Engine Health Monitoring Research Articles

Deep learning-based vehicular engine health monitoring system utilising a hybrid convolutional neural network/bidirectional gated recurrent unit

Vehicles play a pivotal role in the current era of Industry 4.0 by providing passengers with excellent mobility, comfort, and safety while strengthening national and international economies. Unanticipated vehicular engine issues can hinder performance and lead to costly maintenance. As analytics processes become faster, more accurate, and more reliable, intelligent fault prediction and diagnosis for vehicles, particularly engines, is becoming increasingly popular. To date, hybrid deep learning approaches to vehicle engine diagnostics have been limited, and none have used engine health monitoring and categorisation based on vulnerability assessment and vehicle structural information. This paper introduces a hybrid deep learning-based vehicular engine health monitoring system (VEHMS) decision model using Deep CNN (convolutional neural network)-BiGRU (bi-directional gated recurrent unit). This model monitors a vehicle’s engine health in real-time and classifies its status as good, critical, moderate, or minor condition. Several advanced and hybrid deep learning algorithms were applied to monitor engine health and categorise its status by integrating sensor data with evaluated vulnerability information from an infrastructure vulnerability assessment model. The Deep CNN-BiGRU-based VEHMS decision model outperformed other techniques with an accuracy of 0.8897, ensuring minimal decision losses while classifying engine conditions. This study aims to contribute to developing comprehensive vehicle health monitoring systems and advance the automotive industry by incorporating more intelligent features. The proposed approach can enhance vehicle performance, reliability, and efficiency in the transportation sector by improving engine health monitoring.

Improved deep residual shrinkage network for a multi-cylinder heavy-duty engine fault detection with single channel surface vibration

The health monitoring and fault diagnosis of heavy-duty engines are increasingly important for energy storage ecosystem. During operation, vibration characters corresponding to the specific fault need to be extracted from the overall system vibration. Faulty characteristics emanating from one single cylinder are also mixed with those from other cylinders. Besides, the change of working condition brings strong nonlinearities in surface vibration. To solve these problems, an improved deep residual shrinkage network (IDRSN) is developed for detecting diverse engine faults at various degrees using single channel surface vibration signal. Within IDRSN, a wide convolution kernel is utilized in first convolution layer to capture the long-term fault-related impacts and eliminate the short-time random impact. The residual network module is adopted to enhance the focus the relevant components of vibration signals. Mini-batch training strategy is used to improve the model stability. Meanwhile, Gradient-weighted class activation map is adopted to assess the consistency between the learned knowledge and the fault-related information. The IDRSN is implemented to diagnosing a diesel engine under various faults, faulty degrees and operating speeds. Comparisons with existing models are analyzed in terms of hyper-parameters, training samples, noise resistance, and visualization. Results demonstrate the proposed IDRSN's superior performance on fault diagnosis accuracy, stability, anti-noise performance, and anti-interference performance. An average accuracy rate of 98.38 % was achieved by the proposed IDRSN, in comparison to 96.64 % and 93.56 % achieved by the DRSN and the wide-kernel deep convolutional neural network respectively. These results highlight the proposed IDRSN's superiority in diagnosing multiple faults under various working conditions, offering a low-cost, highly effective, and applicable approach for complex fault diagnosis tasks.

Enhancing Aircraft Safety through Advanced Engine Health Monitoring with Long Short-Term Memory.

Predictive maintenance holds a crucial role in various industries such as the automotive, aviation and factory automation industries when it comes to expensive engine upkeep. Predicting engine maintenance intervals is vital for devising effective business management strategies, enhancing occupational safety and optimising efficiency. To achieve predictive maintenance, engine sensor data are harnessed to assess the wear and tear of engines. In this research, a Long Short-Term Memory (LSTM) architecture was employed to forecast the remaining lifespan of aircraft engines. The LSTM model was evaluated using the NASA Turbofan Engine Corruption Simulation dataset and its performance was benchmarked against alternative methodologies. The results of these applications demonstrated exceptional outcomes, with the LSTM model achieving the highest classification accuracy at 98.916% and the lowest mean average absolute error at 1.284%.

