& \quad \sum\limits_{i = 1}^{n} {\beta_{ij} x_{i}^{2} } + \sum\limits_{i = 1}^{n} {\beta_{iii} x_{i}^{3} } , \\ \end{aligned} $$

where xi means input variables, n is the number of input variables and β is polynomial coefficients.

According to the process of the hierarchical optimization method, k, cmax, r and vx, φ, μ, αe were chosen as input variables. 44 sample points were obtained by adopting the optimized Latin hypercube experimental design, which are listed as Table 11 in Appendix. The response surface model was established by least square fitting using the sample points, shown as Eqs. (7)‒(10) in Appendix. The fitting precision of the response surface model was checked by two test methods, namely, root mean square error RMSE and the coefficient of multiple determination R2. Their expressions are shown in Eqs. (4), (5) respectively:

$$ {\text{RMSE}} = \frac{1}{{m\bar{y}}}\sqrt {\sum\limits_{i = 1}^{m} {\left( {y_{i} - \hat{y}_{i} } \right)^{2} } } , $$
$$ R^{2} = 1 - \frac{{\sum\limits_{i = 1}^{m} {\left( {y_{i} - \hat{y}_{i} } \right)^{2} } }}{{\sum\limits_{i = 1}^{m} {\left( {y_{i} - \bar{y}_{i} } \right)^{2} } }}, $$

where m is the number of sample points, y is the output value of sample point, \( \hat{y} \) is the corresponding output value evaluated by responsible surface model, \( \bar{y} \) is the mean of sample points. A little RMSE and a large R2 mean a better model with high fitting precise.

Figure 16 shows the fitting degree by setting the landing indicators obtained from sample points as X-axis and the corresponding ones from response surface model under the same inputs as Y-axis. The closer the scatter points are to the middle diagonal line, the higher the fitting accuracy is.

Figure 16
figure 16

Scatter diagram for fitting precision of response surface

In the figure, subscript A indicates that the values of landing indicators were obtained by simulation and subscript P means that they were from response surface model.

It can be seen from Figure 16 that the R2 of all indicators are higher than 0.97, and RMSE less than 0.05. The fitting accuracy is enough for the response surface models to replace the dynamic one to be computed.

4.3 Optimization in the Followers

After being determined by the leader and transmitted to the followers, the buffer parameters remained unchanged in a round of sub optimization until the next iteration receiving the new parameters from the leader. The followers took landing conditioning factors as design variables, and the four worst landing indicators as objectives. The key parameters of the optimization mathematical models are listed in Table 6. Where qi represent the buffer parameters vx, φ, μ and αe, while q L i , q U i are the lower and upper limits of the corresponding parameters respectively.

Table 6 Mathematical models of the follower optimizations

The evolution algorithm was adopted for the optimization calculation of the followers, and the algorithm parameters are listed in Table 7. Figure 17 shows iterative processes of four follower optimizations under initial buffer parameters (Table 2). Results show that optimization processes of four followers converged well, and the worst conditions could be found within the maximum iteration step.

Table 7 Configuration parameters of evolution algorithm
Figure 17
figure 17

Iteration processes of four follower optimizations

4.4 Optimization in the Leader

Considering the uncertainty of landing conditions, in the premise of ensuring the landing stability, the multi-objective optimization was carried out to minimize Lmax, minimize Smax and maximize Umin; Taking k, cmax and r as design variables. Therefore, the mathematical model of the leader optimization is shown as Eq. (6). Where xi represent the buffer parameters cmax, r and K, while x Li , x Ui are the lower and upper limits of the corresponding parameters respectively. The second generation non inferiority sorting genetic algorithm (NSGA-II) is adopted for the leader optimization, and the parameters of algorithm are listed in Table 8.

$$ \left. {\begin{array}{*{20}l} {\text{min} } \hfill & {L_{\text{max} } ,S_{\text{max} } , - U_{\text{min} } } \hfill \\ {{\text{s}} . {\text{t}}.,} \hfill & {T_{\text{min} } > 1400,} \hfill \\ {} \hfill & {L_{\text{max} } < 15,} \hfill \\ {} \hfill & {x_{i}^{\text{L}} < x_{i} < x_{i}^{\text{U}} .} \hfill \\ \end{array} } \right\} $$
Table 8 Configuration parameters of NSGA-II

After the hierarchical optimization, the Pareto optimal set of the multi-objective optimization problem is obtained (Table 12 of Appendix). Figure 18 shows the Pareto front fitted by the three performance indexes Lmax, Smax and Umin. Considering the acceleration tolerance of precision equipment in the lander and human is sensitive to the change of acceleration, a small overload helps to improve the reliability of the lander. The optimum buffer parameters selected comprehensively from Pareto optimal set is shown in Table 9.

