2.Parameters to be tuned
-
In the calibration of agent-based microscopic model, OD matrix is one of major parameters to be adjusted. Previous implementations of the W-SPSA algorithm used each entry of the dynamic Origin-Destination (OD) matrix as a parameter to be calibrated. Each of the cells of the matrix of N×N for each p period would be bounded by 0 and (eventually) a manually specified upper bound. As SimMobility does not directly use OD as demand input but activity schedules and trip chains, we aggregate the trip chain to generate OD parameter which to be calibrated, then, convert the updated OD parameter (θ ̂_(k+1)) into trip chains by disaggregating through so-called ‘killing-cloning’ process as explained in Section 1.
-
Each activity schedule will be the “parameter” to be calibrated. This means that each individual activity schedule, and therefore each agent, will be calibrated. For this first implementation and as the generation of the activity schedules is modelled at the MT, the parameters that control trips are not considered directly for calibration. Each agent activity schedule (computed by MT) is considered as fixed, but replicates can be generated in the calibration process (clones).
-
One additional temporary table/file must be created in the database/working folder, with the following fields:
<agentID>, <startTime>,…,<OriginID>,<DestinationID>,<loadFactor> -
The above table/file shall be accessible by Matlab at each iteration of the calibration process. The agentID is mapped to its activity schedule (and therefore trip chains) in the original database files. The loadFactor is the value to be calibrated by the algorithm. Note that the loadFactor should be an integer. The Matlab script should then generate the temporary demand that will be used to run SimMobility at each iteration.
- Another parameter in demand side is route choice parameters. The route choice model for private transport corresponds to a Path-Size Logit Model. The detailed description of the attributes of the model can be referred to SimMobility MT manual. In the Short-Term calibration, we include following parameters with the initial setting, perturbation step, and boundary condition as follows:
- For the supply parameters, 112 selected parameters were selected for calibration. Reader can refer
configure_Whole_set.mfor the list of supply parameters. - Initial value of driving behavior parameter:
initial_db = [2.5 ... % min_response_distance [0]
0.045 0.740 0.205 0.750 0.494 ... % ... % CF_parameters_1 [1-5]
-0.40 0.000 0.650 0.850 0.605 ... % CF_parameters_2 [6-10]
0.3091 ... % FF_Acc_Params_b2 [11]
1.3 ... % speed_factor [12]
0.604 0.385 0.323 0.0678 0.217 0.583 -0.596 -0.219 0.0832 0.170 1.478 0.131 0.300 ... % target_gap_acc_parm [13-25]
6.0 5.50 5.0 4.50 4.00 ... % max_acc_car1 [26-30]
-2.5 -2.25 -2.00 -1.75 -1.50 ... % normal_deceleration_car1 [31-35]
-9.00 -8.50 -8.00 -7.50 -7.00 ... % max_deceleration_car1 [36-40]
1.0 ... % hbuffer_lower [41]
1.850875 2.109445 2.263295 2.386205 2.496535 2.60338 2.71371 2.83662 2.99047 3.24904 ... % hbuffer_upper [42-51]
1.0 ... % yellow_stop_headway [52]
2.2352 ... % min_speed_yellow [53]
