Due to the uneven distribution of combustion temperature field in the operation processes of thermal power plants, issues such as over-temperature, tube explosion and flue gas corridor of the furnace heating surfaces may occur. By evaluating the temperature field distribution of the boiler during operation processes, the combustion adjustment can be made to improve the safety, stability and economy of the boiler operation when the operation status can be examined timely. With the boiler of a 660 MW ultra-supercritical power plant as an example, the combustion optimization adjustment test and simulation were carried out. Temperature of heating surfaces and the wall temperature change at the burner were analyzed, and the wind speed and pulverized coal concentration deviations of powder pipes in the pulverizing system, internal and external secondary air volume and intensity of the burner, the adjustment of the direct current air volume, and the wall air in the furnace were measured. Results show that the factors, such as the combination operation modes of different coal mills and the modes of secondary air distribution have a great impact on the boiler operation. In particular, the wall over-temperature, the burner temperature and reducing atmosphere on the heating surface are influenced, which will affect the boiler economic operation. After adjustment, the wall over-temperature phenomenon of the screen superheater, high temperature superheater and high temperature reheater disappears, there is no burnout of burners, and the amount of NO_x generation is also within a reasonable range.

To explore the deep peak shaving capacity of coal-fired boilers, the hydrodynamic force and wall temperature characteristics of the water-cooled wall of a supercritical 630 MW boiler were studied. Mathematical models of the hydrodynamic force and wall temperature characteristics were established, and water-cooled wall pressure drop, flow distribution, outlet working substance temperature and wall temperature distribution along the furnace height between 20% and 30% boiler maximum continuous rating (BMCR) loads were analyzed. Results show that, the helical tube sections of the water-cooled wall exhibit small thermal deviation and uniform heat load distribution under 20% and 30% BMCR loads. The outlet steam temperature deviation of the vertical tube sections of the water-cooled wall is large, and the over-temperature is more likely to occur in this area. Under 30% BMCR load, the boiler could operate safely in dry state condition. Under 28.8% BMCR load, the inlet water temperature of the water-cooled wall is 280.0 ℃, and the maximum wall temperature of the water-cooled wall in dry state condition exceeds 450.0 ℃. Under 27.3% BMCR load, the maximum wall temperature of the water-cooled wall in dry state condition exceeds 500.0 ℃, and the boiler should operate in wet state condition or the wall temperature should be reduced by adjusting the inlet water temperature.

Taking IEA 15 MW floating offshore wind turbine and UMaine VolturnUS-S semi-submersible platform as research subjects, a structure design process of wind turbine floating platform was proposed in this study. The stability analyses on the dimensional parameters of each component in the floating platform were conducted using Ansys Aqwa hydrodynamic analysis software and OpenFAST simulation analysis software. The impact of column spacing, side column height, side column diameter, pontoon height, pontoon width, and heave plate diameter on the overall dynamic response of wind turbine system and mooring cable tension was investigated. The effect of dimensional changes on the stability of wind turbine system was explored. Results show that increasing the column spacing and side column diameter can significantly enhance the overall stability of floating wind turbine system during the operation. The changes of side column height above the waterline have almost no impact on the stability of wind turbine system. Pontoon width and pontoon height primarily affect the motion response in the heave direction, while increasing the heave plate diameter significantly improves the longitudinal and heave direction stability of floating platform.

Influence of hole configuration including cylindrical, fan-shaped and laidback fan-shaped holes on effusion cooling performance under different pressure drops of cooling air was examined using infrared thermal imaging technology. Results show that due to the vortex impact effect, low local cooling efficiency areas appear on the wall surface, which are independent of the hole type. The change in hole type only affects the cooling effect of the wall surface. Increasing the pressure drop of the cooling air can improve the cooling effect, and the area of the low cooling effect zone also decreases accordingly. The difference between the cooling hole shape and the pressure drop of the cooling air causes the movement of the lowest cooling effectiveness position on the wall. As the velocity of cooling air increases, this point moves upstream. In terms of the uniformity, the minimum effectiveness of cylindrical holes is lower by 14.3% to 27.5% than the area-averaged one. The difference between the minimum cooling efficiency of the fan-shaped hole, the expansion hole and the average level is greater, and the uniformity of the cooling effect is worse. The comprehensive cooling efficiency of the expansion hole is the highest, but the uniformity is the worst.

For the supercritical carbon dioxide (S-CO₂) with heated flow in a horizontal tube, to research the influence of the buoyancy effect on flow and heat transfer characteristics, experiments were conducted on flowing and heating in optical and inner coiled wire tubes under the same condition. Investigations were carried out for the influence of the wall heat flux density and mass velocity on the buoyancy effect, while discussions were given for the influence of the inserted coiled wires on the wall temperature distribution, heat transfer mode, etc. Results show that the temperature stratification caused by the buoyancy effect occurs along the gravity direction during SO₂ horizontal flowing with heating, which leads to the severe wall temperature divergence at the top and bottom, with a maximum wall temperature difference of up to 48 K. The inserted coiled wires in tube can effectively inhibit the temperature stratification caused by the buoyancy effect, while the heat transfer coefficients at the bottom and top are increased by more than 60% and more than 140%, respectively. Based on the validation of experimental results with four heat transfer correlations, it is found that Bishop correlation has the best comprehensive performance.

