While model predictive control (MPC) has been extensively studied for energy-efficient indoor thermal environment control in small to medium-scale buildings, existing MPC approaches often prioritize temperature-only optimization, neglecting humidity dynamics and latent heat transfer, which limits energy efficiency and thermal comfort in humid climates. To address this gap, an energy-efficient nonlinear MPC framework for direct expansion (DX) air conditioning systems to simultaneously regulate indoor temperature and humidity is developed. A coupled building thermal-moisture dynamic model with empirical models of the DX system is established to facilitate the design of the nonlinear MPC. The framework employs a second-order moisture model to accurately capture moisture exchange dynamics, moving beyond simplistic insulated-zone assumptions prevalent in prior work. The performances of the proposed MPC for the DX system with different control settings and weather conditions are examined. Results demonstrate that the controller maintains the predicted mean vote (PMV) within the optimal comfort range (−0.2 to +0.2) while achieving significant energy savings. Up to 12.3

Accurate building energy consumption prediction is crucial for optimizing energy efficiency and operational flexibility in modern buildings. However, existing prediction methods struggle to handle the highly complex, nonlinear, and multifactor-influenced nature of building energy data, often failing to fully exploit the intrinsic value and underlying characteristics of varying data components. To address these limitations, this study aims to develop a novel hybrid deep-learning framework that enhances multi-step prediction accuracy by decoupling and independently modeling the underlying patterns of energy time-series data. Methodologically, the proposed approach integrates STL with a divide-and-conquer predictive strategy. First, the original energy consumption series is decomposed into three interpretable components: trend, seasonality, and residual. Next, the random forest algorithm is applied to identify the most relevant input variables from historical energy, calendar, and meteorological data for each specific component. Subsequently, tailored deep-learning models are assigned based on component characteristics: LSTM networks model the long-term trend, the Transformer architecture captures the periodic seasonal patterns, and a Res-LSTM tackles the highly stochastic residual component. The proposed framework is validated using real-world data from the public BDG2 dataset. Experimental results demonstrate that the STL-based hybrid model significantly outperforms single prediction models (including BPNN, SVR, RNN, LSTM, transformer, and Res-LSTM). Specifically, compared to the best-performing single model, the proposed method yields improvements of 20.1%, 12.7%, and 23.9% in mean absolute error (MAE), root mean square error (RMSE), and coefficient of variation of RMSE (CV-RMSE), respectively. With achieved final values of 0.0397 kW for RMSE and 0.94 for the coefficient of determination (R 2), this study concludes that the proposed hybrid methodology effectively mitigates noise and non-stationarity in energy data. It provides a highly accurate, robust, and interpretable tool to support intelligent building energy management and system scheduling.

The thermal management system (TMS) is a critical factor in the development of electric vehicles (EVs). Among various TMS configurations, the direct cooling system has attracted significant research interest due to its compact structure and superior cooling performance. In this study, a simulation model of a typical EV direct cooling TMS was established. To address the control challenges arising from its multiple dynamic characteristics and strong coupling, a simulation-based investigation was conducted using Weight-Based Fuzzy Logic Control (WBFLC) as the basic control algorithm, and two decoupling control strategies (Control Strategy I & II) were proposed. Under dual-cooling conditions, Control Strategy I demonstrated better performance, effectively stabilizing both battery and cabin temperatures near their respective target values (25 degrees C and 24 degrees C in simulations) under both constant-current and variable-current (UDDS and US06 cycles) conditions. The maximum temperature deviation was 0.2 degrees C for the battery and did not exceed 0.15 degrees C for the cabin. Finally, experimental validation of Control Strategy I was performed on an existing test bench. The results indicated that both battery and cabin temperatures could be stably maintained near their target values (20 degrees C for the battery and 24 degrees C for the cabin), with a maximum deviation of no more than 0.25 degrees C. This research presents an effective approach to the decoupling control of direct cooling thermal management systems in electric vehicles.