150.0 ... % driver_signal_perception_distance [54]
0.4 ... % dec_update_step_size or stopped_vehicle_update_step_size [55]
0.5 ... % acc_update_step_size or uniform_speed_update_step_size [56]
750.0 1000.0 2.0 ... % MLC_PARAMETERS [57-59]
4.2656 0.3213 -1.1683 0.0 0.08 -1.0 0.009 -0.2664 -0.012 -3.3754 ... % lane_utility_model [60-69]
10 19 -2.3400 -4.5084 -2.8257 -1.2597 -0.7239 -0.3269 ... % lane_utility_model [70-77]
0.2 -0.231 -2.700 1.112 0.5 0.000 0.2 0.742 0.6 ... % critical_gaps_param [78 - 86]
-0.837 0.913 0.816 -1.218 -2.393 -1.662 ... % Target_Gap_Model [87-92]
1.0 0.5 0.6 0.1 0.5 1.0 30.0 100.0 600.0 40.0 ... % nosing_param [93-102]
0.5 ... % CF_CRITICAL_TIMER_RATIO [103]
0.3 ... % intersection_attentiveness_factor_min [104]
0.6 ... % intersection_attentiveness_factor_max [105]
1.25 ... % minimum_gap [106]
1.0 0.5 ... % critical gap addon [107-108]
0.2 ... % minimum_gap [109]
7 ... % LC Discretionary_Lane_Change_Model_Same [110]
15]; % LC Discretionary_Lane_Change_Model_Diff [111]
- Upper/Lower boundary of driving behavior parameter:
boundary = 0.1;
up_db = [6.0 ... % min_response_distance [0]
0.045+0.045*boundary 0.9 0.205+0.205*boundary 0.8 0.65 ... % CF_parameters_1 [1-5]: alpha beta gama lambda rho : 0.045 0.9 -0.125 0.8 0.65
-0.02 0.000 0.650+0.650*boundary 1.00 0.95 ... % CF_parameters_2 [6-10] -0.02 0.000 -0.05 1.00 0.95
0.45 ... % FF_Acc_Params_b2 [11]
1.3+1.3*boundary ... % speed_factor [12]
0.75 0.6 0.6 0.10 0.35 0.75 -0.40 -0.2 0.10 0.170+0.170*boundary 2.0 0.20 0.45 ... % target_gap_acc_parm [13-25]
6.0+6.0*boundary 5.50+5.50*boundary 5.0+5.0*boundary 4.50+4.50*boundary 4.00+4.00*boundary ... % max_acc_car1 [26-30]
-2.5+2.5*boundary -2.25+2.25*boundary -2.00+2.00*boundary -1.75+1.75*boundary -1.50+1.50*boundary ... % normal_deceleration_car1 [31-35]
-9.00+9.00*boundary -8.50+8.50*boundary -8.00+8.00*boundary -7.50+7.50*boundary -7.00+7.00*boundary ... % max_deceleration_car1 [36-40]
1.0+1.0*boundary ... % hbuffer_lower [41]
1.850875+1.850875*boundary 2.109445+2.109445*boundary 2.263295+2.263295*boundary 2.386205+2.386205*boundary 2.496535+2.496535*boundary 2.60338+2.60338*boundary 2.71371+2.71371*boundary 2.83662+2.83662*boundary 2.99047+2.99047*boundary 3.24904+3.24904*boundary ... % hbuffer_upper [42-51]
1.0+1.0*boundary ... % yellow_stop_headway [52]
2.2352+2.2352*boundary ... % min_speed_yellow [53]
150.0+150.0*boundary ... % driver_signal_perception_distance [54]
0.4+0.4*boundary ... % dec_update_step_size or stopped_vehicle_update_step_size [55]
0.5+0.5*boundary ... % acc_update_step_size or uniform_speed_update_step_size [56]
1000 1200.0 2.0 ... % MLC_PARAMETERS [57-59]
6.0 0.3213+0.3213*boundary -0.75 0.0 0.1 -0.5 0.01 -0.15 -0.002 -1.75 15.0 23.0 -1.0 -2.5 -1.25 -0.75 -0.5 -0.15 ... % lane_utility_model [60-69] % lane_utility_model [70-77]
2.0 -0.1 -1.5 1.75 2.5 0.00 0.7 1.0 6.0 ... % critical_gaps_param [78 - 86]
-0.837+0.837*boundary 0.913+0.913*boundary 0.816+0.816*boundary -1.218+1.218*boundary -2.393+2.393*boundary -1.662+1.662*boundary ... % Target_Gap_Model [87-92]