A turbine with high cycle fatigue failure was used as the simulation model, the coupling method of three-dimensional unsteady flow field and solid vibration response analysis was adopted, the simulation of fault point was carried out. According to the different conditions of turbine expansion ratio and turbine reaction force, the law of aerodynamic load distribution and aerodynamic exciting force caused by aerodynamic state change was compared, at the same time, the relative variation trend of blade vibration stress was obtained. Results show that under unchanged intake conditions, the average exciting force, the amplitude of exciting force and the vibration stress of the blade increase obviously with the increase of the expansion ratio of the turbine. While the turbine reaction force increases, the average aerodynamic exciting force of the blade increases obviously, but the amplitude of the exciting force and the vibration stress have a downward trend.

To solve the problem of ANSYS-Fluent software being unable to provide both high-precision and high-efficiency radiation spectrum models, secondary development was carried out for the full-spectrum correlated-k distribution (FSCK) model based on look-up table method and machine learning method, and the model was embeded into Fluent and coupled with built-in radiative transfer equation (RTE) solvers for radiation heat transfer calculation for mintures of common combustion gases and soots. Radiation heat transfer results of one-dimensional slabs and two group of flames were calculated by the model, and using the line by line (LBL) model as a benchmark, which were compared with the results calculated by gray gas weighted sum (WSGG) model in Fluent. Results show that the FSCK model yields more accurate solutions than the built-in WSGG model in Fluent, regardless of the presence of soot.

To address the response issues caused by time lag and load fluctuations in NO_<i>x emission control of SCR denitrification system, reinforcement learning was employed to adjust the proportional-integral-derivative (PID) parameters. The loss function of Critic network was redesigned according to the deep deterministic policy gradient (DDPG) algorithm, and a delay queue was introduced to simulate system latency. The proposed control strategy has been applied to a 660 MW ultra-supercritical coal-fired power unit in China. Results show that the reinforcement learning control method is superior to traditional PID control in terms of adjustment time, overshoot, and stability. The proposed strategy overcomes the time lag and load fluctuations that traditional PID control cannot resolve, demonstrating the significant theoretical and practical values.

From the perspective of gas turbine blade cooling design, the study and application progresses of data-driven methods of domestic and foreign researchers in recent years was summarized. Three types of data-driven methods including mathematical statistics, machine learing and deep learning were introduced. The advantages of data-driven methods over experimental and numerical simulations were expounded. And the application of data-driven methods was emphatically summarized, mainly including predicition and uncertainty quantification of turbine blade cooling characteristics, improvement of turbulence models in numerical simulation, and data fusion of existing data and knowledge. Based on the current research hotspots and development tendency, further studies focus on data-driven methods in gas turbine blade cooling design were proposed, including exploring the potential of data fusion approches, improving generalization ability, reducing data cost, studying data processing methods, comparing the advantages and disadvantages of different data-driven methods, and developing data-driven approaches for complex cooling designs.

Considering the impact of multi-factor coupling in the power market on power price forecasting, a power market price forecasting model based on long-short term memory (LSTM) neural network was established. The historical new energy output, power transmission, power load and other factors affecting the power supply and demand relationship were set as the boundary factors of LSTM learning parameters, and data preprocessing was carried out. The model parameters of LSTM neural network, including the number of layers, iteration times and learning rate, were optimized to generate a training model for electricity price forecasting, which was then used to forecast the electricity price curves for trading days. The correctness of the proposed method was verified through example simulation, a prediction scenario for electricity trading in a spot market province was constructed, and the evaluation indicators of power price forecasting accuracy were introduced. The results indicate that this method provides a reference for research on day-ahead clearing price prediction and offers market participants in the electricity market effective bidding strategies to achieve substantial marketing revenue.