Ventilation fans are widely used in industry factories and commercial buildings. After experiencing a long-time operation, fans are prone to abnormalities resulting in system performance degradation, energy waste, and even safety issues. Recent works have shown that the machine learning-based techniques outperform most of the traditional vibration signal-based diagnostic method. However, the insufficient number of fault training samples has become the main obstacle for the supervised fault diagnosis. Therefore, in order to improve the fault diagnosis performance with imbalanced training dataset, this paper reports a multi-head self-attention enhanced semi-supervised generative adversarial network (MSA-SGAN) method for ventilation fans. The original unsupervised GAN was improved to a semi-supervised GAN (SGAN) and thus the ability of multi-classification could be achieved. In addition, the SGAN was enhanced by integrating multi-head self-attention (MSA), which allows for increased emphasis on relevant and significant features. An experimental system was established, and different types of fan fault were simulated. Using the experimental data, the fault diagnosis model based on the proposed MSA-SGAN was trained and validated. Results showed that the proposed fault diagnosis method exhibited an excellent performance including overall accuracy, recall, and precision as compared to the other traditional methods. In the case of the imbalanced dataset, the proposed method shows superior performance compared to other traditional supervised and semi-supervised methods.

Indoor thermal comfort and energy efficiency in direct expansion (DX) air conditioning (A/C) systems are critical yet often conflicting objectives. Conventional controllers prioritize temperature and humidity regulation without optimizing energy use. This study addresses this gap by proposing a model predictive control (MPC) strategy to balance thermal comfort and energy efficiency for a single DX A/C system. A hybrid modeling approach is developed to facilitate the design of MPC through integrating a white-box model for the DX cooling coil to capture the cooling and dehumidification characteristics and a gray-box model for the air-conditioned room to predict its thermal dynamics. Using the developed hybrid system model, two MPC schemes are designed: one targeting temperature regulation alone and another incorporating both temperature and humidity in the objective function. Validation demonstrates that the temperature-only MPC can accurately regulate the temperature but resulting in an undesirable humidity level and 28.6 %-36.8 % higher energy consumption. The inclusion of both temperature and humidity significantly reduces energy use while maintaining thermal comfort within a tight predicted mean vote (PMV) range of -0.2 to +0.2. The results highlight that simultaneous temperature-humidity optimization enhances energy efficiency and comfort, resolving the trade-off inherent in traditional systems. The key contribution lies in the novel hybrid modeling framework in MPC design and the demonstration of the superiority of the MPC for temperature-humidity regulation over the temperature-only MPC. This research advances DX A/C system control by providing a scalable, energy-conscious solution for sustainable building operational management.

Achieving simultaneous control over temperature and humidity through a single direct expansion air conditioner would be advantageous, especially considering its widespread use in residential buildings. Traditional control strategies fail to achieve satisfactory control performance due to the nature of complexity and cross-decoupling of the system. Therefore, this research paper presents the advancement of a feedforward decoupling control scheme designed for a direct expansion air conditioning unit to effectively control both temperature and humidity. To facilitate the controller design, a dynamic model considering both sensible and latent heat transfers was developed and verified. The thermal responses of the coupled system were analyzed and their correlation coefficients also examined. Through identifying the transfer function model of the coupled system, a feedforward decoupling controller with two compensators was designed. The proposed decoupling control scheme was tested using the developed model as the plant to be controlled. Test results showed that, in comparison to the transitional PID based dual single-input and single-output controller, the proposed decoupling control scheme was capable of eliminating the interactions between indoor air temperature and humidity under variable speed operation, and realizing the simultaneous control over indoor air temperature and humidity with a desirable control performance.

Accurate measurement of air humidity serves as the foundation for the air conditioning system to achieve and maintain optimal indoor thermal environments. The current humidity sensors, due to their complexity, high cost, and susceptibility to contamination, are deemed unsuitable for humidity testing in the practical deployment of air conditioning systems. This underscores the need for the evolution of novel humidity measurement techniques. This study presented the development of an indirect method for measuring air relative humidity by evaluating certain readily available parameters. The underlying principle is the high degree of coupling in the heat and mass transfer occurring on the evaporator. This coupling facilitates the derivation of the relative humidity of the air at the air conditioning system's inlet from parameters such as temperature and air volume. In an effort to swiftly and accurately derive the relative humidity of the air from numerous parameters, this study employed the artificial neural networks (ANN) methodology and constructed an ANN model. Upon comparison with experimental data, it was discerned that the maximum absolute error in all prediction results was below 4.5% RH, with a substantial proportion falling below 2.5% RH. Further research corroborated that the inclusion of the degree of refrigerant superheat as a training parameter for the model exerted negligible influence on prediction accuracy. The prediction results yielded RMSE values of 0.72% and 0.71% respectively. These findings suggest that the proposed method exhibits a high degree of accuracy, thereby demonstrating its potential applicability in the field testing of air conditioning systems.

