Vehicle Parameter Data of Heterogeneous Fleets for Parameter Optimization Method of Distributed Model Predictive Control Under Multiple Topologies

The heterogeneous dynamic parameters of each vehicle, namely mass, powertrain time constant, aerodynamic drag coefficient and wheel radius, are configured diversely within the typical parameter ranges of passenger cars and light commercial vehicles to reflect the inherent differences in dynamic characteristics of mixed fleets. Specifically, vehicle mass ranges from 1097.5 kg to 1913.4 kg, the inertial time constant of the powertrain varies from 0.52 s to 0.77 s, the comprehensive aerodynamic drag coefficient is set between 0.14 and 1.16 N·s²/m², and the wheel radius is selected from 0.25 m to 0.29 m.

Second, general simulation parameters, including discrete time step, prediction horizon, desired inter-vehicle spacing and acceleration boundaries, are determined in accordance with conventional settings in the field of distributed model predictive control. Finally, the weight parameters of the DMPC cost function are divided into two groups: empirical parameters before optimization and optimized parameters obtained via Particle Swarm Optimization (PSO).

The pre-optimization parameters adopt the empirical values from Zheng et al. (2016). The post-optimization parameters are acquired by offline optimization on the MATLAB platform using the PSO algorithm, where sufficient closed-loop stability conditions are embedded as hard constraints in the optimization process, and a multi-objective fitness function is constructed based on spacing tracking accuracy, velocity consistency and control input smoothness.

All data are generated and validated in a computing environment equipped with MATLAB R2020b or later versions (including the Optimization Toolbox).The time-related information of this dataset is as follows: the total simulation duration is 10 seconds with a discrete time step (i.e., time resolution) of 0.1 second, resulting in a total of 100 simulation steps. In terms of spatial information, the vehicles perform one-dimensional longitudinal motion.

The fleet is composed of one leading vehicle and seven following vehicles arranged with a constant spacing strategy, where the desired inter-vehicle distance between adjacent vehicles is 20 m and the initial velocity of all vehicles is 20 m/s. The spatial resolution of the dataset corresponds to the independent longitudinal position and velocity states of each vehicle, without two-dimensional or three-dimensional spatial grids.The file **PlatoonParameter_data.mat** is saved in MATLAB binary format and contains 14 variables in total:1.

Four vehicle dynamic parameter matrices: *Mass* (7×1 matrix, unit: kg), *Tao* (7×1 matrix, unit: s), *Ca* (7×1 matrix, unit: N·s²/m²) and *R* (7×1 matrix, unit: m). Each row corresponds to one following vehicle (Rows 1 to 7 represent Vehicle 1 to Vehicle 7 in sequence), recording vehicle mass, powertrain inertial time constant, comprehensive aerodynamic drag coefficient and wheel radius respectively.2.

Four simulation configuration parameters: *Num_veh* (scalar, value = 7, denoting the number of following vehicles), *Tim_step* (scalar, value = 0.1, unit: s), *Time_sim* (scalar, value = 10, unit: s) and *Num_step* (scalar, value = 100, denoting the total number of simulation steps).3. Three physical constants: *f* (scalar, value = 0.01, dimensionless, road rolling resistance coefficient), *Eta* (scalar, value = 0.96, dimensionless, mechanical efficiency of the transmission system) and *g* (scalar, value = 9.8, unit: m/s², gravitational acceleration).4.

Two acceleration boundary parameters: *AccMax* (scalar, value = 6, unit: m/s²) and *AccMin* (scalar, value = -6, unit: m/s²).5. *Torquebound* (7×2 matrix, unit: N·m). The first and second elements in the *i*-th row represent the lower and upper bounds of the control input torque for the *i*-th following vehicle, which are calculated by the formulas \(u_{\text{min},i} = \dfrac{m_i a_{\text{min}} R_i}{\eta}\) and \(u_{\text{max},i} = \dfrac{m_i a_{\text{max}} R_i}{\eta}\).This file includes records of seven vehicles with no missing data.The file **Parameter_Initial_data.m** is a MATLAB script-based parameter configuration file.

It contains all the aforementioned vehicle dynamic parameters and simulation configuration parameters, along with additional data as follows: driving cycle data of the leading vehicle (the leading vehicle accelerates at \(2\ \text{m/s}^2\) from 1 s to 2 s under normal acceleration conditions, and brakes at \(-6\ \text{m/s}^2\) from 4 s to 5 s under emergency braking conditions); DMPC controller parameters (prediction horizon \(N_p = 20\) steps, desired spacing \(d = 20\ \text{m}\), initial velocity \(v_0 = 20\ \text{m/s}\)); pre-optimization and post-optimization weight parameters (\(F_i, G_i, Q_i, R_i\)) of the DMPC cost function under four communication topologies (PF, PLF, TPF, TPLF); and configuration parameters of the PSO algorithm (population size: 50, maximum iteration number: 300, individual learning factor \(c_1 = 1.5\), social learning factor \(c_2 = 2.0\), 14-dimensional optimization variables, parameter bounds: [0.1, 20]).

All parameters in this file are stored in the form of MATLAB variable assignment statements, with comments indicating the corresponding paper sections and physical meanings.There are no missing values in this dataset, and all variables are fully assigned. Since the dataset consists of simulation parameter configuration files rather than field measurement data, it does not introduce measurement errors itself. Nevertheless, it should be noted that when these parameters are loaded into the DMPC simulation program, the forward Euler discretization method will introduce numerical truncation errors with an order of magnitude of \(O(\Delta t^2)\).

In addition, the convergence tolerance of the nonlinear optimization solver *fmincon* is set to \(1\times10^{-5}\), which leads to minor numerical solution errors in simulation outputs. Such errors are negligible within the accuracy requirements of engineering simulations.The file **PlatoonParameter_data.mat** can be opened with MATLAB R2020b or later versions, or compatible MAT file readers (e.g., the `scipy.io.loadmat` function in Python).

The file **Parameter_Initial_data.m** can be run directly in MATLAB R2020b or later versions. MATLAB software is available for download from the official MathWorks website: mathworks.com/products/matlab.html