In order to solve the problems of low speed and low accuracy of rolling bearing fault diagnosis caused by the inconsistent sampling frequency of the accelerometer of mechanical vibration monitoring system, a rolling bearing fault diagnosis method based on variational mode decomposition-multi-strategy tuna swarm optimization-extreme gradient boosting (VMD-MTSO-XGBoost) at wide sampling frequency was proposed. Firstly, the vibration signal was de-noised by wavelet and de-sampled to get the de-noised signal at wide sampling frequency. The de-noised signal at wide sampling frequency was processed by variational mode decomposition (VMD), and the intrinsic mode function (IMF) component index was extracted to form the fault feature vector. Secondly, the Circle chaotic map was used to initialize the tuna swarm optimization (TSO) population to increase the richness and diversity of the initial population. In order to improve the ability of the algorithm to jump out of the local optimum and enhance the ability of the algorithm to explore the whole world, the dimension-by-dimension mutation method was used to disturb the optimal individual position. Finally, the modified tuna swarm optimization (MTSO) algorithm was used to optimize the parameters of extreme gradient boosting (XGBoost), and the rolling bearing fault diagnosis model was established. The proposed fault diagnosis method was validated by the Case Western Reserve University dataset, the German University of Paderborn dataset and the measured dataset. Results show that at wide sampling frequency, the fault diagnosis method presented in this paper can identify rolling bearing faults more efficiently and accurately compared with the other three models.

To enhance the energy efficiency of the thermal energy storage system using superheated steam to heat molten salt, a coal-fired unit system configuration integrating the three-tank molten salt thermal energy storage system was proposed. The flexibility improvement potential of coal-fired power plant and the thermodynamic performance of the integrated system were analyzed under typical charging and discharging conditions. Results show that under the condition of 30% THA for the charging process, the maximum output power of coal-fired power plant can be reduced by 81.60 MW, accounting for 12.36% of the rated load, due to the limitation of the minimum steam flow rate through the low-pressure turbine. Under the condition of 50% THA for the discharging process, the maximum output power of coal-fired power plant can be increased to 52.45 MW, accounting for 7.95% of the rated load. The equivalent round-trip efficiency of the integrated system is influenced by the discharging condition. The maximum equivalent round-trip efficiency is 75.35% at the discharging condition of 50% THA, while the minimum equivalent round-trip efficiency is achieved at 59.43% at the discharging condition of 100% THA.

Phase change materials (PCM) are effective materials for thermal energy storage with a wide range of application prospects. However, their inherent low thermal and electrical conductivity greatly hinder their practical application in the field of thermal energy storage. By compositing PCM with different energy conversion materials, efficient mutual conversion among various forms of energy and thermal energy has been achieved. The composite PCM plays a key role in solar photothermal/electrothermal conversion and storage. In the field of photothermal conversion, the compositing of high thermal conductivity photothermal conversion materials with PCM has optimized the thermal conductivity and effectively enhanced the heat storage capacity, photothermal conversion efficiency, and solar energy utilization efficiency. In the field of electrothermal conversion, the preparation of electrothermal composite phase change materials by introducing high electrical conductivity supporting materials has not only improved the electrothermal conversion efficiency but also enriched the utilization methods of electrical energy. A detailed review of the latest research progress on photothermal and electrothermal conversion materials was provided, and the materials were classified. Aiming to offer new insights for the advancement of energy storage technology, the advantages and disadvantages of various types of materials were analyzed, the mechanisms for improving thermal energy conversion efficiency were revealed, and future development directions and challenges were prospected.

To meet the demand for flexible regulation performance of traditional thermal power units, a data-driven nonlinear environmental modeling method for supercritical combined heat and power (S-CHP) and a reinforcement learning control method based on double-delay deep deterministic strategy gradient (TD3) were proposed. Firstly, a data-driven nonlinear model environment was established based on the deep learning algorithm of multi-layer perceptron (MLP) and the dynamic characteristics of the S-CHP unit. Furthermore, based on deep reinforcement learning algorithms, an actor critic strategy value network and S-CHP specific state values and reward functions were designed to conform the dynamic characteristics of the S-CHP unit. A flexible control strategy of TD3 was proposed to achieve the control objectives of rapid response, ensuring heating supply and stable operation. Results show that compared with single-layer networks, the MLP model reduces the root mean square error by 51.7% at the rated power of 52%-93%. Compared with the traditional coordinated control method, the TD3 control strategy has better tracking effect and higher response rate.

To address the problems of insufficient accuracy and stability of the existing photovoltaic (PV) power prediction models, a PV power prediction model was proposed. This model was based on sky image data, the simple model of the atmospheric radiative transfer of sunshine (SMARTS), and the LK optical flow method. The SMARTS was used to calculate the clear sky irradiance and solar position information at a specified time, and the cloud motion vectors were obtained through the LK optical flow method. The cloud motion vectors were used to infer the shading situation of the cloud to the direct sunlight at future times. Then, the clear sky irradiance and the predicted solar shading situation were used to calculate the final power prediction result. The model was validated using the sky image dataset SKIPP'D. In a sunny environment, the proposed method was compared with the Bird model and the Ineichen model. In a cloudy environment, the proposed method was compared with the long short-term memory (LSTM) neural network model and the convolutional neural network (CNN) model. The effectiveness of this method in short-term PV power prediction has been verified. Results show that, the proposed method can accurately capture the effect of cloud changes on the PV power generation in a cloudy environment, the determination coefficient R² of the model is greater than 90%, and both the prediction accuracy and stability are significantly better than those of the control models.

