Under low-temperature conditions in winter, the fin-tube bundles of air-cooled radiators are prone to freezing. Exploring the variation patterns of the critical anti-freezing ambient temperature and critical anti-freezing circulating water flow rate of the indirect air-cooling system at different wind speeds and directions is of great guiding significance for ensuring the safe and stable operation of power plants. Taking a 2×350 MW indirect air-cooled unit as the research object, this study adopts the numerical simulation method, combining with the louver opening adjustment strategy, to investigate the critical anti-freezing characteristics of the indirect air-cooling system under different operating conditions. Through the analysis and calculation of the variation patterns of the flow and heat transfer performance of the indirect air-cooling system under different ambient meteorological conditions in winter, the variation laws of the critical anti-freezing ambient temperature and critical anti-freezing flow rate of the indirect air-cooling system at different wind speeds, wind directions and louver openings are revealed. The results show that an increase in ambient wind speed leads to a decrease in the critical anti-freezing ambient temperature and an increase in the critical anti-freezing flow rate. Reducing the louver opening can lower the critical anti-freezing ambient temperature, while the critical anti-freezing flow rate shows a trend of decreasing at first and then increasing. However, under the condition of high wind speed in the 270° wind direction, the critical anti-freezing flow rate decreases continuously with the reduction of louver opening. The research conclusions can provide operational guidance for the safe and stable operation of indirect air-cooled units in low-temperature winter environments.

The oily dust on photovoltaic (PV) modules significantly reduces power generation efficiency, but conventional cleaning agents suffer from poor cleaning performance, cause environmental pollution, and bring component corrosion risks. This study analyzed the dust composition via XRF, XRD, and ignition methods, revealing that organic matters with a mass fraction of 17.52% cause high adhesiveness. An eco-friendly cleaning agent was developed by optimizing the formulation via a four-factor three-level orthogonal experiment using Class A eco-friendly components. Its degradation and corrosion properties were verified through the continuous activated sludge method and immersion experiments. The results showed 10 g/m² oily dust could reduce the PV module power by 30.28%, while the cleaning agent achieved a cleaning efficiency of 99.84% at 30-fold dilution. The cleaning wastewater with a COD of 3 000 mg/L achieved a biodegradability of over 90.41%. Field application on oil plant rooftops restored 79.4% power efficiency and raised surface temperature by 5.6 ℃. Costing only 0.5 yuan per square meter and causing no corrosion to PV modules, the proposed cleaning agent outperforms commercial alternatives. It enables efficient and eco-friendly cleaning, ensuring stable operation of PV power stations and improving power generation efficiency.

To meet the high requirements of selective catalytic reduction (SCR) systems in coal-fired power plants for accurate and low-latency prediction of nitrogen oxides (NO_x) mass concentrations, this study designs and proposes a soft sensing and deployment framework that balances high accuracy and real-time performance. Using more than 110 000 sets of high-dimensional operational data from a 660 MW coal-fired unit, a systematic comparison of deep learning (DL) and XGBoost models was conducted on a unified platform. Time series cross-validation combined with grid search was employed to optimize hyperparameters, and model performance was comprehensively evaluated in terms of predictive accuracy, computational efficiency, and interpretability via local interpretable model-agnostic explanations. On this basis, an “edge-embedded” collaborative deployment strategy was proposed, in which the DL model is deployed on edge servers to deliver high-accuracy predictions, while the XGBoost model is embedded into the distributed control system (DCS) to ensure real-time responsiveness. The results show that the DL model outperforms XGBoost in dynamic response and predictive accuracy, achieving root mean square errors approximately 10% lower than that of the XGBoost, and maintaining stability under highly fluctuating conditions. Variable importance analysis highlights flue gas oxygen content, burner wall temperature, and total air volume as the dominant factors affecting NO_x formation. The proposed collaborative architecture can theoretically achieve millisecond-level inference and provide offline fault tolerance, offering a practical pathway for intelligent ammonia injection control and combustion optimization.

To address poor control performance of indirect air cooling systems under external disturbances, where conventional PID control fails to meet requirements of large time-delay, strong coupling, and nonlinear systems while generalized predictive control (GPC) offers superior performance but suffers from parameter tuning difficulties, an improved PSO-based GPC parameter tuning method is proposed. Transfer function and environmental wind speed disturbance models are established with a GPC controller using circulating water outlet temperature as the controlled variable. Targeting limitations of conventional PSO prone to local optima and low convergence accuracy, a DE-VPPSO hybrid algorithm combining differential evolution and velocity pausing mechanisms is designed, employing dual-population collaborative search to enhance search capability and convergence precision. The algorithm simultaneously optimizes four key GPC parameters: prediction horizon, control horizon, control weighting, and softening factor. Simulation results demonstrate that under step disturbances, wind speed disturbances, parameter mismatches, noise interference, and comprehensive operating conditions, the proposed method exhibits excellent control performance and robustness, effectively enhancing system performance.