In Situ Assessment of Oil Quality Sensor Performance in Engine Lubricant Flow

Assessing the functional quality of an engine lubricant through real-time sensing could pave the way for development of comprehensive engine health monitoring systems. In this study, a permittivity-based, commercial off-the-shelf (COTS) oil quality sensor was implemented in the lubricant flow of a diesel engine after detailed evaluation on a benchtop test facility. The sensor was mounted on the oil filter housing of the engine in the post-filter oil flow, and its implementation required no modifications to the engine block. Simultaneously, the lubricant flow was visualized by incorporating a novel test cell in the oil flow path. Both the sensor assembly and the flow visualization cell were fully characterized on the benchtop facility prior to implementation on the engine. In these experiments, fresh and used samples of the engine’s recommended oil were tested, and the sensor’s oil quality measurements showed noticeable differences between the engine and benchtop studies, a feature attributable to the observed presence of aeration intrinsic to the engine oil flow. These results prove that the adaptation of permittivity-based sensors for effective real-time engine lubricant quality monitoring will require comparative assessment of oil quality measurements in aerated and nonaerated flow fields.

Virginia Tech Optical Inlet Sensor for Particle Detection: Rolls Royce M250 Turboshaft Demonstration

Abstract Propulsion systems are exposed to environmental ingestion hazards that can cause significant damage and decrease performance. Particles are ingested in a wide range of flight environments that can cause immediate engine failure or long-term damage. An accurate measurement technique has been developed to quantify particle ingestion and aid engine health monitoring. This sensor utilizes scattering and extinction techniques along with machine learning models to measure particle characteristics based on a robust and versatile library. The capabilities of this sensor have been demonstrated using solid quartz particles on the Rolls-Royce M250-C20B particle ingestion turboshaft test engine. To the authors' knowledge, this work presents the first demonstration and validation of optical solid particle sensing in a turbine engine. CSPEC sand (Mil-E-5007C) was ingested for the validation test at two different feed rates using a sand feeder. The sand concentrations were 45 mg/m3 and 22 mg/m3. The sensor outputs the particle characteristics of aspect ratio (AR), size distribution (σ), Sauter mean diameter (D32), and the particle mass flowrate. The Sauter mean diameter and mass flowrate of ingested sand were calculated using the machine learning model outputs and validated by independent measurements. The sensor produced a 0.1 g/min RMS error compared to the validation measurement.

A novel quantum genetic algorithm with the application in LS-SVR

The characteristics of LS-SVR are analyzed. LS-SVR is fitted for modeling small samples and high dimensional data, but the performance of LS-SVR is related to the specific data distribution, the kind of kernel function, its related kernel parameter, and the penalty coefficient. In this paper, the radial basis function is applied as the kernel function of LS-SVR, and the real double-chain coding target gradient quantum genetic algorithm (DCQGA) is applied to optimize the kernel parameter and penalty item coefficient of LS-SVR, then the regression prediction model DCQGALSSVR is proposed. It is of great significance to build an accurate and reliable fault prediction model for the health monitoring and fault diagnosis of liquid rocket engines. The thrust of a liquid rocket engine is an important factor in its health monitoring. By predicting the thrust change value and comparing the predicted value with the engine thrust threshold, it can be predicted whether the engine will fail at a certain time. In this paper, the proposed DCQGALSSVR model is used to model the thrust of a liquid rocket engine. The simulation results show that the average relative error is 0.37% using LSSVR for modeling on 12 test samples, and is 0.3186% using DCQGALSSVR on the same samples. It can be seen that DCQGALSSVR is effective for the health monitoring of liquid rocket engines, so it has a certain promotion significance.