In order to verify the feasibility of blending a large proportion of ammonia in a coal-fired unit, a numerical simulation model was established using a 660 MW wall type tangentially-fired boiler as an example. The impact of ammonia blending ratio, ammonia injection location, and primary and secondary air distribution methods on combustion in the furnace was analyzed. Results show that blending ammonia in coal-fired boiler leads to a decrease in flue gas temperature and an increase in NO_x mass concentration in the furnace. The greater the ammonia blending ratio, the more significant the decrease in flue gas temperature and the increase in NO_x mass concentration. When the ammonia blending ratio increases to 40%, the average flue gas temperature in the furnace drops by approximately 50 K, and the outlet NO_x mass concentration increases by 29.4%. Measures such as blending ammonia in higher-level burners and increasing the primary air ratio can reduce the NO_x mass concentration at the furnace outlet while ensuring complete combustion of ammonia and coal.

To address the difficulty of fault diagnosis in thermal power equipment with multi-parameter coupling and gradual changes, a fault early warning method based on least absolute shrinkage and selection operator (LASSO) regression feature selection and bidirectional long-short term memory (BiLSTM) multivariate regression prediction was proposed. Taking a coal mill in a 1 000 MW power unit as the research subject, feature parameters such as mill current, outlet pressure, and inlet-outlet differential pressure were selected to represent blockage faults. LASSO regression was employed to select the feature variables, and a multivariate regression prediction model was established based on the BiLSTM algorithm. According to the variation mechanism of the feature parameters during mill blockage and the predicted values of the model, a mill blockage fault index was constructed. Finally, the warning threshold was determined using the kernel density estimation method, enabling mill blockage fault warnings. Actual data analysis shows that when the coal mill is operating normally, the average relative error of the BiLSTM multivariate regression prediction model is 1.13%. Compared with the traditional error back-propagation (BP) neural network and support vector regression (SVR) model, it has higher accuracy and the ability to predict the trend of parameter change. When the coal mill is operating abnormally, this method can detect operational abnormalities earlier than the multivariate state estimation technique (MSET) algorithm model, enabling early fault warning under variable operating conditions of the coal mill.

Taking a certain gas-fired heater as the research object, synergistic rules of NO_x and CO₂ production under different proportions of biomass gas co-firing were studied. Results show that compared with pure gas combustion, co-firing biomass gas can reduce furnace flame center, peak temperature of furnace flue gas and outlet flue gas temperature of the convection chamber. When the co-firing ratio is 10%, the volume fraction of carbon decreases by 25.3%, and the volume fraction of NO_x at the outlet is reduced by 67.1%. When the co-firing ratio exceeds 10%, the outlet NO_x production decreases slowly. When the co-firing ratio is 5%-10%, the CO₂ and NO_x production shows a positive synergistic trend, while when the co-firing ratio is within the range of 15% to 30%, the generation amounts of CO₂ and NO_x show a negative synergistic trend. Therefore, to synergistically reduce the generation amounts of CO₂ and NO_x, the optimal co-firing ratio is 10%. At this ratio, the furnace outlet volume fraction of CO₂ is the lowest(2.30%), and the volume fraction of NO_x emissions is 49.8×10^-6.

Wave is a main contributor to the power fluctuations in floating offshore wind turbines (FOWTs), and therefore challenges the large-scale grid integration of FOWTs. To reduce the impact of waves on wind turbine power, a power control method for floating turbines based on wave interval estimation and prediction was proposed. Focusing on the rated power tracking control of the FOWT above the rated wind speed, a FOWT model with input time delay under large inertia and a wave disturbance model were established, a wave predictor based on wave interval estimation was designed, and then a composite time-delay control strategy was developed combining wave compensation and integral sliding mode control. The operation of a semi-submersible FOWT under the present control strategy was simulated using a FAST-Matlab/Simulink co-simulation environment and compared to the system response under traditional PI control. Results show that the proposed power control strategy can effectively compensate for the impact of wave disturbances on FOWTs, significantly reduce power output fluctuations, and provide a reference for the design of control systems for floating wind turbines.

A compressor performance prediction model was proposed for certain sCO₂ coal-fired cycle unit, and the compressor, turbine and boiler models were coupled with the cycle system. Results show that the characteristic sizes and capacity of the compressor follow the 0.5 power law. As the capacity and rated speed increase, the isentropic efficiency improves accordingly. Power consumption is reduced and efficiency is increased when the inlet parameters approach the critical point. When the inlet conditions are fixed, the isentropic efficiency shows a parabolic trend of first increasing and then decreasing with the rise of outlet pressure, with the peak corresponding to the optimal pressure ratio. When the shaft speed of the cycle unit is constant, the thermal efficiency shows a parabolic distribution with the increase of capacity, with the peak appearing at the 300 MW capacity point. If the shaft speed is optimized in coordination with the capacity, the thermal efficiency curve remains parabolic, but the peak shifts to the 600 MW capacity point, due to the weighing effects of turbine efficiency improvement and boiler pressure drop increase.

