Literatura académica sobre el tema "Fuel cell/battery diagnosis/prognosis"

Fuel cell vehicle powertrain is a complex electromechanical system related to electrochemistry, power electronics, power transmission and control systems, network and communications and other disciplines. Therefore, this article has clearly made the concept of fault and fault diagnosis, and conducted the preliminary study of the method of fault diagnosis for the fuel cell vehicle battery, and will adopt the model-based fault diagnosis method to achieve battery fault diagnosis.

Abstract Background: AML is a complex disease that encompasses a huge variation of cytogenetic and mutational backgrounds, which is often complicated by age related functional deficits (Klepin et al. 2013). Given the expanding availability of reduced-intensity treatment options, patient fitness and frailty measures have become increasingly instrumental in the decision between a wide range of treatment options. Furthermore, studies have also demonstrated potential benefits in older patients who receive intensive induction chemotherapy (Julisson 2011), creating a need for additional assessments to identify suitable induction candidates. Some frailty-associated assessments including timed-up-and-go test (TUGT) and the short physical performance battery have been linked to outcomes but are not yet in broad clinical use (Kleplin et al. 2013, Khalaf et al. 2020). To minimize the burden of implementing new patient assessments, we evaluated the utility of commonly available clinical measures of frailty-associated factors including sit-to-stand test and iADL status in predicting patient outcomes. Methods: We performed a retrospective cohort study of elderly patients newly diagnosed with AML at Juravinski Cancer Center between Jan 2019 and Dec 2020. We examined a total of 44 patients aged 65+. The primary outcome was overall survival (OS). Significant risk factors were identified using the Cox proportional hazards model. Results: 43 patients had sufficient data and were included in the analysis. The median age was 70 years (range 65-89), 53% were female and 47% were male. At the time of review, 21 patients were deceased (48.8%). The median survival time was 345 days. 26 patients received full induction treatment (7+3, FLAG-IDA, or Vyxeos), and 18 received conservative treatment (LDAC, azacytidine, or supportive care). The clinical wellness of the patients at diagnosis time was assessed by baseline clinical and laboratory findings. (Table 1) The Cox regression model was used to examine these variables in predicting overall survival (Figure 1). As expected, the ELN risk group was significantly correlated with OS (HR 2.94 95%CI [1.41-6.11]), along with laboratory measures of disease burden including blood blast count (HR 1.02 [1.00-1.03]), WBC, (HR 1.02 [1.01-1.03]) and ANC (HR 1.05 [1.02-1.08]). We then assessed the influence of patient fitness factors on OS. The HCT-CI score was used as an aggregate comorbidities measure. While typically used for post-SCT prognosis, we found that HCT-CI assessed at diagnosis time was a significant predictor of OS (HR 1.22[1.04-1.45]). When added to a multivariate Cox model including ELN and age, the HCT-CI independently predicted OS (p = 0.003) and improved prediction efficiency by the Akaike information criteria (115 vs 122). This corroborates earlier findings by Sorror et al. 2017, who showed incorporation of biochemical and AML-specific variables to HCT-CI yields improved prognostic value. Within the HCT-CI, the AST and cardiac disease scores were the most associated with OS (HR 1.03[1.00-1.05] and 2.27[1.09-4.76]). While clinical and laboratory assessments were readily available, functional assessments of frailty were scarce. Previously reported frailty measures such as KPS score or TUGT were not assessed at the study center. OT/PT routinely administer sit-to-stand or standing balance tests as a part of fall risk assessment. However, this data was only available for 18 patients (41.8%). Independent sit-to-stand was not detected as a significant OS predictor (HR 0.63[0.14-2.83]), possibly related to the very limited sample size. BMI was marginally predictive (HR 1.07[0.99 - 1.08]), but unlike HCT-CI it was not an independent predictor when combined with ELN. Patient age did not significantly predict OS. Conclusion: This single center retrospective study was aimed at examining the role of existing clinical or functional measures of fitness and frailty in predicting overall survival. HCT-CI and ELN were found to be the most predictive factors amongst the variables examined, suggesting that a morbidity index such as HCT-CI could provide prognostic utility. However, functional assessments of frailty were not readily completed, limiting the ability to evaluate their usefulness. A future larger prospective study focused on optimizing and incorporating routine functional assessments of frailty is needed to address this topic. Figure 1 Figure 1. Disclosures Leber: Celgene: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Otsuka: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Astellas: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; TaiHo: Honoraria, Membership on an entity's Board of Directors or advisory committees; AMGEN: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Abbvie: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; BMS: Honoraria, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Pfizer: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Novartis: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Jazz: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau. Khalaf: Novartis: Honoraria; Paladin: Honoraria; Pfizer: Honoraria.