Reliability Assessment of Nuclear Thermal Engine Configuration and Health Monitoring System

Today’s space nuclear technology programs are often confronted with two fundamental challenges early in the project life cycle: 1) development and testing will be more expensive than a non-nuclear alternative, and 2) the consequences of failure will be more severe. As a result, many space nuclear programs have been designed to minimize testing and maximize their probability of success: their reliability. The United States' Nuclear Engine for Rocket Vehicle Applications Program recognized these facts early on, and by 1961 the program’s primary objective set safety and reliability as the overriding considerations. This focus on reliability greatly influenced the engine's design towards minimizing the possible number of catastrophic failures modes. As such, the final configuration heavily relied on duplicate components for redundancy, including duplicate turbopumps. Despite these efforts, at program cancellation in 1972, the non-nuclear subsystem only achieved a mission predicted reliability of 33%. Some of the most significant contributions to rocket engine reliability in the last 50 years have been from advancements in the Health Monitoring System (HMS). Through rigorous instrumentation and control an engine's HMS has the potential to convert over 90% of a system's catastrophic failure modes to a safe shutdown situation. Due to this, many modern rocket engine designs no longer require redundant components and can prioritize performance-enhancing configurations over those that inherently minimize failure modes. Unfortunately, a standard liquid rocket engine HMS will likely not be compatible with a nuclear rocket engine (NRE). The HMS for a liquid rocket engine most often prevents catastrophic failure by shutting off flow to the combustion chamber. This would not work for a NRE, as removing propellant flow to the reactor could result in reactor overheating and meltdown. Maintaining flow to the reactor is often essential for safe NRE operation, such that reliance on an advanced HMS system alone may not be sufficient. This work investigates the feasibility of an NRE HMS by comparing the HMS designs for liquid rocket engines and terrestrial nuclear power plants and evaluates the necessity for redundant components to maximize overall system reliability. Keywords: Nuclear Thermal Propulsion, Nuclear Rocket Engine, Reliability, Fault Prevention, Fault Tolerance, Turbopump

Exploring Prognostic and Diagnostic Techniques for Jet Engine Health Monitoring: A Review of Degradation Mechanisms and Advanced Prediction Strategies

Maintenance is crucial for aircraft engines because of the demanding conditions to which they are exposed during operation. A proper maintenance plan is essential for ensuring safe flights and prolonging the life of the engines. It also plays a major role in managing costs for aeronautical companies. Various forms of degradation can affect different engine components. To optimize cost management, modern maintenance plans utilize diagnostic and prognostic techniques, such as Engine Health Monitoring (EHM), which assesses the health of the engine based on monitored parameters. In recent years, various EHM systems have been developed utilizing computational techniques. These algorithms are often enhanced by utilizing data reduction and noise filtering tools, which help to minimize computational time and efforts, and to improve performance by reducing noise from sensor data. This paper discusses the various mechanisms that lead to the degradation of aircraft engine components and the impact on engine performance. Additionally, it provides an overview of the most commonly used data reduction and diagnostic and prognostic techniques.

Dynamic and high-speed measurement system for blade tip clearance of engine with optical comb

For blade tip clearance measurement, the optical measurement method has the characteristics of high precision, anti-electromagnetic interference, and non-contact measurement. A dynamic and high-speed blade tip clearance measurement system based on time stretch dispersive Fourier transform with the optical comb is proposed and demonstrated. The dynamic experiments are designed to verify the accuracy of the system. The results show that the measuring speed is up to 17.6 MHz and the measurement errors are less than 1 μm under dynamic conditions. The experiments prove that the system has the advantages of high measurement accuracy, high speed, and good repeatability. It can provide a new tool for engine health monitoring.

Feed-Forward Neural Network for health monitoring of a parallel hybrid electric power system

Hybrid engines are becoming more and more widespread. Electric energy instead is a valid help to reduce the environmental impact. In hybrid engines, the number of components is higher and this results in a decrease in reliability. With Engine Health Monitoring (EHM) we mean the set of techniques used to monitor the health status of a system based on the values assumed by some related parameters. Artificial Intelligence (AI) methods are widely used nowadays in this discipline. In this paper, an EHM approach was developed to monitor the health status of some components constituting an hybrid turboshaft. The dynamic model of the hybrid electric power system is described in an accompanying paper. Feed-Forward Neural Network (FFNN) is used as AI tool to built the just cited system. The engine modelled with Simulink, was used to perform a series of steady-state simulations implementing a degradation condition in some selected components. The degradation condition was simulated by changing the value of the Performance Parameters (PPs) related to each of the selected components. The results of the simulation were used to obtain a dataset useful to train the FFNN to predict the values of the same PPs in a degraded case.

