Power Prediction Error Research Articles

Low-carbon unit commitment with intensive wind power generation and carbon capture power plant

The paper proposes a stochastic unit commitment (UC) model to realize the low-carbon operation requirement and cope with wind power prediction errors for power systems with intensive wind power and carbon capture power plant (CCPP). A linear re-dispatch strategy is introduced to compensate the wind power deviation from the spot forecast. The robust optimization technique is employed to obtain a reliable commitment plan against all realizations of wind power within the uncertainty set given by probabilistic forecast. The proposed model is validated with IEEE 39-bus system. The advantages of flexible CCPPs are compared to the normal coal-fueled plants and the impacts of robustness controlling are discussed.

Risk assessment of wind power prediction errors

the requirements on the error evaluation index for wind power prediction(wpp)and drawbacks of the traditional evaluation indexes are summarized.by distinguishing the influences on electric power reliability of a positive error from those of a negative error,a risk assessment index is proposed for wpp errors and its calculation.the risk assessment index is defined as an integration of the probability of a power disturbance event caused by wpp error and the caused by the event.having a monetary dimension,the index can be accumulated as risk cost of wpp error with other costs directly.thus,the confusion on how to trade off ignoring the events of small probability against' attaching importance to the events with heavy losses can be dispelled for those events with big error as well as very small probability.finally,an actual wind farm system in ningxia,china is applied to illustrating the feasibility and availability of the risk index.

Enhancing Regression Models for Complex Systems Using Evolutionary Techniques for Feature Engineering

This work proposes an automatic methodology for modeling complex systems. Our methodology is based on the combination of Grammatical Evolution and classical regression to obtain an optimal set of features that take part of a linear and convex model. This technique provides both Feature Engineering and Symbolic Regression in order to infer accurate models with no effort or designer's expertise requirements. As advanced Cloud services are becoming mainstream, the contribution of data centers in the overall power consumption of modern cities is growing dramatically. These facilities consume from 10 to 100 times more power per square foot than typical office buildings. Modeling the power consumption for these infrastructures is crucial to anticipate the effects of aggressive optimization policies, but accurate and fast power modeling is a complex challenge for high-end servers not yet satisfied by analytical approaches. For this case study, our methodology minimizes error in power prediction. This work has been tested using real Cloud applications resulting on an average error in power estimation of 3.98 %. Our work improves the possibilities of deriving Cloud energy efficient policies in Cloud data centers being applicable to other computing environments with similar characteristics.

Reserve Capacity Optimization of Large Scale Wind Power Integration

The traditional method of power system reserve capacity is not suitable for the power system with large-scale wind power integration. This paper analyzed the prediction error of wind power, the introduction of the system reliability index, established a correlation mode between wind power prediction error and the spare demand change; the introduction of interruptible load to replace part of the reserve capacity, combining probability of the using of spare capacity in historical data, according to the standby capacity demand use of decision theory expected value method is used to solve the optimal reserve. The practical examples prove that the algorithm is feasible.

Biometry and intraocular lens power calculation results with a new optical biometry device: Comparison with the gold standard

To evaluate the agreement in axial length (AL), keratometry (K), anterior chamber depth (ACD) measurements; intraocular lens (IOL) power calculations; and predictability using a new partial coherence interferometry (PCI) optical biometer (AL-Scan) and a reference (gold standard) PCI optical biometer (IOLMaster 500). Service d'Ophtalmologie, Hopital Bicêtre, APHP Université, Paris, France. Evaluation of a diagnostic device. One eye of consecutive patients scheduled for cataract surgery was measured. Biometry was performed with the new biometer and the reference biometer. Comparisons were performed for AL, average K at 2.4 mm, ACD, IOL power calculations with the Haigis and SRK/T formulas, and postoperative predictability of the devices. A P value less than 0.05 was statistically significant. The study enrolled 50 patients (mean age 72.6 years±4.2 SEM). There was a good correlation between biometers for AL, K, and ACD measurements (r=0.999, r=0.933, and r=0.701, respectively) and between IOL power calculation with the Haigis formula (r=0.972) and the SRK/T formula (r=0.981). The mean absolute error (MAE) in IOL power prediction was 0.42±0.08 diopter (D) with the new biometer and 0.44±0.08 D with the reference biometer. The MAE was 0.20 D with the Haigis formula and 0.19 with the SRK/T formula (P=.36). The new PCI biometer provided valid measurements compared with the current gold standard, indicating that the new device can be used for IOL power calculations for routine cataract surgery. No author has a financial or proprietary interest in any material or method mentioned.

