Data-driven Prediction Method Research Articles

Data-driven predictive control for continuous-time linear parameter varying systems with application to wind turbine

A new data-driven predictive control method based on subspace identification for continuous-time linear parameter varying (LPV) systems is presented in this paper. It is developed by reformulating the continuous-time LPV system which utilizes Laguerre filters to obtain the subspace prediction of output. The subspace predictors are derived by QR decomposition of input-output and Laguerre matrices obtained by input-output data. The predictors are then applied to design the model predictive controller. It is shown that the integrated action is incorporated in the control effect to eliminate the steady-state offset. We control the continuous-time LPV systems to obtain the attractive performance with the proposed data-driven predictive control method. The proposed controller is applied to a wind turbine to verify its effectiveness and feasibility.

DCA-Based Real-Time Residual Useful Life Prediction for Critical Faulty Component

Residual useful life (RUL) prediction is significant for condition-based maintenance. Traditional data-driven RUL prediction method can only predict fault trend of the system rather than RUL of a specific system component. Thus it cannot tell the operator which component should be maintained. The innovation of this paper is as follows: (1) Wavelet filtering based method is developed for early detection of slowly varying fault. (2) Designated component analysis is introduced as a feature extraction tool to define the fault precursor of a specific component. (3) Exponential life prediction model is established by nonlinear fitting of the historical RUL and the fault size characterized by the statistics used. Once online detection statistics is obtained, real-time RUL of the critical component can be predicted online. Simulation shows the effectiveness of this algorithm.

Predictive modelling of gold potential with the integration of multisource information based on random forest: a case study on the Rodalquilar area, Southern Spain

Mineral exploration activities require robust predictive models that result in accurate mapping of the probability that mineral deposits can be found at a certain location. Random forest (RF) is a powerful machine data-driven predictive method that is unknown in mineral potential mapping. In this paper, performance of RF regression for the likelihood of gold deposits in the Rodalquilar mining district is explored. The RF model was developed using a comprehensive exploration GIS database composed of: gravimetric and magnetic survey, a lithogeochemical survey of 59 elements, lithology and fracture maps, a Landsat 5 Thematic Mapper image and gold occurrence locations. The results of this study indicate that the use of RF for the integration of large multisource data sets used in mineral exploration and for prediction of mineral deposit occurrences offers several advantages over existing methods. Key advantages of RF include: (1) the simplicity of parameter setting; (2) an internal unbiased estimate of the prediction error; (3) the ability to handle complex data of different statistical distributions, responding to nonlinear relationships between variables; (4) the capability to use categorical predictors; and (5) the capability to determine variable importance. Additionally, variables that RF identified as most important coincide with well-known geologic expectations. To validate and assess the effectiveness of the RF method, gold prospectivity maps are also prepared using the logistic regression (LR) method. Statistical measures of map quality indicate that the RF method performs better than LR, with mean square errors equal to 0.12 and 0.19, respectively. The efficiency of RF is also better, achieving an optimum success rate when half of the area predicted by LR is considered.

Data-Driven High-Throughput Prediction of the 3-D Structure of Small Molecules: Review and Progress

Accurate prediction of the 3-D structure of small molecules is essential in order to understand their physical, chemical, and biological properties, including how they interact with other molecules. Here, we survey the field of high-throughput methods for 3-D structure prediction and set up new target specifications for the next generation of methods. We then introduce COSMOS, a novel data-driven prediction method that utilizes libraries of fragment and torsion angle parameters. We illustrate COSMOS using parameters extracted from the Cambridge Structural Database (CSD) by analyzing their distribution and then evaluating the system's performance in terms of speed, coverage, and accuracy. Results show that COSMOS represents a significant improvement when compared to state-of-the-art prediction methods, particularly in terms of coverage of complex molecular structures, including metal-organics. COSMOS can predict structures for 96.4% of the molecules in the CSD (99.6% organic, 94.6% metal-organic), whereas the widely used commercial method CORINA predicts structures for 68.5% (98.5% organic, 51.6% metal-organic). On the common subset of molecules predicted by both methods, COSMOS makes predictions with an average speed per molecule of 0.15 s (0.10 s organic, 0.21 s metal-organic) and an average rmsd of 1.57 Å (1.26 Å organic, 1.90 Å metal-organic), and CORINA makes predictions with an average speed per molecule of 0.13s (0.18s organic, 0.08s metal-organic) and an average rmsd of 1.60 Å (1.13 Å organic, 2.11 Å metal-organic). COSMOS is available through the ChemDB chemoinformatics Web portal at http://cdb.ics.uci.edu/ .