Centrifugal chillers have been widely used in medium- and large-scale air conditioning projects. However, equipment running with faults will result in additional energy consumption. Meanwhile, it is difficult to diagnose the minor faults of the equipment. Therefore, the Extreme Gradient Boost (XGBoost) algorithm was used to solve the above problem in this article. The ASHRAE RP-1043 dataset was employed for research, utilizing the feature splitting principle of XGBoost to reduce the data dimension to 23 dimensions. Subsequently, the five important parameters of the XGBoost algorithm were optimized using Multi-swarm Cooperative Particle Swarm Optimization (MSPSO). The minor fault diagnosis model, MSPSO-XGBoost, was established. The results show that the ability of the proposed MSPSO-XGBoost model to diagnose eight different states is uniform, and the diagnostic accuracy of the model reaches 99.67%. The accuracy rate is significantly improved compared to that of the support vector machine (SVM) and back propagation neural network (BPNN) diagnostic models.

针对轴流通风机故障诊断中需要大量带标签数据用于分类模型训练的问题,提出基于小波包分解(Wavelet packet decomposition,WPD)与半监督生成对抗网络(Semi-supervised generative adverserial networks,SGAN)的轴流通风机故障诊断方法.首先对预处理后的通风机振动数据进行小波包分解,将提取到的有效频带能量信息作为故障诊断模型的特征输入;其次利用训练数据中的带标签数据与无标签数据,训练 SGAN 的生成器和鉴别器,将训练后的鉴别器作为分类器用于实现少量带标签数据下的轴流通风机故障诊断.搭建了轴流通风机故障诊断试验台,采集了包括正常运行、基座松动与 4种不同程度转子不平衡的 6类状况下通风机振动数据.利用数据训练得到了基于WPD-SGAN的通风机故障诊断模型.故障诊断实验结果显示,在少标签样本情况下,该方法的诊断准确率达到 80%以上.相比传统支持向量机与神经网络监督学习方法,该方法的准确率有大幅提升;与半监督支持向量机方法相比,该方法的准确率提高了 9～14个百分点.

Compressed air is extensively used in manufacturing industries due to its cleanliness, practicality and ease of use, and thus the energy consumed by compressed air systems accounts for a large share of industry electricity. Energy efficient control for compressed air systems will contribute to energy saving. Through modeling the compressed air system equipped with fixed speed compressor, and that with variable speed compressor, their corresponding operating characteristics were examined. Considering the advantages of variable speed compressors at the part-load condition, the variable speed driver technique was introduced to a conventional compressed air system with multi-fixed speed compressor. For balancing the investment cost and operation cost, only one compressor was variable speed controlled and the remaining compressors were load/unload controlled. 14.4% of energy could be saved as compared to the conventional compressed air system with multi-fixed speed compressor under part load condition. Furthermore, by introducing a feedforward loop to compensate for the change of compressed air demand, the proposed control scheme for the compressed air system was improved. Results showed that a better operating performance in terms of an extra 4.1% of energy saving and a smaller pressure fluctuation was achieved under the proposed control scheme. The current study will contribute to energy saving for the compressed air systems with multi-compressor.

较高精度的空调负荷模型是开发实施有效空调控制策略的重要依据,其有利于促进减小电力能源消耗以节约用电成本.首先,通过对建筑构造、室内外环境和气象因素等影响分析,搭建可用于预测空调负荷的灰箱模型,即三阶的等效热参数模型以及二阶的等效湿阻模型;接着,通过最小化模型输出室内温湿度数据与室内实测温湿度采样数据之间的误差建立优化目标函数;然后,提出并使用基于粒子群优化算法的参数辨识方法获取灰箱模型关键参数.实验研究表明,辨识得到的等效热阻和湿阻模型能准确地反映室内温湿度分布和变化特性,具有预测空调负荷的实际应用价值.

