An improved grey wolf optimizer (MGWO) is used to optimize BiLSTM to predict water wall temperature. The improved algorithm adopts nonlinear factor adjustment strategy, adaptive position update strategy and dynamic weight modification strategy to improve the global optimization ability of the GWO. The improved grey wolf optimizer is used to optimize the number of hidden layers, learning rate and regularization parameters of the BiLSTM model to improve the prediction accuracy of the model. The data of a power plant in Xinjiang are used for prediction simulation. The results show that, the improved optimizer has higher prediction accuracy, and can predict the change trend of wall temperature when the unit is lifting and lowering load. Compared with the LSTM and BiLSTM models, the average root mean square error of the model reduces by 9.86% and 3.69%, respectively, and the overtemperature of water wall temperature can be predicted in advance, which is of great significance for the prevention of overtemperature of water wall.

In order to solve the problems of high error and low classification accuracy in the fault diagnosis process of wind turbines caused by the high dimension, feature redundancy and feature correlation of wind turbine supervisory control and data acquisition (SCADA) data, a three-stage feature selection method based on LightGBM-VIF-MIC-SFS is proposed. Firstly, based on the importance calculation of all features implemented by LightGBM, a preliminary feature space is determined. Secondly, a correlation discriminant matrix is constructed based on the variance inflation factor (VIF) and maximum information coefficient (MIC) to evaluate features with similar importance in a single screening, and discard input features with high similarity. Finally, the sequential forward search method is used to process the features for the third time, input the features obtained from the previous two feature selection one by one, and retain the features that can improve the system performance, so as to achieve the final feature selection. After the establishment of the model, the real SCADA data of the wind farm is used for performance evaluation, and the proposed algorithm is compared with the two comparison algorithms on six data sets. The results show that LightGBM-VIF-MIC-SFS has significant advantages over the two comparison feature selection algorithms. A ablation experiment was conducted on the three modules within the proposed algorithm, effectively verifying the effectiveness of each module within the proposed feature selection method and the rationality and accuracy of the optimal feature space obtained based on the proposed method.

In order to understand the key parameters and operating experience on fully burning and high proportion blend burning high-alkali coal in boiler with slag-tap furnace, long-term engineering tests were conducted based on a 300 MW boiler with slag-tap furnace in a power plant. An analysis was conducted on the possible combustion organization, nitrogen oxide control, slag flow, and ash deposition issues. Modifications were made to the combustion system, thermal modification system, and slag flow system. So long-term operational data and key parameter records were conducted, after fully burning and high proportion blending burning of 300 000 tons of high-alkali coal, all operating parameters of the boiler were normal. Based on recent operations and comprehensive tests, the optimized boiler with slag-tap furnace in a specific power plant demonstrates strong adaptability to high-alkali coal. Within the control range of coal ash components with w(Al₂O₃)<25%, w(Fe₂O₃)<15%, 12%<w(CaO)<30%, silica-alumina ratio>1.7, acid/alkali ratio>0.5, the normal operation of the boiler can be ensured without obvious ash deposition. Burning high-alkali coal can effectively reduce the inlet flue gas temperature of the first-level heating surface to below the design value, effectively avoiding the occurrence of slagging on the first-level heating surface of the boiler, and the generation of nitrogen oxides is also reduced by more than 30% compared with using the original design coal.

As a fundamental component of the organic Rankine cycle (ORC), the scroll expander's operating characteristics critically influence the ORC's overall performance. Initially, we establish a three-dimensional transient simulation model of the scroll expander. This allows us to systematically analyze the effects of various operating conditions on aspects such as suction pressure, exhaust pressure, rotational speed, and the output power and isentropic efficiency of the scroll expander, using numerical simulation. Following this, we study the effect of different operating conditions on the transient performance and mechanical properties of the scroll expander, and achieve a more comprehensive understanding of the mechanisms involved. Ultimately, we verify the accuracy of our numerical model using a laboratory-built test bench of the ORC low-temperature waste heat oil-free power generation system. The close correlation between the experimental results and numerical simulation outcomes authenticates the reliability and applicability of our numerical simulation method. In conclusion, this research's findings offer significant referential value for the design and optimization of the scroll expander.