Figure 18
figure 18

Pareto front of three performance indexes

Table 9 The selected buffer parameters

According to the selected optimum buffer parameters (Table 9), the dynamic analysis model was modified correspondingly. Then dynamic simulation was carried out again while using the 36 landing conditions the same as the experimental design in Section 3.2. Table 10 compares the calculation results before and after optimization. The comparative analysis shows that the three landing performances have been improved to a certain extent under the premise of ensuring the stability of the lander not reduced. Under the respective worst conditions, nozzle performance against damage is increased by 5.24%, the acceleration overload is decreased by 5.74%, and the buffer performance of the primary strut is increased by 21.10%. As for the average, nozzle performance against damage is increased by 3.57%, the acceleration overload is decreased by 2.95%, and the buffer performance of the primary strut is increased by 25.38%.

Table 10 Results of the hierarchical optimization

5 Conclusions

  1. 1)

    The dynamic analysis model of the lander which is equipped with adaptive buffer was established. And the semi-active control algorithm is applied to realize the adaptive feedback adjustment of the damping coefficient during soft landing. Based on the dynamic analysis model and experimental design method, the soft landing performances under multiple landing conditions were analyzed.

  2. 2)

    Focusing on the worst landing condition uncertainty caused by the changes of buffer parameters, a hierarchical optimization method was proposed. The method divides the optimization problem into two parts, namely, a multi-objective leader optimization and several follower optimizations. Through this method, the worst condition was updated duly during optimization process. Furthermore, in order to improve the computational efficiency, response surface models were established to replace the dynamic models for iterative calculation.

  3. 3)

    The soft landing performances before and after the optimization were compared. In the premise of ensuring the landing stability not decline, nozzle’s performance against damage, the buffer performance of the primary strut and the acceleration overload performance are all improved after optimizing. And the comparison results indicate the effectiveness of the hierarchical optimization method.

  4. 4)

    Those studies provide guidance to the design of adaptive lander, including the scheme determination, performance analysis and optimal design. And the hierarchical optimization method, which was proposed to solve complex optimization problems, provides a feasible scheme for optimal design of the project with similar properties.

  5. 5)

    The feasibility of the lander with adaptive buffer is validated preliminarily. However, the performances of the buffer such as vibration and response speed may also influence the soft landing performances in some way, which is simplified in this paper. In future studies, we intend to research on those properties respectively, thus validating its practicability further.


  1. Schröder S, Reinhardt B, Brauner C, et al. Development of a Marslander with crushable shock absorber by virtual and experimental testing. Acta Astronautica, 2017, 134: 65–74.

    Article  Google Scholar 

  2. Y Zhang, H Nie, J B Chen. Design and analysis of MR bumper about lunar damper soft landing. Aerospace Shanghai, 2009, 26(1): 48–52. (in Chinese)

    Google Scholar 

  3. G H Lucas, H D Robert, J M Veloria. Modeling and validation of a navy A6-intruder actively controlled landing gear system. NASA Langley Technical Report Server, 1999.

  4. J Holnickiszulc, P Pawłowski, M Mikułowski, et al. Adaptive impact absorption and applications to landing devices. Advances in Science & Technology, 2008, 56: 609–613.

    Article  Google Scholar 

  5. M Ostrowski, J Holnicki-Szulc. Adaptive impact absorption controlled via pyrotechnic devices. 4th European Conf. on Structural Control, Petersburg, Russia, 2008.

  6. J H Hu. Research on magnetorheological fluid damping semi-active control system. Zhejiang University, 2007. (in Chinese)

  7. Q Zhou. Design and analysis of magneto rheological dynamic vibration absorber. Huazhong University of Science and Technology, 2013. (in Chinese)

  8. L Sa. Design of magnetorheological buffer for lunar lander soft landing gear. Beijing Jiaotong University, 2016. (in Chinese)

  9. B Y Niu. Structure design and optimization of magnetorheological buffer for lunar lander landing system. Chongqing University, 2016. (in Chinese)

  10. G M Mikulowski, J Holnickiszulc. Adaptive landing gear concept—feedback control validation. Smart Materials & Structures, 2007, 16(6): 2146.

    Article  Google Scholar 

  11. A L Wang. Research on dynamics and semi-active control of lunar lander soft landing. Nanjing University of Aeronautics & Astronautics, 2011. (in Chinese)

  12. G Mikulowski, L Jankowski. Adaptive landing gear: Optimum control strategy and potential for improvement. Shock & Vibration, 2015, 16(2): 175–194.

