Equivalent circuits are one of the most used models for Li-ion cells in the automotive area. However, it is a challenge to these models to be able to capture the cell discharge capacity under different loads, while still being accurate on both continuous charge and dynamic tests, fast to compute, and easy to parametrize from non-specialized data. To tackle this challenge, this paper proposes an extension of the nonlinear double capacitor model by increasing its order, parameter dependency with C-rate, and an identification procedure that exploits the pseudo-linear nature of the problem to find the parameter maps. An analogy between the parts of the circuit and the single particle model is also presented to reduce the search space of the identification algorithm and to enhance model interpretability. The performance of the proposed model extension is analyzed and compared to a state-of-the-art model on a challenging LiFePO4 dataset with different characteristics and validated on a realistic drive cycle, obtaining a mean absolute average error of around 20 mV for both training and validation tests.

Background: The phenomenon of calendar ageing continues to have an impact on battery systems worldwide by causing them to have undesirable operation life and performance. Predicting the degradation in the capacity can identify whether this phenomenon is occurring for a cell and pave the way for placing mechanisms that can circumvent this behaviour. Methods: In this study, the machine learning algorithms, Extreme Gradient Boosting (XGBoost) and artificial neural network (ANN) have been used to predict the calendar ageing data belonging to six types of cell chemistries namely, Lithium Cobalt Oxide, Lithium Iron Phosphate, Lithium Manganese Oxide, Lithium Titanium Oxide, Nickle Cobalt Aluminum Oxide and Nickle Manganese Cobalt Oxide. Results: Prediction results with overall Mean Absolute Percentage Error of 0.0126 have been obtained for XGBoost algorithm. Among these results, Nickle Cobalt Aluminum Oxide and Nickle Manganese Cobalt Oxide type cell chemistries stand out with their mean absolute percentage errors of 0.0035 and 0.0057 respectively. Also, algorithm fitting performance is relatively better for these chemistries at 100% state of charge and 60°C temperature compared to ANN results. ANN algorithm predicts with mean absolute error of approximately 0.0472 overall and 0.0238 and 0.03825 for Nickle Cobalt Aluminum Oxide and Nickle Manganese Cobalt Oxide. The fitting performance of ANN for Nickle Manganese Cobalt Oxide at 100% state of charge and 60°C temperature is especially poor compared to XGBoost. Conclusions: For an electric vehicle battery calendar ageing prediction application, XGBoost can establish itself as the primary choice more easily compared to ANN. The reason is XGBoost’s error rates and fitting performance are more usable for such application especially for Nickel Cobalt Aluminum Oxide and Nickel Manganese Cobalt Oxide chemistries, which are amongst the most demanded cell chemistries for electric vehicle battery packs.

This paper aims to investigate how some features commonly associated with more generic time-series analysis are associated with capacity fade in lithium-ion cells and how they can be used to create simple but effective machine-learning models. This is done by processing the current, voltage, and temperature measurements, which span around two hundred cells for roughly two years, with a popular automated time-series analysis routine that extracts a significant number of different characteristics from the dataset for each signal. The most promising factors associated with the capacity fade are obtained by using a feature selection technique that is simple, quick and does not depend on a specific model structure. An analysis of the most relevant results is done, together with a standard hyperparameter search strategy using bayesian optimization for different classical regression models. With this step-by-step approach, the most promising features were investigated and an average error smaller than 5% was obtained on previously unseen validation data.

In modern Battery Management Systems (BMSs), it is significant to obtain an accurate battery model to estimate the states of the battery such as State of Charge (SoC), State of Health (SoH), State of Power (SoP), State of Safety (SoS) etc. Traditional lithium-ion batteries (LIBs) have some drawbacks in terms of safety and energy density. To overcome these drawbacks, all-solid-state batteries (ASSBs) are being developed as an alternative solution for conventional lithium-ion batteries. The focus of this study is on all-solid-state batteries and their modeling based on the equivalent circuit model. On the other hand, the modeling of a cell needs an immense amount of data and long test duration time. Instead of cell characterization test data, the all-solid-state cell is modeled by using DC internal resistance (DC-IR) information. During this study, two different equivalent circuit models containing series-connected RC pairs with and without ohmic resistance are investigated. In addition, the equivalent circuit model parameters are derived via Genetic Algorithm. Moreover, measured and simulated resistance values are compared with Mean Absolute Error (MAE) criteria for two different equivalent circuit models. Finally, the plausibility of the obtained models are analyzed and compared with experimental Hybrid Pulse Power Characterization (HPPC) test results.

Parameter identification of an electrical battery model is significant for the analysis of the performance of a battery. In order to obtain an accurate electrical battery model, a series of cell characterization tests should be conducted which will take a considerable amount of time. In this study, in order to identify the parameters of the electrical battery model in a short amount of time with an acceptable accuracy, DC-IR data is used. DC-IR test will take less time compared to the cell characterization tests. For the parameter identification, one of the most commonly used evolutionary algorithm (EA), Genetic Algorithm (GA) is used for the curve fitting problem and its performance is compared with the Levenberg-Marquardt algorithm.