As a kind of slag-tap boiler, cyclone-fired boilers exhibit significantly different aerodynamics comparing with pulverized coal (PC) boilers. However, by far there are still lack of CFD models that are able to provide effective guidance to the design and operation of cyclone-fired boilers. A CFD model of cyclone-fired boilers was developed in which the capture of coal particles by the molten slag layer was considered through a slag layer coal particle capture model. This model was then employed to investigate the aerodynamics and flue gas recirculation (FGR) optimization design of a 550 MW cyclone-fired boiler. The results demonstrate the highly nonuniform characteristics of furnace aerodynamics of cyclone-fired boilers due to the strong swirling flows created by the cyclones leads to the formation of localized high temperature zones and severe boiler fouling problems at the entrance of boiler convection pass. Thus, it is critical to adapt the FGR design with the nonuniform furnace flow and temperature distributions. The simulation results show that, with the optimized FGR design, the high temperature zones and the resulting severe fouling problems were effectively mitigated.

In order to study release characteristics of sodium during thermal conversion of high-alkali coals, the release characteristics of sodium in high-sodium coal and low-sodium coal were compared and analyzed through combustion experiments and pyrolysis experiments of raw coal and water-washed coal, so as to explore the release law changes of different forms of sodium in coal samples during combustion and the influence of atmosphere changes on sodium release. The results show that, during the combustion experiment, the release of sodium from high-alkali coal increases slowly at 300~500 ℃ and rapidly at 500~1 100 ℃, and the release of sodium from low-alkali coal increases rapidly at 300~500 ℃ and slowly at 500~1 100 ℃. It can be seen that, coal quality is one of the main reasons affecting sodium release, and sodium release will be greatly affected by the difference in coal composition. The release law of sodium during combustion and pyrolysis is basically the same, the sodium release rate changes slowly during the pyrolysis process, and is about 7.0% lower than that of the combustion process. The release characteristics of organic sodium and water-soluble sodium are different due to different release routes.

To enhance the whole process safety of fan operations and ensure accurate fault diagnosis and long-term production income of thermal power plants, predicting these risk issues is crucial to enhance the safety of the unit. In this paper, we proposed a fan fault diagnosis model of big data platform that integrates multilayer perceptron and polynomial fitting. The fan early warning model was established by multilayer perceptron and polynomial fitting modeling technology, and integrated into the big data platform to find abnormalities which were difficult to find manually during the operation of the fan. By combining data mining with mechanism analysis and feature value knowledge base, the parameters boundary information of fan stall could be excavated, the stall boundary conditions of the fan were accurately configured under various working conditions, and a stall boundary condition diagram was created. By combining those informations with normal operating conditions, the early stall zone can be obtained. Finally, a fault diagnosis model that covers the entire working condition of the fan can be established. Utilizing the comprehensive big data platform that covers, circulates, and maintains fan operation data, a system of intelligent fan patrol model was constructed. The intelligent patrol disk model which replaces the operator was then used to monitor and diagnose the fan running state regularly, which can achieve accurate and safe diagnosis of fan faults, minimize the fault incidence and maximize the personnel reuse rate.

With the continuous increase of installed capacity of new energy generation in power grid, thermal power units have to undertake more peak shaving. However, the flexibility and peak shaving capacity of current thermal power units are generally insufficient. A subcritical 300 MW coal-fired unit is retrofitted for molten salt energy storage. Six heat storage strategies and two heat release strategies are proposed and investigated. The influence of heat storage and release process on the peak shaving capacity and thermal performance of the unit under three working conditions is analyzed, the technical and economic analysis is performed in terms of the net present value. The results show the feasibility of extracting reheated steam for heat storage is higher, with a peak shaving depth of 58.9%. However, the coal consumption would increase at the same time. During the releasing heat stage, the maximum increment of power generation up to 11.3% of the rated power generation is achieved by heating the water supply to generate steam, while the higher temperature of the molten salt is required. Using high-temperature molten salt instead of low-pressure heater to preheat water supply is proved to have more advantages, while the power generation increment is relatively small. During the whole process of heat storage and release, the maximum circulating electricity efficiency can reach 0.987. The economic analysis of heat storage transformation is conducted. The dynamic investment payback period is 11.65 years, and the net present value is 49.118 million yuan. Therefore, the renovation scheme is feasible.