Aero engine health monitoring, diagnostics and prognostics for condition-based maintenance: an overview

Abstract Aero engine performance deterioration highly influences its reliability, availability and life cycle. Predictive maintenance is therefore a key figure within Industry 4.0, which guarantees high availability and reduced downtime thus reduced operational costs for both military and civil engines. This leads to maintenance on demand and needs an effective engine health monitoring system. This paper overviews the work carried out on aero engine health monitoring, diagnostic and prognostic techniques based on gas path performance parameters. The inception of performance monitoring and its evolution over time, techniques used to establish a high-quality data base using engine model performance adaptation, and effects of computationally intelligent techniques on promoting the implementation of engine fault diagnosis are reviewed. Generating dependable information about the health condition of the engine is therefore a requisite for a successful implementation of condition-based maintenance. Based on this study, further research can be attempted to predict residual life of critical components using degradation pattern from aero engine performance data bank which will be an invaluable asset for engine designers as well as for operators.

Aero engine health monitoring, diagnostics and prognostics for condition-based maintenance: an overview

Abstract Aero engine performance deterioration highly influences its reliability, availability and life cycle. Predictive maintenance is therefore a key figure within Industry 4.0, which guarantees high availability and reduced downtime thus reduced operational costs for both military and civil engines. This leads to maintenance on demand and needs an effective engine health monitoring system. This paper overviews the work carried out on aero engine health monitoring, diagnostic and prognostic techniques based on gas path performance parameters. The inception of performance monitoring and its evolution over time, techniques used to establish a high-quality data base using engine model performance adaptation, and effects of computationally intelligent techniques on promoting the implementation of engine fault diagnosis are reviewed. Generating dependable information about the health condition of the engine is therefore a requisite for a successful implementation of condition-based maintenance. Based on this study, further research can be attempted to predict residual life of critical components using degradation pattern from aero engine performance data bank which will be an invaluable asset for engine designers as well as for operators.

Intelligent Combined Neural Network and Kernel Principal Component Analysis Tool for Engine Health Monitoring Purposes

An efficient maintenance plan is an important aspect for aeronautical companies to increase flight safety and decrease costs. Modern technologies are widely used in the Engine Health Monitoring (EHM) discipline to develop intelligent tools capable of monitoring the health status of engines. In this work, Artificial Neural Networks (ANNs) and in-detail Feed-Forward Neural Networks (FFNNs) were exploited in addition to a Kernel Principal Component Analysis (KPCA) to design an intelligent diagnostic tool capable of predicting the Performance Parameters (PPs) of the main components used as their health index. For this purpose, appropriate datasets containing information about degraded engines were generated using the Gas Turbine Simulation Program (GSP). Finally, the original datasets and the reduced datasets obtained after the application of KPCA to the original datasets were both used in the training and testing process of neural networks, and results were compared. The goal was to obtain a reliable intelligent tool useful for diagnostic purposes. The study showed that the degraded component detection and estimation of its performance achieved by using the hybrid KPCA–FFNNs were predicted with accurate and reliable performance, as demonstrated through detailed quantitative confusion matrix analysis.

Development of a combined Artificial Neural Network and Principal Component Analysis technique for Engine Health Monitoring

In aerospace sector, reliability is a crucial point. Modern technologies widely use Artificial Intelligence (AI) algorithms together with detections by sensors in order to design a health-based maintenance plan which stops an aircraft only when needed. In this work, an Engine Health Monitoring (EHM) system was developed by exploiting AI algorithms as Artificial Neural Networks (ANNs) trained to estimate a series of Performance Parameters (PPs) used as index of the health status of the main components constituting an engine. A neural network called Feed-Forward Neural Network (FFNN) in combination with a Principal Component Analysis (PCA) for feature reduction was used in this paper. The software Gas turbine Simulation Program (GSP) was used to generate a series of data containing information about engine performance under different flight conditions and compressor degradation levels. The datasets were subsequently used to train the neural networks to estimate the PPs of the degraded component. The final purpose of the present work is to develop an efficient diagnostic system useful to increase flight safety and decrease maintenance costs and fuel consumption.

