Selecting the gas-steam hybrid cycle as the research target, the mathematical model of the unit and the system is created. To compensate for the weaknesses of the BP neural network, the gravitational search method is employed to optimize the BP neural network, thus creating a GSA-BP gas unit condition monitoring and prediction model. The power station’s operational data over a period of around one year was selected, and 2,732 effective samples were obtained as research materials. Tests show that the forecasting error of the gas turbine’s output power can be steadily maintained below 0.5 MW, and the GSA-BP model is capable of accurately reflecting the characteristics of the unit under different loads. Based on the basic principles of the bidirectional RRT algorithm, the improved bidirectional RRT algorithm is proposed based on artificial potential field theory. The simulation test in the scenario of gas unit inspection shows that the improved bidirectional RRT algorithm performs better than the traditional RRT algorithm and bidirectional RRT algorithm in terms of optimal performance and planning speed. Finally, the intelligent inspection cockpit platform of gas units is developed. This intelligent inspection cockpit platform is of great significance to the transformation of the operating and maintenance shift mode and improving the automation level of operating and maintaining.