The diagnosis and prognosis of deployed battery energy storage systems is difficult because the battery might never experience controlled conditions through normal operation as both the day-to-day charge and discharge duty cycles might be sporadic. To circumvent this issue new methodologies must be developed, and we recently proposed one for photovoltaic connected batteries. The method is using real observed solar irradiance, modeled clear sky irradiance and synthetically generated battery data from a battery digital twin to diagnose the battery degradation. The approach was demonstrated to be effective for opportunistic diagnosis without the need for any maintenance cycles for nickel manganese cobalt oxide-based batteries for days with over 50% clear sky or with an average irradiance over 650 W/m2. This work is validating the approach further for lithium iron phosphate-based batteries. We applied our data-driven diagnosis technique to synthetically generated PV connected LFP cells and tested more than 10,000 different degradations on 720 unique days. Overall, the results showcase that, similarly to what was observed for the NMC cell, diagnosis is possible for days above 50% clear sky or for high average irradiance. The main difference was that LAMPE estimation was proven much more problematic for LFP Cells, especially for small LAMPEs. This, however, should not hamper deployability since the analysis of the impact of the degradation composition and extent indicated that the accuracy of the diagnosis is restored when LAMPE is prominent, and thus a potential safety hazard. This additional analysis showcases the benefits of using varied synthetic datasets to investigate the impact of all possible degradation on the cells.

The global transition to electric power, aimed at mitigating climate change and addressing fuel shortages, has led to a rising usage of lithium-ion batteries (LIBs) in different fields, notably transportation. Despite their many benefits, LIBs pose a critical safety concern due to the potential for thermal runaway (TR), often triggered by spontaneous internal short circuit (ISC) formation. While extensive research on LIB fault diagnosis and prognosis exists, forecasting ISC formation in batteries remains unexplored. This paper presents a new methodology that combines the extended Kalman filter (EKF) algorithm for real-time estimation of ISC state with an adaptive linear regressor model for forecasting remaining useful life (RUL). This approach is designed for seamless integration into actual battery management systems, offering a computationally efficient solution. Numerical validation of the framework was conducted due to the current lack of experimental data in the literature. The significance of this work lies in its contribution to ISC prognosis, providing a practical solution to enhance battery safety.

The proton battery possesses water electrolysis, proton storage and discharging functions simultaneously, and it can be manufactured without expensive metals. Use the principle of proton exchange membrane water electrolysis for charging, store it in the activated carbon on the hydrogen side and use the principle of proton exchange membrane fuel cell for discharge when needed. According to the latest literature, it is difficult to obtain the exact important physical parameters inside the proton battery (e.g., voltage, current, temperature, humidity and flow), and the important physical parameters are correlated with each other, which have critical influence on the performance, lifetime and health status of the proton battery. At present, the condition of the proton battery is judged indirectly only by external measurement, the actual situation inside the proton battery cannot be obtained accurately and instantly. Therefore, this study uses micro-electro-mechanical systems (MEMS) technology to develop a flexible 5-in-1 microsensor, which is embedded in the proton battery to obtain five important physical parameters instantly, so that the condition inside the proton battery can be mastered more precisely, so as to prolong the battery life and enhance the proton battery performance.

Fuel cells, as clean and efficient energy conversion devices, hold great potential for applications in the fields of hydrogen-based transportation and stand-alone power systems. Due to their sensitivity to load parameters, environmental parameters, and gas supply, the performance monitoring and fault diagnosis of fuel cell systems have become crucial research areas. Electrochemical impedance spectroscopy (EIS) is a widely applied analytical method in fuel cell systems. that can provide rich information about dynamic system responses, internal impedance, and transmission characteristics. Currently, EIS detection is primarily implemented by using simple topologies such as boost circuits. However, the injection of excitation signals often results in significant power fluctuations, leading to issues such as uneven temperature distributions within the cell, unstable gas supply, and damage to the proton exchange membrane. To address this issue, this paper proposes a real-time EIS detection technique for a proton exchange membrane fuel cell (PEMFC) system that connects a lithium-ion battery and injects the load voltage perturbation through a triple active bridge (TAB) converter. By applying the small-signal model of the TAB converter and designing a system controller using a decoupling control method, the PEMFC power remains stable after the disturbance injection across the entire frequency range under tests. Furthermore, the lithium-ion battery can instantly track load changes during fluctuations. The proposed EIS detection method can acquire EIS data in real time to monitor the state of the PEMFC. Simulation results validate the effectiveness and accuracy of the proposed method for EIS detection.

Fast-switching semiconductors induce ripple currents on the high-voltage DC bus in the electric vehicle (EV). This paper describes the methods used in the project SiCWell and a new approach to investigate the influence of these overlaid ripples on the battery in EVs. The ripple current generated by the main inverter is demonstrated with a measurement obtained from an electric vehicle. A simulation model is presented which is based on an artificial reference DC bus, according to ISO 21498-2, and uses driving cycles in order to obtain current profiles relevant for battery cycling. A prototype of a battery cycling tester capable of high frequency and precise ripple current generation was developed and is used to cycle cells with superimposed ripple currents within an aging study. To investigate the impact of the frequency and the amplitude of the currents on the battery’s lifetime, these ripple parameters are varied between different test series. Cell parameters such as impedance and capacity are regularly characterized and the aging of the cells is compared to standard DC cycled reference cells. The aging study includes a total of 60 automotive-sized pouch cells. The evaluation of ripple currents and their impact on the battery can improve the state-of-health diagnosis and remaining-useful life prognosis. For the development and validation of such methods, the cycled cells are monitored with a measurement system that regularly measures current and voltage with a sampling rate of 2 MHz. The resulting dataset is suitable for the design of future ripple current aging studies as well as for the development and validation of aging models and methods for battery diagnosis.