Development of big data lean optimisation using different control mode for Gas Turbine engine health monitoring

In the era of big data lean optimisation, a technologically advanced tool for Gas Turbine (GT) performance analysis is necessary. This need is due to big data taken every 5 min of about 4-years, totalling about 6 million data points (1.5×4 data points) for a single unit of GE 7FA engine of 211MW capacity. The use of actual engine data to quantify degradation using a time-base is limited and challenging. This restraint is due to the influence of varying ambient conditions, compressor inlet conditions with different power settings and loads. This study presents a software development kit (SDK) for real-time processing and evaluation of engine data using different control modes with a statistical method of filtering the data using the inlet guide vane (IGV) opening. The study also examines the analysis of GT health monitoring using time-based machine-generated data. The output of this investigation includes obtaining an average degradation trend line for the power output and heat rate increase as a function of time that serves as the basis for the investigation of the economic analysis. Implementing the SDK has reduced the labour cost, enhanced the accuracy of the analysis, and improved the efficiency of service delivery. The validation of the result with the available literature indicated a fair deviation of 1.4%. The utilisation of data for fuel flow recorded for the standard and extended correction was 24.3% and 16.3% at a range of 9.06–9.43 kg/s and 8.95–9.31 kg/s, respectively. The average percentage of fouling degradation level amounted to a 7.2% reduction of power output and a 1.6% increase in heat rate for one year of operation.

Optimized age dependent clustering algorithm for prognosis: A case study on gas turbines

This paper proposes an Age Dependent Clustering (ADC) structure to be used for prognostics. To achieve this aim, a step-by-step methodology is introduced, that includes clustering, reproduction, mapping and finally estimation of Remaining Useful Life (RUL). In the mapping step, neural fitting tool is used. Considering age based clustering concept, determination of main elements of the ADC model is discussed. Genetic algorithm (GA) is used to find the elements of the optimal model. Lastly, fuzzy technique is applied to modify the clustering. The efficacy of the proposed method is demonstrated with a case study on the health monitoring of some turbofan engines. The results show that the concept of clustering even without optimization processes is efficient even for the simplest form of performance. However, by optimizing structure elements and fuzzy clustering, the prognosis accuracy increases up to 71%. The effectiveness of age dependent clustering in prognosis is proven in comparison with other methods.

Discrimination of Rapid and Gradual Deterioration for an Enhanced Gas Turbine Life-cycle Monitoring and Diagnostics

Advanced engine health monitoring and diagnostic systems greatly benefit users helping them avoid potentially expensive and time-consuming repairs by proactively identifying shifts in engine performance trends and proposing optimal maintenance decisions. Engine health deterioration can manifest itself in terms of rapid and gradual performance deviations. The former is due to a fault event that results in a short-term performance shift and is usually concentrated in a single component. Whereas the latter implies a gradual performance loss that develops slowly and simultaneously in all engine components over their lifetime due to wear and tear. An effective engine life-cycle monitoring and diagnostic system is therefore required to be capable of discriminating these two deterioration mechanisms followed by isolating and identifying the rapid fault accurately. In the proposed solution, this diagnostic problem is addressed through a combination of adaptive gas path analysis and artificial neural networks. The gas path analysis is applied to predict performance trends in the form of isentropic efficiency and flow capacity residuals that provide preliminary information about the deterioration type. Sets of neural network modules are trained to filter out noise in the measurements, discriminate rapid and gradual faults, and identify the nature of the root cause, in an integrated manner with the gas path analysis. The performance of the proposed integrated method has been demonstrated and validated based on performance data obtained from a three-shaft turbofan engine. The improvement achieved by the combined approach over the gas path analysis technique alone would strengthen the relevance and long-term impact of our proposed method in the gas turbine industry.