Heat exchangers (HX) are often utilized in industry, and the optimization of the performance of HX is a key area of research. In this study, EVAP-COND software 4.0 and genetic algorithm (GA) based optimization methods were proposed to optimize the circuitry and fin pitch of a finned tube heat exchanger for an air conditioner. A simulation model for a multi-circuit finned-tube evaporator used in an air conditioning unit was developed using the EVAP-COND software, and further validated based on the experimental data. Considering the refrigerant flow maldistribution of the original HX, four different circuit arrangements, i.e., types A, B, C, and D, were designed and optimized circuitry obtained. Based on both simulation and experimental results, D-type HX with 1.8 mm fin pitch was selected as 10% tubes could be saved with no significant loss of heat transfer capacity. Then the fin pitch was further optimized using the multi-objective GA method, with both Colburn factor j and friction factor f being considered. Optimization results showed that, in Pareto front, points 1 to 4 showed the increase in the Colburn factor j was negative, while the decrease in the friction factor f was positive. The friction factor decreased by 3.5% as one moved from Point 1 to Point 4, but the Colburn factor rose by 1.02%. Points 5 to 10 demonstrated that, while the decrease in the friction factor was negative, the increase in the Colburn factor was positive. The friction factor decreased by 5.31%, but the Colburn factor increased by 1.51% when going from Point 5 to Point 10. The results of optimization demonstrated that the objective function performed at its optimum when the fin pitch was around 1.77 mm.

In order to solve the problem that the traditional AGV path planning algorithm is difficult to deal with the complex and dynamic environment and some reinforcement learning algorithms have discrete actions and sparse rewards, this paper proposes an AGV path planning method based on the PPO (Proximal Policy Optimization) algorithm. This method firstly establishes an action decision-making system based on normal distribution through the output mean and variance of the neural network, which solves the problem of discrete output actions. Then, the method adds auxiliary rewards to the main line rewards to effectively alleviate the sparse rewards. Finally, a path planning method based on this algorithm is designed. In order to verify the performance of the proposed method, the method is compared and analyzed with the DQN algorithm in three simulation environments of Gazebo. The results show that the convergence speed and stability of the method based on the PPO algorithm are better than the DQN algorithm. The method based on the PPO algorithm has obvious advantages in the task completion time.

In this study, grid refinement strategies are studied to improve computing accuracy and efficiency of the complex optimal control problems. Firstly, a solution method based on multiple variable time nodes was proposed for the optimal control problem with Bang-Bang characteristics. The problem was transformed into the Mayer formula by adding relaxed variables to eliminate integral terms. And the implementation procedure for the algorithm was also defined. Then, for optimal control problems with continuous slope changes and key switching points, a slope-based adaptive grid refinement strategy was also proposed. The time grids were able to automatically increase or decrease based on the slope information, and a new technique was used to judge the switching points and refine the corresponding time grids. Finally, the performances of the proposed methods were verified with six classical examples. The results demonstrate that they comprehensively surpass traditional methods in computing accuracy and efficiency.

Ice slurry is a solid-liquid phase fluid consisting of a liquid solution and ice parti-cles. It is widely used in life and engineering because of its excellent cold-carrying capacity. In this paper, a genetic algorithm is used to optimize the ice slurry flow with the minimum pumping power as the objective function. The results show that the genetic algorithm can be effectively applied to the optimization of ice slur-ry flow characteristics within reasonable parameters. In addition, the transport characteristics of ice slurry are also analyzed. The selection of suitable ice mass fraction values under different working conditions can make the transport charac-teristics optimal.

Based on the mathematical modeling and operational optimization studies of reverse osmosis (RO) and multistage flash (MSF) desalination, the structural optimization of the hybrid process was specially studied in this work with the consideration of reducing comprehensive expenses under given operational conditions. Firstly, according to the process mechanism and flowchart of the RO and MSF seawater desalination technologies, seven hybrid structures with different feed conditions were designed, and their connection equations were established for numerical calculation. Then, in order to evaluate the economic performance of the hybrid systems with different structures, the hourly average operational cost equations of RO and MSF processes were established and formulated as the comprehensive evaluation indicators. Next, with a given water production requirement, simulation calculations of the hybrid system with seven different structures were performed. The results show that the hybrid system with the fourth structure has the lowest operational cost of 4.6834 CNY/m3, and at the same time it has the lowest blowdown. However, if we take GOR or production water temperature as the target, the optimal structure of the hybrid system is the fifth or the seventh option. The obtained results are helpful in structural optimization of the hybrid system with aspects of operational cost reduction, maximum GOR, or minimizing the wastewater discharge.