Development and validation of a gray box model to predict thermal behavior of occupied office buildings

Due to the development of energy performance contracting and the needs for peak electric demand reduction, the interest for building energy demand prediction is renewed. Gray-box models are a solution for energy demand prediction. However, it is still difficult to find the best level of model complexity and the good practices for the training phase. Since models’ order and parameter identification method have a strong impact on the forecasting precision and are not intuitive, a comparative design approach is used to find the best model architecture and an adequate methodology for improving the training phase. The gray box models are compared on their ability to forecast heating and cooling demands and indoor air temperature. An objective function is proposed aiming to minimize both power and indoor temperature prediction errors. Moreover, for each model, several training period durations are tested. First, this study shows that a R6C2 (second order model) model is adapted to predict the building thermal behavior. Furthermore, the best fits are obtained with two weeks of data for the identification process. Second, a sensitivity analysis using total Sobol index calculation leads to validate the objective function and identify the most important parameters for prediction.

Multi-Scenario Based Generation Scheduling of Power System with Wind Farms

Wind power penetration to power systems is increasing rapidly in the recent years due to its environmental benefit, while wind power fluctuation also brings some problems to power system operation which impacts the generation of conventional power units. For this reason, probabilistic optimal dispatching model based on multi-scenasio is developed in this paper. With the discretization of wind power fluctuations range, the scenario probability can be get by discretizing wind power prediction error distribution curves, at the same time considering the relevance of the prediction error of the adjacent periods in the time scale. By means of leading probability adjustment costs into objective function, the optimization result can consider the cost due to the fluctuation of wind power. The rationality and effectiveness of the proposed method is verified by testing and demonstrating IEEE-39 bus system with a wind farm.

An energy‐saving algorithm for cloud resource management using a Kalman filter

SUMMARYCloud computing has emerged as a promising technique to provide storage and computing component on‐demand services over a network. In this paper, we present an energy‐saving algorithm using the Kalman filter for cloud resource management to predict the workload and to further achieve high resource availability with low service level agreement. Using the proposed algorithm, one can estimate the potential future workload trend then predict the computing component workload utilizations and further retrench energy consumption and achieve load balancing in a cloud system. Experimental results show that the proposed algorithm achieves more than 92.22% accuracy in the computing component workload prediction, improves 55.11% energy in energy consumption, and has 3.71% in power prediction error rate, respectively. Copyright © 2013 John Wiley & Sons, Ltd.

Rolling Generation Dispatch Based on Ultra-short-term Wind Power Forecast

The power systems economic and safety operation considering large-scale wind power penetration are now facing great challenges, which are based on reliable power supply and predictable load demands in the past. A rolling generation dispatch model based on ultra-short-term wind power forecast was proposed. In generation dispatch process, the model rolling correct not only the conventional units power output but also the power from wind farm, simultaneously. Second order Markov chain model was utilized to modify wind power prediction error state (WPPES) and update forecast results of wind power over the remaining dispatch periods. The prime-dual affine scaling interior point method was used to solve the proposed model that taken into account the constraints of multi-periods power balance, unit output adjustment, up spinning reserve and down spinning reserve.

Revision of reserve requirements following wind power integration in island power systems

This contribution studies the impact of wind power on the operation of island power systems. The analysis focuses on the available flexibility of thermal generation to balance wind power variations and prediction errors. This issue is highly relevant for small and medium sized islands where interconnections are absent and the smoothing of variations is limited due to a small geographical surface. The main objective is to determine if additional reserve requirements are necessary for ensuring reliable wind power integration in an isolated transmission system. In this context, a case study is performed for Cyprus, a medium-sized island, where wind developments towards 2020 schedule an installed capacity of 300 MW, reaching 7% of the annual electricity consumption. Simulations with installed wind power capacities up to 400 MW show that the current available flexibility in the generation system is inadequate to balance real-time wind power fluctuations and prediction errors. Consequently, large amounts of wind curtailment and demand shedding may be expected. Therefore, current reserve requirements should be revised, in order to reliably facilitate wind power into the system. Furthermore, the impact of introducing natural gas for electricity generation in Cyprus on the reserve requirements, following wind power integration, is examined.

