<p>Considering that it is difficult to directly and accurately measure the swing angle of the load of bridge crane during operation, and for the defects of most of today’s crane anti-swing control strategies that require system-wide state information feedback, this study proposes a neural network robust control strategy based on the radial basis function (RBF), which is based on a two-type Lagrangian to establish the crane system dynamic equations, and then derive a control strategy for adaptive updating of neural network weights. The significant advantages of this strategy is that it does not require state feedback information related to the load, and only requires partial state feedback from the trolley as input, different from the general neural network that needs to be pre-trained to optimize the weights repeatedly, the strategy designed in this study does not need to train the weights in advance, and the values can be adaptively updated with the operation of the crane control system, and the closed-loop error stability of the system is proved by the relevant theories. Finally, the comparative experiments with the existing strategy and the robustness experiments are built to confirm that the proposed method in this study has a good anti-swing effect on the load and at the same time, it is robust to the changes of the system parameters, finally, the designed control system does not need the swing angle information of the load, which reduces the cost of development of the control system, and it is very practical.</p>