Abstract: As the main transportation equipment for large open-pit mines, the development of mine truck overload warning function is one of the effective technical means to ensure mine safety production. In response to the current situation that mining truck loading and transportation rely on personnel's work experience, as well as the common occurrence of overloading and overloading of mining trucks, a mining truck indirect weighing method based on front oil and gas suspension nitrogen pressure measurement and middle and rear steel plate spring suspension displacement measurement is proposed, with a focus on vehicle mounted weighing data acquisition and filtering technology, as well as intelligent weighing algorithm research. Firstly, an indirect weighing method was selected for the three bridge rigid mining truck, and a mining truck mounted weighing system was designed. The corresponding software and hardware solutions were studied; Secondly, on-site collection experiments were conducted on the weighing signals of mining trucks, and 9 types of weighing signals were measured by simulating the actual transportation conditions of mining trucks; Then, in response to the shortcomings of traditional Kalman algorithm in filtering out high amplitude impulse noise in vehicle signals, algorithm optimization was carried out by adding outlier judgment and Kalman gain adjustment formula, and the data foundation of vehicle weighing training model was obtained; Finally, in response to the shortcomings of slow learning rate and easy falling into local minima in the application of BP network in weighing systems, the Tuna swarm optimization (TSO) algorithm was introduced to optimize it. A mining truck intelligent vehicle weighing prediction model based on TSO−BP neural network was constructed, and the original weighing signal dataset and the filtered denoised weighing signal dataset were made. Comparative experiments of weighing algorithms were conducted based on vehicle weighing signals. The results show that for the original weighing signal dataset, the maximum weighing error of the TSO−BP prediction model is reduced from 25.57% to 9.06% compared to the traditional BP prediction model. The TSO−BP vehicle weighing model has better robustness and stability; For the denoised weighing signal dataset, the overall error of the TSO−BP mining vehicle onboard weighing model is less than 1%, the weighing error during loading and full load stationary is about 0.1%, and the weighing error during full load operation is less than 0.5%, which can meet the accuracy requirements of dynamic onboard weighing for mining vehicles.