Mining Truck Suspension Parameter Identification for Time-Varying Stiffness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mining truck suspension simulation technologies fail to accurately model or quantify the time-varying stiffness and damping characteristics, relying on statistical data rather than precise identification.
Innovation Solution
A method integrating a deep learning network with a physical model to establish a longitudinal-vertical dynamics model, using a three-layer LSTM network and fully connected network to predict suspension parameters, and a loss function to update network parameters based on sensor data, enabling accurate identification of time-varying suspension characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If statistical data is used to model suspension characteristics, then the model is simple to implement, but the accuracy of time-varying stiffness and damping characteristics is insufficient
Solution Approach 1:
The patent introduces a deep learning network as an intermediary between the physical model and sensor data. The neural network processes raw sensor measurements and outputs estimated suspension parameters, acting as a mediator that transforms complex sensor data into meaningful physical parameters while maintaining model interpretability.
Solution Approach 2:
The patent segments the modeling approach into two distinct components: a physical dynamics model that provides the theoretical framework and interpretability, and a deep learning network that handles the complex data processing and parameter estimation. This segmentation allows each component to specialize in its strength while working together to solve the overall problem.
2Measurement precision
If a physical model is used to model suspension characteristics, then the model has interpretability, but it cannot accurately capture time-varying nonlinear characteristics
Solution Approach 1:
The patent creates a composite modeling approach by combining a physical dynamics model with a deep learning network. The physical model provides the theoretical foundation and interpretability, while the neural network adds the capability to capture complex nonlinear time-varying behaviors. The fusion of these two different 'materials' (modeling approaches) creates a hybrid model that exhibits both interpretability and high accuracy.
Solution Approach 2:
The patent makes the model dynamic by using a recurrent neural network that processes sequential sensor data over time. The network adapts to changing operating conditions by continuously learning from new data, allowing the model to capture time-varying characteristics while maintaining the interpretability of the underlying physical model through its structured architecture.
3Measurement precision
If multiple sensors are installed to collect comprehensive data, then the measurement accuracy improves, but the system complexity and cost increase
Solution Approach 1:
The patent extracts and utilizes only the essential information from a single IMU sensor to estimate all four suspension parameters. By carefully selecting and processing the available data from one sensor, the system achieves comprehensive parameter estimation without requiring multiple sensors, thereby reducing system complexity while maintaining measurement accuracy.
Solution Approach 2:
The patent makes the single IMU sensor perform multiple functions: it simultaneously provides data for estimating body acceleration, pitch rate, and ultimately all four suspension parameters (two stiffness and two damping coefficients). This multi-functional use of a single sensor reduces the overall sensor count while maintaining comprehensive measurement capability.
Data Source
AI summary
A method and system are provided for identifying time-varying suspension characteristics of heavy-load vehicles. The method includes collecting sequential control state data of a mining truck using sensors, predicting parameter-related factors through a deep learning network, estimating suspension stiffness and damping coefficients via a linear dynamic model considering longitudinal-vertical coupling, and predicting future system states through a nonlinear dynamic model based on the estimated parameters and learned factors. According to the method, a deep learning network is integrated into a physical model of the mining truck, an accurate longitudinal-vertical dynamical model of the mining truck is established, accurate suspension parameters are identified, the stiffness damping time-varying characteristics of the suspension of the mining truck are given through a physical model-data driving method, and the model has certain interpretability and generalization; the rigidity and damping of the four suspensions can be obtained only through sprung information.