1.0+1.0*boundary 0.5+0.5*boundary 0.6+0.6*boundary 0.1+0.1*boundary 0.5+0.5*boundary 1.0+1.0*boundary 30.0+30.0*boundary 100.0+100.0*boundary 600.0+600.0*boundary 40.0+40.0*boundary ... % nosing_param [93-102]
0.5+0.5*boundary ... % CF_CRITICAL_TIMER_RATIO [103]
0.3+0.3*boundary ... % intersection_attentiveness_factor_min [104]
0.6+0.6*boundary ... % intersection_attentiveness_factor_max [105]
1.25+1.25*boundary ... % minimum_gap [106]
1.0+1.0*boundary 0.5+0.5*boundary ... % critical gap addon [107-108]
0.2+0.2*boundary ... % minimum_gap [109]
7+7*boundary ... % LC Discretionary_Lane_Change_Model_Same [110]
15+15*boundary]; % LC Discretionary_Lane_Change_Model_Diff [111]
lb_db = [2.0 ... % min_response_distance [0]
0.015 0.1 0.205-0.205*0.1 0.4 0.45 ... % CF_parameters_1 [1-5]: 0.015 0.1 -0.300 0.4 0.45
-0.95 0.000 -0.75 0.05 0.5 ... % CF_parameters_2 [6-10]
0.05 ... % FF_Acc_Params_b2 [11]
1.3-1.3*0.1 ... % speed_factor [12]
0.50 0.2 0.1 0.03 0.10 0.25 -0.75 -0.5 0.05 0.170-0.170*0.1 0.8 0.05 0.15... % target_gap_acc_parm [13-25]
6.0-6.0*boundary 5.50-5.50*boundary 5.0-5.0*boundary 4.50-4.50*boundary 4.00-4.00*boundary ... % max_acc_car1 [26-30]
-2.5-2.5*boundary -2.25-2.25*boundary -2.00-2.00*boundary -1.75-1.75*boundary -1.50-1.50*boundary ... % normal_deceleration_car1 [31-35]
-9.00-9.00*boundary -8.50-8.50*boundary -8.00-8.00*boundary -7.50-7.50*boundary -7.00-7.00*boundary ... % max_deceleration_car1 [36-40]
1.0-1.0*boundary ... % hbuffer_lower [41]
1.850875-1.850875*boundary 2.109445-2.109445*boundary 2.263295-2.263295*boundary 2.386205-2.386205*boundary 2.496535-2.496535*boundary 2.60338-2.60338*boundary 2.71371-2.71371*boundary 2.83662-2.83662*boundary 2.99047-2.99047*boundary 3.24904-3.24904*boundary ... % hbuffer_upper [42-51]
1.0-1.0*boundary ... % yellow_stop_headway [52]
2.2352-2.2352*boundary ... % min_speed_yellow [53]
150.0-150.0*boundary ... % driver_signal_perception_distance [54]
0.4-0.4*boundary ... % dec_update_step_size or stopped_vehicle_update_step_size [55]
0.5-0.5*boundary ... % acc_update_step_size or uniform_speed_update_step_size [56]
75.0 500.0 2.0 ... % MLC_PARAMETERS [57-59]
2.0 0.3213-0.3213*boundary -1.50 0.0 0.03 -1.5 0.002 -0.35 -0.015 -5.50 5.0 15.0 -3.5 -6.5 -3.75 -2.00 -1.0 -0.5 ... % lane_utility_model [60-69] % lane_utility_model [70-77]
0.1 -0.35 -4.0 0.75 0.2 -0.15 0.15 0.5 0.0 ... % critical_gaps_param [78 - 86]
-0.837-0.837*boundary 0.913-0.913*boundary 0.816-0.816*boundary -1.218-1.218*boundary -2.393-2.393*boundary -1.662-1.662*boundary ... % Target_Gap_Model [87-92]
1.0-1.0*boundary 0.5-0.5*boundary 0.6-0.6*boundary 0.1-0.1*boundary 0.5-0.5*boundary 1.0-1.0*boundary 30.0-30.0*boundary 100.0-100.0*boundary 600.0-600.0*boundary 40.0-40.0*boundary ... % nosing_param [93-102]
0.5-0.5*boundary ... % CF_CRITICAL_TIMER_RATIO [103]
0.3-0.3*boundary ... % intersection_attentiveness_factor_min [104]
0.6-0.6*boundary ... % intersection_attentiveness_factor_max [105]
1.25-1.25*boundary ... % minimum_gap [106]
1.0-1.0*boundary 0.5-0.5*boundary ... % critical gap addon [107-108]
0.2-0.2*boundary ... % minimum_gap [109]
7-7*boundary ... % LC Discretionary_Lane_Change_Model_Same [110]
15-15*boundary]; % LC Discretionary_Lane_Change_Model_Diff [111]