    Article  Google Scholar 

  13. T Maeda, M Otsuki, T Hashimoto, et al. Attitude stabilization for lunar and planetary lander with variable damper. Journal of Guidance, Control, and Dynamics, 2016: 1790–1804.

  14. J J Wang, C J Wang, S G Song. Performance optimization of lunar lander based on response surface methodology. Journal of Beijing University of Aeronautics and Astronautics, 2014, 40(5): 707–711. (in Chinese)

    Google Scholar 

  15. H Y Wu, C J Wang, J Z Ding, et al. Soft landing performance optimization for novel lander based on multiple working conditions. Journal of Beijing University of Aeronautics and Astronautics, 2017(4): 776–781. (in Chinese)

    Google Scholar 

  16. Q X Qin, X Qin, J R Xiao, et al. Robust optimization design of the mars airbag lander. China Mechanical Engineering, 2017, 28(1): 20–26. (in Chinese)

    Google Scholar 

  17. J Z Yang, F M Zeng, J F Man, et al. Design and verification of the landing impact attenuation system for Chang’E-3 lander. Scientia Sinica (Technologica), 2014(5): 440–449. (in Chinese)

    Google Scholar 

  18. H Y Chai, Y H Deng, C Sheng. Design and realization of structure subsystem for the Chang’E-3 lunar lander. Scientia Sinica (Technologica), 2014(4): 391–397. (in Chinese)

    Google Scholar 

  19. X X Bai, N M Wereley, W Hu. Maximizing semi-active vibration isolation utilizing a magnetorheological damper with an inner bypass configuration. Journal of Applied Physics, 2015, 117(17): 288.

    Article  Google Scholar 

  20. X X Bai, W Hu, N M Wereley. Magnetorheological damper utilizing an inner bypass for ground vehicle suspensions. IEEE Transactions on Magnetics, 2013, 49(7): 3422–3425.

    Article  Google Scholar 

  21. A L Wang, H Nie, J B Chen. State-jump semi-active control of lunar lander soft landing. Acta Aeronautica Et Astronautica Sinica, 2009, 30(11): 2218–2223. (in Chinese)

    Google Scholar 

  22. Z W Li, Z J Li. Status of researching on dynamical models of MR damper. Machine Building & Automation, 2012, 41(1): 142–145. (in Chinese)

    Google Scholar 

  23. Y H Guo, E W Chen, Y M Lu, et al. Calculation of equivalent linear damping coefficient of a magnetorheological damper. China Mechanical Engineering, 2014, 25(13): 1719–1723. (in Chinese)

    Google Scholar 

  24. Z Q Deng, S C Wang, H B Gao, et al. Upper limit amplitude of lunar lander based on ADAMS software. Journal of Harbin Institute of Technology, 2003, 35(12): 1492–1495. (in Chinese)

    Google Scholar 

  25. “LORD Technical Data RD-8040-1 and RD-8041-1 Dampers,” Accessed on November 2015,

  26. R J Muraca, J W Campbell, C A King. A Monte Carlo analysis of the viking lander dynamics at touchdown. National Aeronautics and Space Administration, 1975.

    Google Scholar 

  27. R E Lavender. Monte Carlo approach to touchdown dynamics for soft lunar landings. NASA-TN-D-3117, 1965.

Download references

Authors’ Contributions

CW was in charge of the whole trial; ZD wrote the manuscript; HW and JD assisted with sampling and laboratory analyses. All authors read and approved the final manuscript.

Authors’ Information

Zongmao Ding, born in 1993, is currently a master candidate at School of Mechanical Engineering and Automation, Beijing University of Aeronautics and Astronautics, China. He received his bachelor degree from Beijing University of Aeronautics and Astronautics, China, in 2016.

Hongyu Wu, born in 1993, is currently a PhD candidate at Department of Mechanical Engineering, Tsinghua university, China. He received his master degree from Beijing University of Aeronautics and Astronautics, China, in 2018.

Chunjie Wang, born in 1955, is currently a professor at State Key Laboratory of Virtual Reality and Systems, Beijing University of Aeronautics and Astronautics, China. She received her PhD degree from China University of Mining and Technology, China, in 1997.

Jianzhong Ding, born in 1991, is currently a PhD candidate at School of Mechanical Engineering and Automation, Beijing University of Aeronautics and Astronautics, China. He received his bachelor degree from China University Of Petroleum, China, in 2013.

Competing Interests

The authors declare that they have no competing interests.


Supported by National Natural Science Foundation of China (Grant No. 51635002).

Author information

Authors and Affiliations


Corresponding author

Correspondence to Zongmao Ding.



