The power grid equipment supply chain is affected by multi-level coupled structures and environmental disturbances, making it prone to dynamic risks such as supplier disruptions and logistics delays. Traditional methods (e.g., ARIMA-GARCH) have limitations in handling high-dimensional sparse risk data of the power grid equipment supply chain, including insufficient modeling capabilities and delayed response to sudden risks. To address this, this paper proposes a Compression Coefficient-based Stochastic Configuration Network model (CC-SCN). The model dynamically optimizes the network structure through adaptive compression coefficients, effectively suppressing redundant nodes and enhancing the feature extraction capability for high-dimensional sparse risk data. Additionally, the sensitivity of compression coefficients to abrupt changes in network residuals strengthens the ability to identify sudden risks. Experiments on the power grid equipment supply chain dataset show that the proposed model reduces the misrecognition rate to 0.87% and achieves a recognition rate of 96.3%, which is significantly better than traditional methods. It realizes multi-dimensional feature fusion and real-time recognition in complex risk scenarios, and solves the problems of inefficiency and poor generalization caused by redundant nodes in traditional SCN.