Based on Aspen Plus software, a carbon capture system model of high-temperature Ca-based adsorbent was established, and simulation study was carried out for CO₂ capture process of flue gas in a 1 000 MW coal-fired unit. Aiming at the problems of high energy consumption and pollutants emission in energy supply of oxygen-enriched coal combustion in calciner, external electric heating energy supply was proposed to provide heat for calciner. The changes in key performance parameters after flue gas entering the carbon capture system were investigated in detail. On this basis, the conservation of energy and mass, as well as the effects of different operating parameters on system performance were analyzed. Results show that after the carbon capture system is added in the 1 000 MW coal-fired generating unit, the thermal efficiency of electricity generation will decrease by 26.11%. When the temperature of the carbonization furnace is maintained at 645 ℃, CO₂ capture efficiency will be 92.13%. CaCO₃ will be decomposed more thoroughly when the calcination furnace temperature is set at 900 ℃. The increase of temperature in carbonation furnace will reduce energy consumption of calcination and improve thermal efficiency of power generation, while the increase of calcination temperature will increase energy consumption and reduce thermal efficiency of power generation. The increase in calcium-carbon molar ratio leads to the increase of solid circulation in the calcium-based carbon capture system, while average carbon conversion of calcium-based adsorbents shows negative correlation.

Aiming at the problem of early fault diagnosis of rolling bearings, a fault diagnosis method of red-tailed hawk (RTH) algorithm optimization with feature mode decomposition (FMD) and 1.5-dimensional spectrum was studied. Firstly, through theoretical analysis, pulse energy factor index (PEFI) was designed and used as fitness function. Secondly, RTH algorithm combining with FMD was used to search the key parameters of the optimal decomposition in parallel. Then, the optimal signal component after decomposition was selected by PEFI, and the envelope demodulation was performed. Finally, the 1.5-dimensional spectrum of envelope signal was calculated, and the characteristic frequency information of bearing fault extracted was analyzed in the spectrum diagram to realize the accurate diagnosis of early weak bearing fault. The results of simulated fault experiments and engineering case analysis show that the proposed method can solve the problem of parameter self-adaptation, greatly reducing the impact of noise and other interfering components on diagnosis. It possesses good robustness and can effectively extract weak feature information from early bearing fault signals. It has important practical engineering reference value.

To solve the problems of the high energy consumption of CO₂ regeneration and the superior difficulty of CO₂ utilization in the flue gas decarbonization system with traditional monoethanolamine (MEA) method, a novel ammonia decarbonization system for producing NH₄HCO₃ was proposed based on the water photovoltaic electrolysis for hydrogen production and Haber-Bosch ammonia synthesis process.This system utilized photovoltaic electrolysis of water to produce green hydrogen, which was further synthesized with N₂ to produce green ammonia (NH₃). The NH₃, in solution form, reacted with CO₂ in flue gas in an absorption tower to generate NH₄HCO₃. Simultaneously, the rich liquor (containing NH₄HCO₃) was cooled and separated to produce NH₄HCO₃, achieving the co-production of electricity, heat, and NH₄HCO₃ with low carbon emissions, and avoiding the challenges associated with large-scale storage and utilization of CO₂. Based on a 300 MW coal-fired unit, a simulation model of the system was established in Aspen Plus V11 software. Calculation and analysis results indicate that the system can remove 1.546 8 million tonnes of CO₂ annually from flue gas (with a decarbonization rate of 85%), and produce 2.792 million tonnes of NH₄HCO₃, 1.034 2 million tonnes of by-product O₂, and 4.731 4 million tonnes of hot water annually. It requires the configuration of a photovoltaic power unit with an installed capacity of 2 594 MW and an ammonia production system with an annual output of 620 500 tonnes. The equivalent cost of carbon dioxide removal is 233.96 yuan per tonne.

Most existing data-driven energy optimization and energy-saving diagnostic methods were based on offline global models for analysis. Complex and variable states of the units during actual operation could degrade the performance of these models. Regarding this issue, a method for economic efficiency diagnosis of the unit was proposed based on just-in-time learning (JITL) and condition matching. This method involved preprocessing historical operational data of the unit, selecting boundary parameters and relevant parameters related to the target parameter. The conditions were divided into different cells under various boundaries through grid partitioning and the samples with optimal performance were selected as the benchmark condition sample set, and the target condition database was established. Condition matching was conducted based on this database. If the matching was successful, the optimal performance value under the corresponding boundary was selected as the parameter benchmark. Otherwise, a JITL approach was employed, which extracted similar condition samples through similarity measurement, and established a local online model to output the predicted value as the benchmark for performance evaluation and control adjustment. Results show that benchmark values of plant electricity consumption rate for different operating conditions and optimization strategies for desulfurization electricity consumption rate can be obtained based on the proposed method, providing reliable guidance for unit operation optimization.