Selection of fuel channels for Thermal Power Measurement in 700MWe Indian PHWR by evolutionary algorithm

Abstract This paper presents studies on the design of Thermal Power Monitoring System (TPMS) for the forthcoming 700 MWe Indian Pressurized Heavy Water Reactor (PHWR). This reactor contains total 392 horizontal fuel channels. Each channel contains clustered natural Uranium fuel along with associated heavy water coolant placed inside a pressure tube. The coolant in different fuel channels is physically and thermally isolated from each other inside the core. It is necessary to select 44 fuel channels (out of 392) for keeping instrumentation to measure flow and temperature of coolant. The reactor is logically divided into 7 radial zones each containing certain number of fuel channels. The selection of instrumented channels is to be made such that power measured by them in terms of per unit basis represents the true zone-wise and global powers fairly accurately. This should be possible for a large number of reactor configurations that can occur because of the movement of reactivity devices in the core. Such a study is useful to make the TPMS more accurate means to measure the reactor bulk power and zone powers. The choice of 44 channels is an optimization problem in which the error in zonal and global power prediction is to be minimized. There are several constraints on the selection of instrumented channels. Therefore, a constrained combinatorial optimization problem has to be solved. An evolutionary technique based on Estimation of Distribution Algorithm (EDA) is used for this purpose. A suitable pattern of 44 instrumented channels is obtained for which errors in zonal and global powers are less than 0.5% and 2.0% respectively. The equilibrium state of the core with 357 possible device configurations is considered for the optimization.

Optimal operation of a pumped-storage hydro plant that compensates the imbalances of a wind power producer

Abstract The participation of wind energy in electricity markets requires providing a forecast for future energy production of a wind generator, whose value will be its scheduled energy. Deviations from this schedule because of prediction errors could imply the payment of imbalance costs. In order to decrease these costs, a joint operation between a wind farm and a hydro-pump plant is proposed; the hydro-pump plant changes its production to compensate wind power prediction errors. In order to optimize this operation, the uncertainty of the wind power forecast is modeled and quantified. This uncertainty is included in an optimization problem that shifts the production of the hydro-pump plant in an optimal way, aiming at reducing the imbalance costs. The result of such a method is profitable for both participants, the wind farm and the hydro-pump plant. A realistic test case is used to evaluate the proposed method.

Optimal Participation of Wind Power in the Liberalized Power Market

AbstractWhile installing wind power into the liberalized power market can reduce the operation and fuel costs of the power system, it also increases the imbalance cost and need for reserve. Due to this contradictory, there is a need for planning the amount of wind power in the power system. This paper proposes a model following American power market rules to find the optimal participation of wind power. This model assumes that wind power producer can bid into both day-ahead and real-time market and pay imbalance cost due to wind power prediction errors. A case study will be presented where the proposed model will be tested with simulation results on a practical power system.

Improving the prediction accuracy of the SRK/T formula: The T2 formula

To investigate the causes of nonphysiologic behavior of the SRK/T formula, assess their clinical significance, and develop and evaluate solutions. Two NHS ophthalmology departments, United Kingdom, and a private practice, United States. Evaluation of technology. The individual steps of the SRK/T formula were examined for nonphysiologic behavior, and the clinical significance of behaviors was assessed with reference to a database of biometry and refractive outcomes in 11 189 eyes. The full data set was divided into 2 subsets, the first to develop solutions to nonphysiologic behavior of the SRK/T formula and the second to evaluate their performance. The SRK/T formula showed nonphysiologic behavior in the calculation of corrected axial length and corneal height. Although the former is of little clinical significance, the latter showed a systematic error that contributes to inaccurate intraocular lens (IOL) power prediction. The T2 formula was developed using a regression formula for corneal height derived from the development subset. Comparison of the performance of the T2 and SRK/T formulas using the evaluation subset showed significant improvement in the mean absolute error with the T2 formula (0.3064 diopter [D] versus 0.3229 D; P<.0001). On average, the prediction error with the T2 formula was 9.7% less than with the SRK/T formula, with significantly higher proportions of eyes within ±0.50 D of target (P<.0001). The SRK/T formula has nonphysiologic behavior that contributes to IOL power prediction errors. A modification to the formula algorithm, the T2 formula, can be directly substituted for SRK/T, resulting in significantly improved prediction accuracy.

Establishing expert system for prediction based on the project-oriented data warehouse

Many studies have generated cost estimating relationships (CERs) for transportation projects via data analysis. Some studies collected data from databases, while others sourced data from conventional paper-based formats. When cost data were not in a consistent format, many studies failed to discuss the streamlining of pattern recognition, ranging from generating a problem statement, data warehouse and prediction modeling to information management. This study adopts a standard procedure of identifying CERs for transportation projects. For the proposed dimensional data warehouse, a pavement maintenance and rehabilitation project was selected as a case study for extracting data and concealed prediction rules. Linear and log-linear statistical approaches were adopted to create most advantageous models, defined based on their explanatory power and mean absolute prediction error. The resulting favorable estimation models created from the proposed cost data warehouse were integrated into an expert system to facilitate information management and generate preliminary budgets for transportation agencies.