See Tables 11 and 12.

$$ \begin{aligned} U = \,& 511.59 - 1.1048\varphi - 199.74\mu + 205.45r \\ & - 6.9518\alpha_{e} - 24.145v_{x} + 0.015802\varphi^{2} + 68.985\mu^{2} \\ & - 253.31r^{2} + 0.43316\alpha_{e}^{2} + 0.76387\varphi \mu + 2.3551{\text{E}} \\ & - 6\varphi k + 7.8933{\text{E}} - 4\mu c_{\text{max} } + 2.3759{\text{E}} - 4\mu k \\ & + 19.753\mu v_{x} - 2.2695{\text{E}} - 6c_{\text{min} } + 3.9774\alpha_{e} c_{\text{max} } \\ & + 3.8669r\alpha_{e} - 3.2941{\text{E}} - 4rk + 2.1381\alpha_{e} k \\ & - 0.76569\alpha_{e} v_{x} + 121.85r^{3} , \\ \end{aligned} $$
$$ \begin{aligned} L = \,& 26.775 - 5.0093r - 0.48742\alpha_{e} + 14.621\mu^{2} \\ &+ 0.11214\alpha_{e}^{2} - 6.7505{\text{E}} - 9k^{2} - 3.4301v_{x}^{2} \\ &+ 0.071897\varphi \mu - 0.13375\varphi r - 8.0912{\text{E}} \\ &- 5\mu c_{\text{max} } - 1.4029\mu \alpha_{e} - 2.9661\mu v_{x} \\ &- 2.8717{\text{E}} - 5c_{\text{max} } r - 1.8059{\text{E}} - 6\alpha_{e} c_{\text{max} } \\& + 2.5342rv_{x} - 0.10076\alpha_{e} v_{x} + 5.0199{\text{E}} - 5kv_{x} \\ &+ 3.1638{\text{E}} - 5\varphi^{3} + 1.1452{\text{E}} - 14c_{\text{max} }^{3} \\& + 1.1324r^{3} + 6.5243{\text{E}} - 14k^{3} + 0.67529v_{x}^{3} , \\ \end{aligned} $$
$$ \begin{aligned} S = \,& 482.67 - 0.0052582c_{\text{max} } - 615.95r + 193.76\mu^{2} \\ &+ 4.2096{\text{E}} - 8c_{\text{max} }^{2} + 777.24r^{2} - 33.976v_{x}^{2} \\ &- 1.7764\varphi \mu + 1.0444{\text{E}} - 5\varphi c_{\text{max} } \\ &+ 0.41149\varphi r + 9.5027\mu \alpha_{e} - 44.733\mu v_{x} \\& + 4.9696{\text{E}} - 4c_{\text{max} } r - 6.6522{\text{E}} - 5c_{\text{max} } \alpha_{e} \\ &- 5.0897r\alpha_{e} + 5.7483{\text{E}} - 4rk + 1.6141\alpha_{e} v_{x} \\& + 3.8669r\alpha_{e} - 5.0606{\text{E}} - 4kv_{x} - 7.8106{\text{E}} \\ &- 5\varphi^{3} - 357.00r^{3} + 1.5838{\text{E}} - 13k^{3} + 7.9416v_{x}^{3} , \\ \end{aligned} $$
$$ \begin{aligned} T = \,& 15524 - 0.35713\varphi + 1102.9\mu + 239.20r \\& - 11976v_{x} + 8.9406{\text{E}} - 9c_{\text{max} }^{2} - 593.63r^{2} \\ &- 3.3219\alpha_{e}^{3} + 3437.5v_{x}^{2} - 0.22014\varphi \alpha_{e} \\ &- 0.0014584\mu c_{\text{max} } - 11.325\mu \alpha_{e} - 0.0027737\mu k \\ &- 218.04\mu v_{x} - 6.3259r\alpha_{e} + 29.639rv_{x} - 1.9831{\text{E}} \\ &- 4\alpha_{e} k - 213.08\mu^{3} + 315.93r^{3} + 0.15724\alpha_{e}^{3} \\ &+ 2.9960{\text{E}} - 13k^{3} - 325.27v_{x}^{3} . \\ \end{aligned} $$
Table 11 Sample points for response surface
Table 12 Pareto optimal set

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ding, Z., Wu, H., Wang, C. et al. Hierarchical Optimization of Landing Performance for Lander with Adaptive Landing Gear. Chin. J. Mech. Eng. 32, 20 (2019).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


  • Landing gear
  • Soft landing
  • Sensitivity analysis
  • Response surfaces
  • Hierarchical optimization