The purpose of this study is to address the progressively increasing accuracy requirements for force measurement in the manufacturing and operation of modern mechanical equipment as well as the development of high-end equipment. Focusing on the four-fulcrum dynamometer – a novel structure increasingly applied in critical force measurement scenarios – this research emphasizes that the configuration of its four force measurement units serves as a crucial factor affecting the dynamometer's force measurement performance.
A mechanical model was established based on the structure of the four-fulcrum dynamometer, and output performance calibration experiments were conducted. The sparrow search algorithm-backpropagation (SSA-BP) neural network was constructed using the existing experimental data and verified. The optimization objective was to minimize the output fluctuation of multipoint loading within the fixed loading area of the dynamometer. Relying on the favorable nonlinear optimization characteristics of the BP neural network, the SSA was adopted to optimize the weights and thresholds of the BP neural network, addressing the issue of the BP neural network getting trapped in local optima. The output model of the dynamometer under different spans and different point loadings was established, and enabling predictions across multiple spans.
The prediction results indicate that the output performance within the fixed loading area of the dynamometer gradually diminishes with the increase in the sensor span, and the output fluctuation reduces from a maximum of 241.9 N (4.8% FS) to 57.99 N (1.15% FS), signifying a remarkable enhancement in performance. The maximum error is 28 N, accounting for 0.57% of the full scale, which confirms the high accuracy and feasibility of the network. Consequently, on the condition of satisfying the structural stability of the dynamometer, the arrangement span of the sensor should be augmented. This approach shortens the experimental design process and saves a considerable amount of variable-span experimental procedures. The verification results of the network output confirm that the network accuracy satisfies the usage requirements, with the experimental results exhibiting high credibility.
This study provides a methodological foundation for the design of four-fulcrum dynamometers and the enhancement of force measurement accuracy.