To meet the basic requirements of performance regulation for centrifugal compressors, the coupling effect between performance regulation and anti-surge control under different operation conditions of the compressor were deeply analyzed. A control strategy for decoupling the performance controller was proposed based on conditions such as the compressor's working point and load increase/decrease. Additionally, by adopting the quasi-nonlinear modeling method, the compressor system and control models were developed, which could be solved in real time on a unified platform. Model solving was conducted via a multidisciplinary simulation platform, yielding the compressor's dynamic characteristics under variable conditions and verifying the control effects of the performance control and decoupling strategy. Results show that the proposed method achieves performance control, realizing the design concept of prioritizing anti-surge valve closure during load increase and rotational speed reduction during load decrease, ensuring stable unit operation when the compressor runs under variable loads.

For the forecasting of ultra-short-term photovoltaic power, a hybrid prediction model based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), variational mode decomposition (VMD) and bidirectional gated recurrent unit (BiGRU) was developed. The photovoltaic power generation signal was decomposed using CEEMDAN, and the decomposed signals were clustered and reconstructed using sample entropy and the K-means methods. Then, the VMD was applied for the secondary decomposition of complex signals to mitigate signal non-stationarity. The decomposed signal components were employed as inputs for training, validation and prediction in the BiGRU model. Subsequently, the predicted results from each signal component were linearly combined to obtain the final forecasting results. Results show that the hybrid model outperforms single models, confirming the effectiveness of the model. By comparing the forecasting performances under typical weather conditions and various evaluation metrics, the generality of the proposed method was validated.

Hydrogen-blended gas turbines represent an innovative energy conversion technology integrating hydrogen with electricity. Developing a dynamic model for these turbines is crucial for understanding hydrogen blending characteristics and precisely adjusting control parameters. Based on mechanism modeling method, detailed component models for the compressor, combustion chamber (for hydrogen and natural gas mixtures), and turbine were established. Using transfer learning methods, models for hydrogen-blended combustion efficiency and other gas turbine component characteristics were developed, which were then integrated to form a thermodynamic dynamic model of hydrogen-blended gas turbines. The model's steady-state and dynamic validation was conducted using on-site operation and test data of a small F-class hydrogen-blended gas turbine. The results demonstrate that the constructed model is not only applicable for simulating gas turbines using hydrogen-blended natural gas fuel, but also for simulating load increase and decrease during dynamic hydrogen blending ratio adjustments. It can provide an accurate simulation model for research on hydrogen blending ratio control strategies for gas turbines.

Based on the 660 MW coal-fired generator unit, a new solar-coal complementary power generation system was proposed, and the key equipment model and subsystem model were established and verified. The dynamic characteristics and peak regulation performance of the system were studied by simulation method. Results show that after configuring the heat storage system, the system can restore balance in about 10 minutes with minimal changes in the main parameters. The system can achieve continuous and stable operation on typical daily after configuring a thermal storage subsystem with a storage duration of 10 hours. By the configuration of heat storage system and extraction steam storage process, the load range of the system is expanded from 198.00-660.00 MW to 187.62-723.13 MW, and the peak regulation performance is significantly improved.

Based on the operation mechanism of combined heat and power (CHP) system, focus product concept was introduced by combining multidisciplinary concerns with heat and power products generated by the system, and the focus products were aligned with E modules. By leveraging the relationship between product quality analysis and output evaluation, a comprehensive multi-E evaluation system was constructed, and a holistic multi-E analytical theoretical system was objectively established. Using the concepts of universal set and subset from set theory, evaluation indicators were treated as elements of a set, therefore the process of E module determination was transformed into the filtration process from the universal set to its subsets. Taking a typical CHP system as the example, combined with the multi-E theoretical system, the subset filtration processes of CHP system under various optimization objectives were analyzed. Results show that the constructed multi-E evaluation analysis method can provide valuable reference for similar research.

A T-shaped molten salt receiver was designed and modelled. A 3D numerical model of the reciver was established, coupling the boundary conditions of the beam-down concentrated solar radiation with molten salt flow, heat transfer, and solar energy absorption. The energy loss mechanisms of the receiver were revealed and the impacts of the molten salt's extinction coefficient, flow distribution, and filter mesh location on the internal temperature distribution of the receiver were analyzed via this approach. Results show that the outlet molten salt temperature reaches 560-570 ℃. The main energy losses are re-radiation loss, convection loss and optical reflection loss. The receiver with smaller extinction coefficient exhibits higher internal temperatures compared with receivers with larger extinction coefficient and shallow filter mesh, as the substantial heat absorbed by deep-layer molten salt cannot be rapidly discharged, leading to a heat accumulation inside the receiver.

