Adaptive sliding mode observer with Metaheuristic optimization for accurate state estimation in distributed electric drive plant protection vehicles

• Proposed a CEPSO-ASMO framework for nonlinear agricultural vehicle state estimation. • Reduced tire force estimation errors by 38.32% and 42.10% under two road conditions. • Field tests on DDEPPVs achieved below 5% error in key state and tire force estimates. • Provides a low-cost real-time estimator for safer autonomous agricultural vehicle control. In agricultural plant protection, plant protection vehicles (PPV) face significant stability challenges due to complex road conditions and the unique characteristics of specialized tires and high ground clearance. Accurate measurement of key parameters, such as longitudinal and lateral speeds and vehicle sideslip angle, is essential for advanced stability control . However, the high cost of precision sensors limits their use in agricultural vehicles. To address this challenge, this paper proposes a Chaotic enhanced particle swarm optimization adaptive sliding mode observer (CEPSO-ASMO) for distributed electrically driven propelled plant protection vehicle (DDEPPV). Firstly, the special agricultural tires used in DDEPPV were nonlinearly modeled in conventional hard soil road and soft soil road environments, and the accuracy of the model was verified by theoretical analysis. Secondly, a tire sliding mode observer with an adaptive feedback gain mechanism was designed. Then, a chaos mechanism and dynamic adjustment strategy were introduced to optimize the feedback gain, sliding mode surface parameters, and other aspects of the adaptive sliding mode observer, which improved the accuracy and robustness of the observer. The real vehicle experiments show that, compared with traditional methods, the CEPSO-ASMO significantly improved the tire force and driving state estimation accuracy of the DDEPPV. This method reduced the tire force estimation error by 38.32% and 42.1% under two typical road conditions, respectively, and effectively suppressed the chattering phenomenon. In addition, the normalized root mean square error of the vehicle’s driving state estimation was less than 5%. By using standard high-frequency onboard sensors, the CEPSO-ASMO can accurately estimate tire forces and dynamic driving state parameters of DDEPPV, offering promising engineering application prospects.