In view of the limitations of excessive reliance on subjective experience in reactor fault risk assessment, a comprehensive evaluation method based on fault tree and fuzzy Bayesian network is proposed. A fault tree model covering three fault manifestations, abnormal vibration, abnormal temperature rise, and abnormal oil chromatography, is built to clarify the causal logic of 22 risk factors and then mapped into a Bayesian network. Trapezoidal fuzzy numbers are used to handle linguistic uncertainty in expert judgment, and expert weights are combined to obtain basic event probabilities. Forward reasoning yields the overall reactor fault probability, while backward reasoning and fault importance indices are simultaneously involved to identify key fault causes. Case analysis shows that the method accurately assesses fault risk, identifies fault causes, and provides a basis for operation and maintenance decisions. The proposed comprehensive evaluation method based on fault tree and fuzzy Bayesian network effectively reduces subjective dependence in reactor fault risk assessment, offering high practical value and broad application prospects.

The 800H alloy is a critical high-temperature material used for the main steam header of high-temperature gas-cooled reactor steam generators and is prone to creep damage and failure under prolonged high-temperature and high-pressure service conditions. In this study, tensile and creep tests were systematically performed on 800H alloy under various temperatures and stress levels to obtain its mechanical properties and detailed creep curves. The results indicate that the alloy exhibits an anomalous creep behavior in the early stage of creep, characterized by an initial decrease followed by an increase in creep rate, which is more pronounced under lower stress conditions. The true steady-state creep rate, accounting for this anomalous behavior, was determined using the Monkman-Grant relationship in combination with stress exponent analysis. Further investigation shows that this anomalous creep behavior significantly reduces the creep rate and influences the subsequent creep process. Quantitative analysis reveals that the creep rate at the anomalous point increases with increasing temperature and stress, while the duration of the anomalous stage increases with applied stress. These findings provide valuable insights into the creep behavior and strengthening mechanisms of the 800H alloy.

The ultra-supercritical double-reheat power generation technology has become an important development direction of thermal power generation technology due to its high efficiency and low emissions. A dynamic simulation model of the double-reheat boiler was established by the GSE software based on a 660 MW ultra-supercritical double-reheat coal-fired boiler with triple-rear passes. The variation laws of the main, primary, and secondary reheat steam parameters after disturbance by coal flow rate, feed-water flow rate, feed-water temperature and excess air coefficient under 100%THA working condition were calculated in detail. Moreover, the variation laws of steam temperature at the boiler outlet under the working conditions of 100%THA, 75%THA and 50%THA were compared. The simulation results show that the response of steam parameters at the boiler outlet is the slowest after feed-water temperature is disturbed, and the stability time is approximately 2 300 s under 100%THA working condition. The response of steam parameters at the boiler outlet is the fastest after excess air coefficient is disturbed, and the stability time is approximately 720 s. The stability times of the steam parameters at the boiler outlet are approximately 1 040 s and 1 250 s respectively after coal flow rate and feed-water flow rate are disturbed. In the initial stages of transient processes after coal flow rate, excess air coefficient and feed-water temperature are disturbed, there are short-term “reverse” changes in the steam temperature at the boiler outlet. In addition, the thermal inertia of the double-reheat boiler will increase as the boiler load decreases. The simulated dynamic characteristics of the double-reheat coal-fired boiler with triple-rear passes can provide the basis for further optimization of the unit operation control strategies.

Driven by the “dual-carbon” goals, the establishment of integrated energy systems in industrial parks, and the large-scale renewable integration have imposed heightened flexibility requirements on coal blending for thermal power units. The coal blending process comprises two stages: pre-furnace and in-furnace operations. During pre-furnace blending, a minimum coal quality deviation model addresses low-calorific-value coal utilization. Chaos search-based adaptive mutation particle swarm optimization blends such coal into furnace-compliant mixtures meeting boiler specifications. For in-furnace blending, dynamic adjustment of coal ratios across load ranges ensures load stability while minimizing fuel costs. A two-stage optimization model resolves circulating fluidized bed (CFB) boiler blending: Stage 1 selects coal feeder combinations according to weekly peak chemical plant loads and PV generation scenarios; Stage 2 optimizes coal feed rates under load-balance constraints, incorporating desulfurization-driven sulfur content limits. Comparative analysis under spring irradiance conditions reveals that in-furnace blending of two coals reduces daily combustion costs by 4.36×10⁵ yuan. Post-retrofit evaluation of blending of three coals demonstrates a further reduction in daily fuel costs.