Municipal solid waste incineration state recognition system based on deep convolutional stochastic configuration...

To address the fluctuating combustion caused by the complex and variable composition of municipal solid waste incineration (MSWI), this paper proposes a state recognition system for MSWI based on a deep convolutional stochastic configuration machine (DCSCM). The system consists of high-temperature cameras, a network switch, an industrial control computer, and a server, which employs DCSCM for intelligent assessment of combustion states. The DCSCM simulated expert prior knowledge via a linear predictive basis model, incrementally constructed convolutional structure, and incorporated an adaptive optimization mechanism based on error feedback. The mechanism dynamically evaluated the model convergence and enabled adaptive construction of model width and depth before convolutional kernels were populated. The results demonstrated that the trained DCSCM achieved a parameter size of 376 KB and a recognition accuracy of 97.32%, with its operational deployment and self-optimization in complex environments improving the average accuracy by 1.20%. The study provides support for precise combustion parameter control in MSWI, enabling automated monitoring of combustion states and reduced pollutant emissions.