Focusing on the problems of opaqueness and high energy consumption in coal-fired power plant wastewater recycling processes, this paper studies the simulation and operational optimization of coal-fired power plant wastewater treatment by taking a coal-fired power plant system in Inner Mongolia as an example. Firstly, based on the solution–diffusion theory, pressure drop, and osmotic concentration polarization, a mechanistic model equation for coal-fired power plant wastewater treatment is developed. Secondly, the equation fitness and equation parameters are calibrated to obtain an accurate model. Thirdly, the system is simulated and analyzed so as to obtain the influence and change trajectories of different feed flowrates, temperatures, pressures, and concentrations on various aspects of the system’s performance, such as water recovery rate, salt rejection rate, and so on. Finally, in order to reduce the operating cost of the system, an optimization analysis is carried out, with the lowest specific energy consumption and average daily operating cost as optimization goals, and the performance changes of the system before and after optimization under three different working conditions are compared. The results show that adopting the given optimal strategy can significantly reduce the system’s operational cost. This research is helpful for the digitization and low-carbon operation of coal-fired power plant wastewater treatment systems.

冷水机组故障状态下故障样本数据具有非高斯的特性,传统的线性降维方法难以应用于其故障数据的特征提取.针对此问题,综合均匀流形逼近和投影(UMAP)方法适用于高维非高斯数据的特性与树突网络(DD)对大样本数据的良好分类能力,并使用自适应矩估计(Adam)在DD权重更新过程中替代其原先的梯度下降法,提出了一种基于UMAP-AdamDD的冷水机组故障诊断方法.将所提方法的诊断结果与其它常见非线性降维方法和分类器下的诊断结果进行比较,UMAP-AdamDD方法对冷水机组常见故障的诊断效果显著,综合故障诊断结果大于96％.

Incipient fault detection and diagnosis for centrifugal chillers is significant for maintaining safe and effective system operation. Due to the advantages of simple learning algorithm and high generalization capability, the extreme learning machine (ELM) can identify faults quickly and precisely in comparison to conventional classification methods such as back propagation neural network (BPNN). This paper reports an effective diagnosis method for incipient chiller faults with the integration of kernel entropy component analysis (KECA) and voting based ELM (VELM). KECA was first performed to reduce the dimensionality of the original input data so as to minimize the model complexity and computational cost. Instead of using a single ELM, multiple independent ELMs were adopted in VELM, and then the class label could be predicted based on the majority voting method. Using the experimental data of seven typical faults together with a normal operation, the proposed KECA-VELM fault diagnostic model was trained and further validated. The results show that a better fault diagnosis performance can be achieved using the KECA-VELM classifier compared with the conventional BPNN, ELM and VELM based classifiers. The overall average fault diagnosis accuracy for the faults at the least severity level was reported over 95% based on the proposed method.

Temperature and humidity are two important factors that influence both indoor thermal comfort and air quality. Through varying compressor and supply fan speeds of a direct expansion (DX) air conditioning (A/C) unit, the air temperature and humidity in the conditioned space can be regulated simultaneously. However, most existing controllers are designed to minimize the tracking errors between the system outputs with their corresponding settings as quickly as possible. The energy consumption, which is directly influenced by the compressor and supply fan speeds, is not considered in the relevant controller formulations, and thus the system may not operate with the highest possible energy efficiency. To effectively control temperature and humidity while minimizing the system energy consumption, a model predictive control (MPC) strategy was developed for a DX A/C system, and the development results are presented in this paper. A physically-based dynamic model for the DX A/C system with both sensible and latent heat transfers being considered was established and validated by experiments. To facilitate the design of MPC, the physical model was further linearized. The MPC scheme was then developed by formulating the objective function which sought to minimize the tracking errors of temperature and moisture content while saving energy consumption. Based on the results of command following and disturbance rejection tests, the proposed MPC scheme was capable of controlling temperature and humidity with adequate control accuracy and sensitivity. In comparison to linear-quadratic-Gaussian (LQG) controller, better control accuracy and lower energy consumption could be realized when using the proposed MPC strategy to simultaneously control temperature and humidity.