Comparison of 2 laser instruments for measuring axial length

To compare axial length (AL), anterior chamber depth (ACD), and keratometric (K) measurements of 2 laser biometers. Private practices, Lynwood and Santa Monica, California, USA. In this prospective comparative observational study of eyes with cataract and eyes with a clear lens, AL, ACD, and K measurements were performed using an IOLMaster biometer, which uses partial coherence interferometry (PCI), and a Lenstar LS 900 biometer, which uses optical low-coherence reflectometry (OLCR). Intraocular lens (IOL) power calculation was performed using the Haigis formula. The IOL prediction error was calculated for each eye. The study evaluated 50 eyes with cataract and 50 eyes with a clear lens. There was a good correlation between AL, ACD, and K measurements in the cataractous eyes (r = 0.9993, 0.9667, and 0.9959, respectively) and in eyes with a clear lens (r = 0.9995, 0.8211, and 0.9959, respectively). The OLCR unit measured a slightly longer AL in the cataract group and clear lens group (mean difference 0.026 mm and 0.023 mm, respectively), a deeper ACD (0.128 mm and 0.146 mm, respectively), and a flatter K (-0.107 diopter [D] and -0.121 D, respectively). The differences were statistically significant (P<.0001). The mean absolute error in IOL power prediction was 0.455 D +/- 0.32 (SD) with the OLCR unit and 0.461 +/- 0.31 D with the PCI unit (P>.1). Measurements were comparable between the OLCR device and the PCI device. A slight decrease (0.050) in the a(0) constant is recommended if the Haigis formula is used.

Accuracy of preoperative biometry in vitrectomy combined with cataract surgery for patients with epiretinal membranes and macular holes: Results of a prospective controlled clinical trial

To evaluate how adding vitrectomy to cataract surgery affects the accuracy of preoperative biometry and postoperative refractive outcomes. Department of Ophthalmology, Ludwig Boltzmann Institute of Retinology and Biomicroscopic Lasersurgery, Rudolf Foundation Clinic, Vienna, Austria. This study comprised 40 patients with vitreoretinal pathology and coexisting significant cataract (study group) and 40 patients with significant cataract only (control group). The main outcome measure was intraocular lens (IOL) power prediction error. Secondary outcome measures were spherical equivalent, anterior chamber depth, axial length, keratometry values, and intraocular pressure. A multivariate regression analysis was performed. There was a significant difference in prediction error between the study group and control group (P< .05). The addition of vitrectomy was associated with induced myopia of approximately -0.4 diopter, more so in patients who had epiretinal membrane removal than in patients with macular hole surgery (P= .04). There was no significant difference in prediction error between the 3 IOLs used (P= .2). The myopic shift was strongly dependent on the diagnosis and the need for intraocular tamponade, indicating that a slightly hyperopic IOL should be used in patients having combined surgery.

Intraocular Lens Power Calculation After Automated Lamellar Keratoplasty for High Myopia

To report intraocular lens (IOL) power calculation in 2 eyes that were highly undercorrected by previous myopic automated lamellar keratoplasty (ALK). A 35-year-old man underwent bilateral myopic ALK, which caused high residual myopia (-9.0 -4.0 x 171 and -9.5 -4.5 x 74). The patient then underwent cataract surgery with IOL implantation for cataract development. The double-K clinical history method was utilized, and satisfactory IOL power prediction results were obtained. Two no-history IOL power calculation methods (Rosa correcting factor method and Ferrara theoretical variable refractive index method), which involved axial length-dependent modification of the keratometer-measured corneal radius, and 1 no-history IOL power calculation method (Shammas' method), which involved axial length-independent modification of the keratometer-measured corneal power, were tested on these 2 eyes. In both eyes, the double-K SRK-T clinical history method gave small IOL prediction errors (-0.66 and -0.81 D). The Shammas' no-history method gave a slightly higher IOL prediction error in the right eye (-1.67 D) and a small IOL prediction error in the left eye (-0.74 D). Unacceptable IOL power prediction errors would have resulted if Rosa's correcting factor method (-8.07 and -8.35 D) or Ferrara's theoretical variable refractive index method (-17.56 and -18.51 D) had been applied. When we utilized Rosa's method for the IOL power calculation by assuming that the previous ALK had fully corrected the refractive error, the predicted IOL powers were very close to the benchmark IOL powers (IOL power prediction errors: 1.16 and 0.37 D). When we utilized Ferrara's method with the same assumption, the IOL power prediction errors remained high (-6.32 and -7.16 D). For patients who have had previous myopic ALK (and whose eyes are highly undercorrected) and who require cataract surgery and for whom the pre-ALK history is available, the double-K method appears to yield excellent predictive results. However, if the pre-ALK history is not available, the Shammas' no-history method appears to yield better results than the Rosa's or the Ferrara's method. High undercorrection by the previous ALK might have been one of the major reasons why Rosa's method resulted in a high IOL prediction error in these 2 eyes. The cause for the marked IOL prediction error by Ferrara's method in this case, however, remains to be determined.

