基于多通道的ICU脑血管疾病死亡风险预测模型

死亡风险预测指根据病人临床体征监测数据来预测未来一段时间的死亡风险。对于ICU病患，通过死亡风险预测可以有针对性地对病人做出临床诊断，以及合理安排有限的医疗资源。基于临床使用的MEWS和Glasgow昏迷评分量表，针对ICU病人临床监测的17项生理参数，提出一种基于多通道的ICU脑血管疾病死亡风险预测模型。引入多通道概念应用于BiLSTM模型，用于突出每个生理参数对死亡风险预测的作用。采用Attention机制用于提高模型预测精度。实验数据来自MIMIC [Ⅲ]数据库，从中提取3?080位脑血管疾病患者的16?260条记录用于此次研究，除了六组超参数实验之外，将所提模型与LSTM、Multichannel-BiLSTM、逻辑回归（logistic regression）和支持向量机（support vector machine, SVM）四种模型进行了对比分析，准确率Accuracy、灵敏度Sensitive、特异性Specificity、AUC-ROC和AUC-PRC作为评价指标，实验结果表明，所提模型性能优于其他模型，AUC值达到94.3%。     

Operational time-series data modeling via LSTM network integrating principal component analysis based on human experience

Today’s information technologies involve increasingly intelligent systems, which come at the cost of increasingly complex equipment. Modern monitoring systems collect multi-measuring-point and long-term data which make equipment health prediction a “big data” problem. It is difficult to extract information from such condition monitoring data to accurately estimate or predict health statuses. Deep learning is a powerful tool for big data processing that is widely utilized in image and speech recognition applications, and can also provide effective predictions in industrial processes. This paper proposes the Long Short-term Memory Integrating Principal Component Analysis based on Human Experience (HEPCA-LSTM), which uses operational time-series data for equipment health prognostics. Principal component analysis based on human experience is first conducted to extract condition parameters from the condition monitoring system. The long short-term memory (LSTM) framework is then constructed to predict the target status. Finally, a dynamic update of the prediction model with incoming data is performed at a certain interval to prevent any model misalignment caused by the drifting of relevant variables. The proposed model is validated on a practical case and found to outperform other prediction methods. It utilizes a powerful deep learning analysis method, the LSTM, to fully process big condition monitoring series data; it effectively extracts the features involved with human experience and takes dynamic updates into consideration.     

Deformation-Strength Relationships in Short-Term Tension of Heat-Resistant Materials

The analysis of 124 curves obtained in short-term tensile tests demonstrate that they can be described by varying strain hardening and softening characteristics. Different stress–strain curves can be produced at invariable yield strength and ultimate strength and interrelated proportional variations of the above characteristics. To determine some specific stress–strain curve, it is necessary to take account of yield strength and ultimate strength as well as strain corresponding to the latter. The relations between yield strength, ultimate strength and hardening and their practically complete absence between these parameters and softening were statistically established.     

Treatment of depressive symptoms during short-term rehabilitation: An attempted replication of the DOUR project.

Objective: To replicate and extend P. A. Lichtenberg and colleagues' (1996) cross-disciplinary intervention to improve physical and mental health among older adults. Participants: 14 depressed older adults (6 treatment, 8 control). Setting: The short-term rehabilitation unit of an urban nursing home. Intervention: Occupational therapists were trained to treat depression using pleasant events and cognitive-behavioral therapies. Outcome Measures: Geriatric Depression Scale, the Short Form-12, and the Multi-Level Assessment Instrument: Activities of Daily Living. Results: No significant group differences were found in physical or mental health. However, more control group members (75%) than treatment group members (33%) were depressed at study completion. Conclusions: The treatment of depressive symptoms can be integrated with a nonmental health treatment modality. (PsycINFO Database Record (c) 2011 APA, all rights reserved)     

The Benefits of a "Sidelong" Approach to Self-Esteem Need Satisfaction: Comment on Crocker and Park (2004).

J. Crocker and L. E. Park (2004) have achieved an admirable integration of the self-esteem literature with their claim that self-esteem is better conceived of as a dynamic human striving, rather than as a passive state or personality characteristic. However, the costs of self-esteem striving may be overstated--these costs may arise only in certain constrained cases. Also, although Crocker and Park suggested that self-esteem is not a true psychological need, there is evidence that humans in all cultures need to feel a positive sense of self-worth (K. M. Sheldon, in press). Problems may arise only when people strive too directly for this feeling, rather than deriving it as a natural concomitant of non-self-focused goals. A "sidelong" approach to self-esteem need satisfaction is advocated in this commentary. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Discrete-time adaptive sliding mode power system stabilizer with only input/output measurements