A vortex generator was installed on a flat film experimental platform to generate streamwise vortices, and the effect of secondary flow in the blade end region on the film cooling effect of multi-row fan-shaped holes was studied. The adiabatic film cooling efficiencies of four row film holes were experimentally measured, and the influence of the film cooling efficiency of each row of holes was analyzed based on the Sellers formula. Results show that the air film coverage distorts along the motion path of the vortex due to the influence of the vortex direction, leading to a decrease in the cooling effect of the air film and the formation of hot spots at the core of the second row of vortices. The influence degree of vortices is related to the air film blowing ratio, and near wall vortices mainly reduce the air film cooling effect of the first two rows of holes. The cooling efficiency of the first row of holes significantly decreases, while the decrease value of the second row of holes is decreasing and the impact on the rear row of holes is gradually weakened. The air film after the third row of holes is basically not affected.

The occurrence modes of mercury in middlings from different origins were studied by using sequential chemical extraction, and the thermal stability of various occurrence forms was investigated by combining temperature programmed decomposition-atomic fluorescence spectrometry (TPD-AFS) technique. Results show that mercury in middlings mainly exists in the forms of exchangeable Hg, carbonate and oxide surface bound Hg, organic bound Hg, sulfide bound Hg, silicate bound Hg and residual Hg. In middlings, the dominant form of mercury is the sulfide bound Hg, accounting for 60.67%-62.93% of the total mercury content, there are two release peaks, mainly released at 150-400 ℃, and less at 400-600 ℃. The content of residual Hg accounts for 14.63% to 25.31%, mainly released within the temperature range of 150-300 ℃. The content of organic bound Hg accounts for 3.98% to 7.99%, mainly released within the temperature range of 400-600 ℃. The content of silicate bound state Hg ranges from 2.18% to 3.34%, and it is released after 600 ℃. The content in the exchangeable state Hg ranges from 1.01% to 3.20%, and it shows no obvious release during pyrolysis. The content in the surface bound state Hg of carbonate and oxide is the least, ranging from 0.46% to 1.33%, and it is mainly released within the temperature range of 180-340 ℃.

To address future large-scale uncertainties in long-distance power delivery and safe economic grid connection of new energy, it is necessary to promote the efficient joint operation with new energy and flexible energy storage to meet the channel capacity requirements, while considering the flexible retrofitting of thermal power. Taking a large-scale renewable energy base in northern China as an example and considering the randomness and volatility of wind and solar energy, a multi-energy complementary capacity configuration optimization model was constructed based on the joint probability distribution of wind and solar energy using the Copula function. The improved sparrow search algorithm was used to solve the optimal capacity configuration scheme, and simulation analysis was performed for typical days in summer and winter. The influence of pumped storage power stations with different installed capacities on the volatility of thermal power units, system costs, operating conditions and environmental benefits was compared. Results show that the multi-energy complementary power generation system can both improve operating conditions and enhance the economy of operation, with significant environmental benefits.

Aiming at the problem of insufficient feature learning ability and poor diagnosis effect of traditional deep neural network under the variable operating conditions of wind turbine rolling bearings, a fault diagnosis model was proposed, which combined channel attention mechanism (CA) and capsule residual network (CPRN),to achieve high performance fault diagnosis of rolling bearings under varying working conditions. Firstly, the corresponding time-frequency map was generated by continuous wavelet transform of the vibration signal. It was used as a training sample after matrix reconstruction. And the weight of different features was assigned through the channel attention module to weaken the influence of redundant features on the recognition results. Then It was input into the CPRN model built by stacking convolutional layers, residual blocks and capsule layers, and an affine transformation matrix with weights sharing was adopted to replace the fully connected capsule layers to reduce the number of parameters, improve the training speed, and finally output the diagnostic results. CWRU bearing data set and test bench simulation data were used for experiments. Results show that the average diagnostic accuracy of CA-CPRN under variable working conditions reaches 97.63% and 98.23% respectively, indicating better generalization ability compared with other models. In addition, the diagnostic accuracy on both datasets is better than that of other models in the case of noise, and the average accuracy reaches 99.09% and 96.32% respectively, which proves the superiority of the model in noise resistance.

Characteristics such as intermittency and volatility of renewable energy pose challenges to grid scheduling. Liquid air energy storage system is one of the effective technical measures to solve this problem, not only in terms of large scale and long storage time, but also in terms of high energy storage density and not limited by geographical environment. Firstly, the principles of five classical air liquefaction cycle technologies were introduced, and the characteristics of different systems in terms of air liquefaction were analyzed. Secondly, the improved technologies of two air liquefaction cycles were analyzed, and comparisons were conducted on the advantages and shortcomings of different air liquefaction technologies in terms of liquefaction capacity and economy. It is concluded that Linde-Hampson cycle and Claude cycle should be adopted based on the comprehensive consideration of liquefaction rate, economic cost and safety. Finally, the problems of existing air liquefaction technologies in terms of liquefaction rate and cooling capacity gap have been analyzed, and the future development trend of air liquefaction technologies for liquid air energy storage systems has been discussed.