Intraocular Lens Calculations after Hyperopic Refractive Surgery

To evaluate the effect of hyperopic refractive surgery on intraocular lens (IOL) power calculation, compare published methods of IOL power calculation after refractive surgery, evaluate the effect of prerefractive surgery refractive error on IOL deviation, and introduce a new alternative formula for IOL calculation in patients who have had refractive surgery for hyperopia. Retrospective noncomparative case series. Twenty eyes from 13 patients who had undergone cataract surgery after previous hyperopic refractive surgery. Seven different methods of IOL calculation were performed retrospectively: clinical history (IOL(hisK)), clinical history method at spectacle plane (IOL(hisKs)), vertex (IOL(vertex)), back calculated (IOL(BC)), calculation based on average keratometry (IOL(avgK)), calculation based on steepest keratometry (IOL(steepK)), and calculation based on the double K formula (IOL(doubleK)). Each method's result was compared with an exact IOL (IOL(exact)), which would have resulted in emmetropia. Each method was then compared with change in spherical equivalent induced by refractive surgery (SE(h)). A paired t test was used to determine statistical significance. Mean error in IOL power prediction for each method when compared to IOL(exact). When evaluating different methods of IOL calculations, IOL(vertex) was the most accurate, with a mean deviation from emmetropia of 0.42+/-1.75 diopters (D), followed by IOL(BC) (+0.54+/-1.86 D), IOL(hisK) (+1.56+/-2.35 D), IOL(hisKs) (+1.57+/-2.35 D), IOL(steepK) (+1.59+/-2.25 D), IOL(doubleK) (+1.65+/-2.56 D), and IOL(avgK) (+2.24+/-2.46 D). There was no statistical difference between IOL(vertex), IOL(BC), and IOL(exact). The power of IOL(avgK) would be inaccurate by 0.27x+1.53, where x = SE(h). Thus, most patients without the adjustment to IOL(avgK) would be left myopic. However, when IOL(avgK) is adjusted with this formula, there is no statistical difference to IOL(exact). For IOL power selection in previously hyperopic patients, a predictive formula based only on SE(h) and current average keratometry readings was not found to statistically differ from IOL(exact). The IOL(vertex) and IOL(BC), which also did not statistically differ from IOL(exact), require prerefractive surgery keratometry readings that are often not available to the cataract surgeon.

Improved accuracy of intraocular lens power calculation with the Zeiss IOLMaster

This study aimed to demonstrate how the level of accuracy in intraocular lens (IOL) power calculation can be improved with optical biometry using partial optical coherence interferometry (PCI) (Zeiss IOLMaster) and current anterior chamber depth (ACD) prediction algorithms. Intraocular lens power in 461 consecutive cataract operations was calculated using both PCI and ultrasound and the accuracy of the results of each technique were compared. To illustrate the importance of ACD prediction per se, predictions were calculated using both a recently published 5-variable method and the Haigis 2-variable method and the results compared. All calculations were optimized in retrospect to account for systematic errors, including IOL constants and other off-set errors. The average absolute IOL prediction error (observed minus expected refraction) was 0.65 dioptres with ultrasound and 0.43 D with PCI using the 5-variable ACD prediction method (p < 0.00001). The number of predictions within +/- 0.5 D, +/- 1.0 D and +/- 2.0 D of the expected outcome was 62.5%, 92.4% and 99.9% with PCI, compared with 45.5%, 77.3% and 98.4% with ultrasound, respectively (p < 0.00001). The 2-variable ACD method resulted in an average error in PCI predictions of 0.46 D, which was significantly higher than the error in the 5-variable method (p < 0.001). The accuracy of IOL power calculation can be significantly improved using calibrated axial length readings obtained with PCI and modern IOL power calculation formulas incorporating the latest generation ACD prediction algorithms.