In this paper a discrete-time adaptive sliding mode control method is newly developed and applied to the power system stabilization problem. A controllable canonical form of state space realization is constructed using the parameters identified by the on-line recursive least squares method and the system state is estimated from the input/output measurements and the simple state transformations. The identified parameters and the estimated state are then used by the discrete-time sliding mode control, which is suitable for the digital equipment. The most important advantage of the proposed power system stabilizer (PSS) is that it is able to maintain its regulating performance with a slower sampling period than that of the conventional sliding mode PSS because it is developed in a pure discrete-time domain. Another advantage of the proposed PSS is that it needs neither a mathematical model of the power system nor the full-state measurements because they are identified through on-line identifications. Several computer simulations for the linear power system are performed to verify the performance of the proposed PSS. In the computer simulations for various circumstances which are probable in a power system are considered, such as transitions of the active and reactive powers, change of parameters of the synchronous machine, line-to-ground faults and measurement noise. As a result, a new power system stabilizer which can operate in a wide range of operating conditions and can overcome various disturbances and measurement noises is proposed.     

Intelligent tool wear monitoring based on parallel residual and stacked bidirectional long short-term memory network

Effective tool wear monitoring (TWM) is essential for accurately assessing the degree of tool wear and for timely preventive maintenance. Existing data-driven monitoring methods mainly rely on complex feature engineering, which reduces the monitoring efficiency. This paper proposes a novel TWM model based on a parallel residual and stacked bidirectional long short-term memory (PRes–SBiLSTM) network. First, a parallel residual network (PResNet) is used to extract the multi-scale local features of sensor signals adaptively. Subsequently, a stacked bidirectional long short-term memory (SBiLSTM) network is used to obtain the time-series features related to the tool wear characteristics. Finally, the predicted tool wear value is outputted through a fully connected network. A smoothing correction method is applied to improve the prediction accuracy. The proposed model is experimentally verified to have a high prediction accuracy without sacrificing its generalization ability. A TWM system framework based on the PRes–SBiLSTM network is proposed, which has a certain reference value for TWM in actual industrial environments.     

Machine-Learning based methods in short-term load forecasting

Short-term load forecasting is of great significance to the secure and efficient operation of power systems. However, loads can be affected by a variety of external impact factors and thus involve high levels of uncertainties. So it is a challenging task to achieve an accurate load forecast. This paper discusses three commonly-used machine-learning methods used for load forecasting, i.e., the support vector machine method, the random forest regression method, and the long short-term memory neural network method. The features and applications of these methods are analyzed and compared. By integrating the advantages of these methods, a fusion forecasting approach and a data preprocessing technique are proposed for improving the forecasting accuracy. A comparative study based on real load data is performed to verify that the proposed approach is capable of achieving a relatively higher forecasting accuracy.     

城市社区尺度降雨径流快速模拟——以福州市一排水小区为例

随着城市化进程的不断加快,我国城市正面临着越来越严峻的洪涝问题。本文在社区尺度上构建雨洪管理模型（SWMM）,使用遗传算法率定SWMM模型参数;在对研究区降雨分析的基础上,采用模糊识别法筛选出最具代表性的两种雨型;基于不同的降雨情景与SWMM模拟值组成的数据集,建立长短期记忆神经网络（LSTM）模型模拟研究区降雨径流关系,并使用不同工况评估了LSTM模型效果。结果表明,LSTM模型对降雨径流的模拟与SWMM模型基本吻合,而其对洪峰流量的拟合略有偏差。在较小降雨下,LSTM模型模拟洪峰流量较SWMM输出结果偏小;在较大降雨下,模拟结果偏大;在中等降雨时,模拟效果最好。此外,50个隐含层单元的拟合效果更好,但同时更多的隐含层单元对洪峰流量拟合效果更好。     

基于多模型融合的CNN-LSTM-XGBoost短期电力负荷预测方法

短期电力负荷的精准预测可以有效指导机组组合调度、经济调度与电力市场运营。针对输入数据特征量受限时负荷预测的低精度问题,提出一种基于多模型融合的CNN-LSTM-XGBoost短期电力负荷预测方法。通过建立融合局部特征预提取模块的LSTM（long short term memory）网络结构,并将其与XGBoost（eXtreme boosting system）预测模型并行结合,之后结合MAPE-RW（mean absolute percentage error-reciprocal weight）算法进行模型融合初始权重设置,对最佳权重进行搜索,构建最佳融合模型。通过运用电力负荷数据对所提方法进行预测实验,结果表明CNN-LSTM- XGBoost模型的MAPE（mean absolute percentage error）与RMSE（root mean square error）分别为0.377%与148.419 MW,相比于单一网络模型与融合模型结构实现了误差指标的显著降低,验证了基于多模型融合的CNN-LSTM-XGBoost短期电力负荷预测方法具有较快的模型训练速度、较高的预测准确度与较低的预测误差。