Redox flow batteries have the potential to meet long duration energy storage requirements of the implementation of intermittent renewable energy technologies such as solar and wind power. However, the current process for validation of new energy storage technologies requires extensive and time-consuming testing to determine the battery performance over long-term cycling and the lifetime of the battery. In this study, we develop machine learning models for forecasting long term battery performance as well as for the diagnosis and prognosis of the degradation modes driving this performance. To train these models, we leverage a semi-analytical model to generate a large set of synthetic data across a diverse range of operating conditions, cell properties, initial conditions, degradation mechanisms, and degradation models. Based on this data, we develop deep learning models for forecasting voltage curves, cycle-level performance metrics including charging capacity, discharging capacity and energy efficiency, and degradation parameters such as membrane conductivity and electrode porosity. We compare Transformer architectures designed for time series data such as Autoformer and PatchTST, with an operator learning method called Deep Operator Networks (DeepONets) which can predict battery voltage based on the operating parameters and state of charge. By providing insights into expected model performance and lifetimes, such models have the capacity to accelerate the technology validation process, reduce the real-time testing requirements, and shorten the timeline for commercial deployment.

This paper presents a model of an energy system for a private household extended by a lifetime prognosis. The energy system was designed for fully covering the year-round energy demand of a private household on the basis of electricity generated by a photovoltaic (PV) system, using a hybrid energy storage system consisting of a hydrogen unit and a lithium-ion battery. Hydrogen is produced with a Proton Exchange Membrane (PEM) electrolyser by PV surplus during the summer months and then stored in a hydrogen tank. Mainly during winter, in terms of lack of PV energy, the hydrogen is converted back into electricity and heat by a fuel cell. The model was created in Matlab/Simulink and is based on real input data. Heat demand was also taken into account and is covered by a heat pump. The simulation period is a full year to account for the seasonality of energy production and demand. Due to high initial costs, the longevity of such an energy system is of vital interest. Therefore, this model was extended by a lifetime prediction in order to optimize the dimensioning with the aim of lifetime extension of a hydrogen-based energy system. Lifetime influencing factors were identified on the basis of a literature review and were integrated in the model. An extensive parameter study was performed to evaluate different dimensionings regarding the energy balance and the lifetime of the three components, electrolyser, fuel cell and lithium-ion battery. The results demonstrate the benefits of a holistic modelling approach and enable a design optimization regarding the use of resources, lifetime and self-sufficiency of the system.

Le concept des systèmes d'énergie hybrides a amélioré considérablement les performances dynamiques, l'efficacité énergétique et la durée de vie des systèmes énergétiques, ce qui est très prometteur pour les chaines de conversion électrifiées. Ces dernières années, l'hybridation pile à combustible à membrane d'échange de protons (PEMFC) avec des batteries lithium (Libs) a été particulièrement mise en évidence par de nombreux projets de recherche. La stratégie de gestion d'énergie (EMS) joue un rôle essentiel au niveau de la supervision de ce type de système. Elle pilote les flux d'énergie entre les différentes sources pour satisfaire la charge tout en améliorant l'efficacité opérationnelle. Une EMS intelligente et performante doit être supportée par des résultats fiables relatifs au diagnostic / pronostic de chaque source, à savoir le diagnostic / pronostic de la PEMFC et les Libs qui est indispensable pour la synthèse d'une stratégie EMS consciente de leur santé. Cependant, les travaux de recherche actuels s'appuient souvent sur une décorrélation entre l'EMS et le diagnostic / pronostic associé. Ainsi, l'objectif principal de ces travaux est le développement d'un module de diagnostic/pronostic de chaque source et son intégration dans la synthèse de l'EMS. Les approches seront fondées sur des techniques d'intelligence artificielle au moyen d'un processus d'auto-apprentissage des données opérationnelles
The concept of hybrid energy systems have considerably improved dynamic performance, energy efficiency and lifespan of energetic systems, which is very promising for the powertrain system in electrified vehicles. Especially the effectiveness of hybrid system based on polymer electrolyte membrane fuel cells (PEMFCs) and lithium-ion batteries (LIBs) has been highlighted by many research projects in recent years. In such system, energy management strategy (EMS) plays a critical role in the supervisory level. It manage the power flow among the different power sources, to meet the power demand and to improve the operation efficiency. A smart EMS should be supported by reliable diagnostic/prognostic results of the single power source. Namely, the topic of diagnostic/prognostic of PEMFCs and LIBs is indispensable when designing a health-conscious EMS. However, the current research work of designing EMS and executing diagnostic/prognostic of ESS is usually independent of each other. Therefore, in this PhD project, integrating the diagnosis/prognosis module of each power source into EMS design is the primary aim. Especially, the proposed health-aware EMS and the module of diagnostic/prognostic will be based on artificial intelligence techniques through operation data self-learning process