In order to explore the effects of mixed coal combustion on reducing the carbon content of fly ash in a subcritical 600 MW natural circulation balanced draft boiler, three types of coal with significant differences in characteristics were used, and numerical simulations were combined with experiments to study the mechanism of mixed coal combustion and its relationship with fuel characteristics, burnout rate, and carbon content in fly ash in the stratified combustion process of the balance boiler. Based on the results, a set of coal blending principles are proposed to improve the combustion characteristics of coal powder in power station boilers. Firstly, three extreme operating conditions were set up, and high-quality coal was respectively focused on the upper, middle, and lower layers of the burner for combustion. And then, through computational fluid dynamics (CFD) simulations, it was found that when the coal powder airflow of the balance combustion boiler flowed upward, and the high-quality coal was distributed in the upper layer of the combustion chamber while the poor-quality coal was distributed in the lower layer, the residence time of poor-quality coal in the furnace increased significantly, allowing it to fully combust in the high-temperature area of the upper layer, resulting in a low carbon content in fly ash. Furthermore, based on the analysis of the heat value, volatile matter and ash content of coal powder in different combustion layers under various working conditions, it was found that the difference in heat value and ash content is the main factor influencing combustion characteristics. Based on the actual operating conditions of the coal-fired power plant, a more realistic set of coal blending scenarios was established, and numerical simulations and on-site experiments were conducted, which showed that the overall combustion characteristics of coal powder under each working condition followed the aforementioned rules, and the carbon content in the fly ash reduced compared with the original operating conditions. Finally, the coal blending principles were established, which take into full consideration of the actual operating conditions of the power plant, recommend high-quality coal with low ash content to be used in the upper and middle layers of the combustion chamber while poor-quality coal with high ash content should be used in the middle and lower layers of combustion burner.

High-alkali coal such as Zhundong coal has huge reserves in Xinjiang, in which rich alkali metal elements can easily lead to fouling and slagging problem on heating surface of the furnace, thus to decrease the safety of boiler. It is of great significance to develop efficient and clean combustion power generation technology for high-alkali coal to achieve the “double-carbon” goal. The research progress of the online monitoring technology in high-alkali coal combustion furnace based on spontaneous emission radiation analysis is summarized. The development trend and dynamics are discussed focusing on the research status and application of emission spectrum technology and spontaneous emission radiation imaging and image processing technology in high-alkali coal combustion monitoring. Emission spectroscopy can obtain the temperature and component concentration by processing the spectral radiation signal emitted by the flame at different wavelengths. Recently, it has been widely used to measure the combustion temperature and the gaseous alkali metal concentration in the industrial furnace and judge the slagging trend in the furnace qualitatively. Different from emission spectroscopy technology, spontaneous emission radiation imaging and image processing technology has the ability to analyze the spatial distribution of signals. The technology obtains the spontaneous radiation image in the furnace via CCD, CMOS and other surface array sensors. Based on image processing technology and thermal radiation imaging theory, the temperature distribution of the combustion field in three-dimensional space can be obtained combining with solving radiation inverse problem, which makes it possible to monitor the three-dimensional visual of slagging formation. In the future, monitoring of the fouling and slagging on the heating surface in two or three dimensions should be carried out based on emission spectroscopy technology, spontaneous emission radiation imaging and image processing technology. Combined with the distribution of parameters such as combustion temperature and gaseous alkali metal concentration in the furnace, a quantitative judgmental index of fouling and slagging on the heating surface of high-alkali coal combustion should be established to achieve the goal of online prediction of fouling and slagging on the heating surface.