Turbojet Engine Parameters Calculation Based on Fuel Flow and Exhaust Gas Temperature

<div class="section abstract"><div class="htmlview paragraph">The aircraft jet engine is one of the most complex multivariable systems with multiple inputs and multiple outputs. To attempt to optimize control functions or to address diagnostic problems, a detailed knowledge of all jet engine design parameters and performances is required. Although jet engines have been around for almost a century, there are only a few companies in the world presently designing and manufacturing them; as such these companies possess detailed knowledge of all relevant design characteristics and performance parameters. In the event where jet engine technical details are unknown, or only a few of them are known from manufacturer’s catalogues, the challenge becomes how to calculate and extrapolate critical performance parameters based on only fuel flow, jet exhaust temperature and total thrust. The brief engineering concept of calculation steps presented herein has the goal of identifying unknown turbojet engine parameters and design characteristics based on limited measurements and test data obtained at static conditions and single operating point. Test data included steady state fuel flow, turbine gas exit temperature and total jet engine thrust. For the purpose of jet engine parameters estimate and identification, the Rolls-Royce VIPER 11 Mk 22-6 jet engine was used. Calculated and estimated jet engine design parameters were evaluated and compared with measured parameters under the same initial conditions. The results show that calculation and identification processes are accurate with an overall error of less than 1 percent. The presented numeric algorithm combined with off design parameters calculation, can be used in jet engine health monitoring and fault detection processes.</div></div>

Aero-Engine Health Monitoring with Real Flight Data Using Whale Optimization Algorithm Based Artificial Neural Network Technique

Health monitoring of an aero-engine assumes importance in the light of primary requirements of flight safety and reliability. This paper proposes a novel, simple method for monitoring aircraft engine health using Whale Optimization Algorithm based Artificial Neural Network (WOANN) technique, for analyzing the data downloaded from the health and usage monitoring system (HUMS) of a military aircraft. The actual engine data recorded during 47 different flights of eight different engines (of the same type) have been considered in this work. Thirteen engine parameters have been used to determine and monitor the health of the engine. The efficiency of the WOANN technique for engine health monitoring, is compared with that of three other common machine learning algorithms: Probabilistic based Neural Network (PNN), K-Nearest Neighbour (KNN), and Back propagation based Artificial Neural Network (BPANN). The results show that WOANN algorithm classifies and predicts engine health far more accurately as compared to PNN, KNN and BPANN. The values obtained for the metrics of Accuracy, Error, False Positive Rate, F1 score, Mathews Correlation Coefficient, Specificity, Kappa coefficient are found to be the best for WOANN algorithm. The WOANN achieved overall prediction accuracy of 95%, thus presenting itself as a very useful tool for day-to-day monitoring of aircraft engine health using the data downloaded from the aircraft’s HUMS.

In-cylinder pressure reconstruction by engine acoustic emission

The importance of the in-cylinder pressure transducer has been proven in revealing the information about combustion and exhaust pollution formation, as well as for its capability to classify knock. Due to their high price, they are not used commercially for engine health monitoring, which is of significant importance. Hence, this study will investigate the reconstruction of the in-cylinder pressure trace using a structure-borne acoustic emission (AE) sensor, which are relatively low cost sensors. As shown in the literature, AE indicators show a strong correlation with in-cylinder pressure parameters in both time and crank angle domain. In this study, to avoid the effect of engine speed fluctuations, the reconstruction is done in the crank angle domain by means of the Hilbert transform of AE. Complex cepstrum signal processing analysis with a feed-forward neural network is used to generate a reconstruction regime. Furthermore, the reconstructed signals are used to determine some of the important in-cylinder parameters such as peak pressure (PP), peak pressure timing (PPT), indicated mean effective pressure (IMEP) and pressure rise rate. Results showed that the combination of cepstrum analysis with neural network is capable of reconstructing pressure using AE, regardless of engine load, speed and fuel type. The reconstructed pressure can be used to reliably determine PP and PPT. IMEP can be estimated as well in a reasonable range.