A study was conducted on the modification of SGT5-2000E gas turbine by blending natural gas fuel with hydrogen rich synthesis gas. The blending ability of hydrogen-rich synthesis gas in this type of gas turbine was analyzed, and modification measures were proposed to address technical issues caused by H₂ and CO in hydrogen-rich synthesis gas. The performance parameters of the gas turbine unit before and after blending were compared. Results show that this type of gas turbine can mix up to volume fraction of 50% hydrogen rich synthesis gas. When blending 50% hydrogen rich synthesis gas, the performance parameters of the gas turbine meet the design requirements, and the relevant modifications and improvement measures have achieved the expected results, improving the operational flexibility of this type of gas turbine unit and providing reference for similar blending research.

Steam-water film interfacial energy and mass transfer mechanism of direct contact condensation is the basis for the optimization design of mixing condensers. Taking steam-water film direct contact condensation as the research object, a film interfacial condensation energy and mass transfer model was established based on the volume of fluid (VOF) model. The impact of varying steam and water mass flow rate ranging in 0.002-0.010 kg/s and 0.30-0.83 kg/s respectively, as well as the influence of flow pattern, on the heat exchange of steam-water film direct contact condensation were analyzed. Results show that the heat transfer coefficient increases with the water and steam mass flow rate. The steam mass flow rate has little influence than the water mass flow rate on direct contact condensation. Compared with the parallel flow,cross flow proves to be more favorable for the heat transfer process. Due to the increased interphase force and irregular shape of the water film, the risk of water film breakage increases in the countercurrent flow condition.

The flexibility enhancement will significantly affect the thermal comfort and operation economy of combined heat and power (CHP) units. In order to optimize the comprehensive performances of the unit in the process of rapid load change in a large range, a multi-objective cooperative control strategy based on the predictive control was proposed. Firstly, a multi-objective evaluation index system of the unit was established, which considered flexibility, thermal comfort, economy, and stationarity. In order to achieve multi-objective coordination, a power-heat-coordination control strategy based on the predictive control was proposed. The coal consumption index was introduced into the rolling optimization to achieve economic optimization, and the solution method of multi-objective control rates was given. Then, with the multi-objective evaluation index as the objective function of the particle swarm optimization, the optimal parameter setting of the control strategy was obtained based on the offline optimization of the simulation model, and the comprehensive performance of the unit was optimized. Finally, the simulation test was conducted on a reference 300 MW unit. Results show that the proposed strategy can enhance the operational flexibility of the combined heat and power unit, and ensure the thermal comfort and operation economy of the unit to a considerable extent, which provides a basis for the multi-objective optimization control of the combined heat and power unit.

In order to analyze the flow field, flame structure and emission patterns in the secondary zone of axial staged combustion of methane and ammonia fuels, a high-frequency particle image velocimetry (PIV)analysis and OH^*-based self-luminous techniques were used to investigate the secondary flow field structure, reburning flame morphology and pollutant emission characteristics of the premixed secondary burner with a jet angle of 135° at a momentum flux ratio of 6 using different types of fuels under atmospheric pressure conditions. Results show that when the jet angle is 135°, the secondary flow field will form a large low-speed recirculation zone downstream of the jet outlet, which is conducive to stable combustion of fuels with slower flame propagation speed. When methane mixed with ammonia is used as the secondary fuel, the position of the reburning flame shifts upstream, resulting in a longer high temperature chemical residence time. At the beginning of ammonia-doped combustion, when the ammonia-doped ratio increases, the jet flame transforms from a continuous flame to a disintegrated flame. The reaction zone of the ammonia jet flame is elongated, leaving a heavy sufficient reaction distance to reduce emissions. Therefore, the emission gradually decreases in the detached flame form.

In order to improve the efficiency and accuracy of gear fault identification under variable speed conditions, an intelligent fault identification method of variable speed gear, namely INGO-CSA-LSTMN, was proposed based on improved northern goshawk optimization (INGO) algorithm to optimize convolutional self-attention long short-term memory network (CSA-LSTMN). In response to the problems that the traditional northern goshawk optimization algorithm taking too long training time and being easy to fall into local optimization, an INGO algorithm was proposed by introducing sinusoidal pulse modulated chaotic map and random Levy flight strategy, and it was applied to optimize the key parameters of the CSA-LSTMN model to improve the stability and training efficiency of the model. The validation through test functions demonstrates that the INGO algorithm has a faster convergence speed and can find the optimal solution more accurately. The analysis of gear datasets from two different experimental rigs reveals that compared to other commonly used network models, the INGO-CSA-LSTMN model has higher recognition accuracy for gear faults under different operating conditions, with the accuracy rates exceeding 99.9%.