CTLs' repertoire shaping in the thymus: A Monte Carlo simulation

Motivation: The human immune system evolved a multi-layered control mechanism to eliminate self-reactive cells. Of these so-called tolerance induction mechanisms, lymphocytes T education in the thymus gland represents the very first one. This complicated process is not fully understood and quantitative models able to help in this endeavor are lacking. Here, we present a stochastic computational model of the thymus which combines data-driven prediction methods and a novel method based on protein–protein potential measurements for assessing molecular binding among cell receptors, major histocompatibility complex (MHC) molecules, and self-peptides. Results: Of all possible specificities of immature T cells entering the thymus, only a small fraction is actually selected for maturation. Monte Carlo simulations of thymocytes selection in the thymus are performed varying the size of the self and a parameter determining the number of encounter with antigen-presenting cells (APCs). We score the fraction of self-reacting thymocytes leaving the thymus as mature naive T cells and show that self-reactivity is only marginally dependent on the number of self-molecules presented by APCs, while it is strongly affected by a parameter proportional to the time spent in the thymus. We study how this measure changes when we vary the number of MHC alleles and found an optimal number not too different from what we have in reality. The main result of this study is more methodological than biological as we show that immunoinformatics data and methods can be used in systemic level simulation of immune processes.

Data-driven prediction model of indoor air quality in an underground space

Several data-driven prediction methods based on multiple linear regression (MLR), neural network (NN), and recurrent neural network (RNN) for the indoor air quality in a subway station are developed and compared. The RNN model can predict the air pollutant concentrations at a platform of a subway station by adding the previous temporal information of the pollutants on yesterday to the model. To optimize the prediction model, the variable importance in the projection (VIP) of the partial least squares (PLS) is used to select key input variables as a preprocessing step. The prediction models are applied to a real indoor air quality dataset from telemonitoring systems data (TMS), which exhibits some nonlinear dynamic behaviors show that the selected key variables have strong influence on the prediction performances of the models. It demonstrates that the RNN model has the ability to model the nonlinear and dynamic system, and the predicted result of the RNN model gives better modeling performance and higher interpretability than other data-driven prediction models.

Computational Immunology Meets Bioinformatics: The Use of Prediction Tools for Molecular Binding in the Simulation of the Immune System

We present a new approach to the study of the immune system that combines techniques of systems biology with information provided by data-driven prediction methods. To this end, we have extended an agent-based simulator of the immune response, C-ImmSim, such that it represents pathogens, as well as lymphocytes receptors, by means of their amino acid sequences and makes use of bioinformatics methods for T and B cell epitope prediction. This is a key step for the simulation of the immune response, because it determines immunogenicity. The binding of the epitope, which is the immunogenic part of an invading pathogen, together with activation and cooperation from T helper cells, is required to trigger an immune response in the affected host. To determine a pathogen's epitopes, we use existing prediction methods. In addition, we propose a novel method, which uses Miyazawa and Jernigan protein–protein potential measurements, for assessing molecular binding in the context of immune complexes. We benchmark the resulting model by simulating a classical immunization experiment that reproduces the development of immune memory. We also investigate the role of major histocompatibility complex (MHC) haplotype heterozygosity and homozygosity with respect to the influenza virus and show that there is an advantage to heterozygosity. Finally, we investigate the emergence of one or more dominating clones of lymphocytes in the situation of chronic exposure to the same immunogenic molecule and show that high affinity clones proliferate more than any other. These results show that the simulator produces dynamics that are stable and consistent with basic immunological knowledge. We believe that the combination of genomic information and simulation of the dynamics of the immune system, in one single tool, can offer new perspectives for a better understanding of the immune system.

Data-driven predictive control for blast furnace ironmaking process

High performance control of blast furnace (BF) ironmaking process is a difficult problem due to the high temperature and hostile measurement conditions for measuring devices in the process. Previous research focused on developing of accurate predictive models for silicon content in hot metal ([SI]) while control of the whole process is seldom discussed. In the present work, a data-driven predictive control method based on subspace method is presented for the blast furnace ironmaking process. The algorithm is based on input–output data and easy to implement. Simulation results show the algorithm is effective for the control application. Finally, various practical issues concerning predictive control of blast furnace ironmaking process are also addressed, such as constraint handling, control objective and output set-point selection, adaptive strategy, etc.

Architecture for development of adaptive on-line prediction models

This work presents an architecture for the development of on-line prediction models. The architecture defines unified modular environment based on three concepts from machine learning, these are: (i) ensemble methods, (ii) local learning, and (iii) meta learning. The three concepts are organised in a three layer hierarchy within the architecture. For the actual prediction making any data-driven predictive method such as artificial neural network, support vector machines, etc. can be implemented and plugged in. In addition to the predictive methods, data pre-processing methods can also be implemented as plug-ins. Models developed according to the architecture can be trained and operated in different modes. With regard to the training, the architecture supports the building of initial models based on a batch of training data, but if this data is not available the models can also be trained in incremental mode. In a scenario where correct target values are (occasionally) available during the run-time, the architecture supports life-long learning by providing several adaptation mechanisms across the three hierarchical levels. In order to demonstrate its practicality, we show how the issues of current soft sensor development and maintenance can be effectively dealt with by using the architecture as a construction plan for the development of adaptive soft sensing algorithms.