In recent years, especially considering environmental awareness and low pollution, economical and safe driving is crucial in automotive applications. In this study, a driving rating method based on interval type-2 fuzzy logic system is proposed. Aggressive and normal driving behaviors are considered. Then, the driving behaviors are analyzed and compared in terms of engine speed, acceleration, accelerator pedal position, engaged gears, and fuel consumption values. The simulation studies indicate that the proposed driving rating method is efficient in rating aggressive and normal driving behaviors. In addition, driving behavior impact on fuel consumption is analyzed.

In this study, a novel approach is described to the design of an interval type-2 fuzzy neural system (IT2 FNS). It differs from the classical IT2 FNS in its use of parameterized conjunctors. In the optimization of the IT2 FNS, the membership functions are kept fixed and only the parameters of the conjunctors and the parameters in the consequent are tuned. In this study, the gradient based learning algorithm is used. The approach is tested for the modeling of a benchmark nonlinear function and for the wheel slip control of a quarter car model (QCM). In the stated applications, in the absence of any expert knowledge, some knowledge about the system is gained by the use of the interval type-2 fuzzy c-means (IT2 FCM) clustering algorithm. Nevertheless, this requires the number of classes to be known beforehand. To alleviate this problem, some validity indices that have been suggested in the literature and a novel validity index that carries less computational burden are considered to determine the number of classes and the number of fuzzy rules. Simulation studies are presented and compared with the results from the literature.

This paper focuses on the design of a novel control approach. Its contribution to the existing literature is that in the design of an interval type-2 fuzzy neural system, recursive fuzzy c-means clustering algorithm is used and the designed algorithm is applied in control applications. The center and the standard deviation values of the interval type-2 Gaussian membership functions at the antecedent part of the Takagi-Sugeno-Kang type fuzzy rules are determined by the use of the recursive fuzzy c-means clustering algorithm. The parameters at the consequent parts are tuned based on the gradient descent approach. The effectiveness of the designed algorithm is tested by simulation studies on a 2-DOF helicopter system and by experimental studies on a real-time servo system. The performance of the proposed method is compared with a traditional neuro-fuzzy structure and an interval type-2 fuzzy neural system, which are both adopted from the literature. The results obtained illustrate the efficacy of the proposed control approach.

In this study, the aim is to track the desired pitch and yaw axis trajectories of a 2-DOF helicopter system. For this purpose, neuro-fuzzy system with parameterized conjunctors is used and its performance is compared with a conventional control approach, namely a PID controller. In neuro-fuzzy methods, in order to obtain an optimal fuzzy model, the most commonly used approach is to tune the parameters of the membership functions at the antecedent part of the fuzzy rules. This adaptation process may lead to loss or distortion of the knowledge that is carried by these membership functions. To alleviate this problem, the parameters of the parameterized conjunctors are tuned instead of the parameters of these membership functions.

Model-free approaches such as Artificial Neural Networks and Fuzzy Controllers are widely used in the control of Antilock Braking System (ABS) due to its strongly nonlinear structure and uncertainties involved. In this paper the design of a Spiking Neural Network (SNN) controller is considered for the regulation of the wheel slip value at its optimum value. For the training of the network a gradient descent based approach is followed. To formulate the generation of a new spike train from the incoming spikes, the Spike Response Model (SRM) is used. Delay coding is utilized to convert real numbers into spike times. The control algorithm is applied to a quarter vehicle model, and it is verified through simulations indicating fast convergence and good performance of the designed controller.

In this study, two fuzzy algorithms, type-1 fuzzy algorithm with parameterized conjunctors and a novel approach interval type-2 fuzzy algorithm with parameterized conjunctors are used in the modeling application for nonlinear functions. The aim of using parameterized conjunctors as fuzzy operators in these algorithms is not to lose or distort the expert knowledge about the system during the optimization process. In this study, this linguistic information about the system is obtained by using fuzzy c-means clustering algorithms. Then, the designed fuzzy algorithms are tested on two benchmark nonlinear functions in modeling application.

In conventional fuzzy modeling and control, to obtain an optimal fuzzy system, a commonly used approach is to tune the parameters of the membership functions. However, if the membership functions carry significant expert knowledge about the system, this may be lost or distorted during the optimization process. In order to prevent such a loss of valuable information, parameterized conjunction operators may be used and their parameters can be tuned instead. In this paper such an approach is adopted to optimize a type-1 fuzzy neural system (FNS), used for slip control of a Quarter Car Model (QCM). The simulation results presented indicate the efficacy of the approach in meeting the desired objectives even under noisy conditions.

The control of nonlinear systems is a challenging task in control engineering and the use of type-1 Fuzzy Logic Controllers (FLCs) has been proposed as a possible approach. However, traditional type-1 FLCs can prove to be inadequate in dynamically unstructured environments that include large amount of uncertainties. Under such circumstances, type-2 fuzzy logic controllers can be a viable alternative. In this paper, an Anti Lock Breaking System (ABS) is controlled both by a type-1 and an interval type-2 fuzzy logic controller with and without noisy input measurement. The performances of both controllers are approximately the same without noise in the input measurement. However, with noisy input measurements, interval type-2 fuzzy logic controller results in better performance, indicating its superiority when there exist considerable amount of uncertainties in the system to be controlled.

Control of nonlinear systems is a challenging task in control engineering and the use of type-2 fuzzy logic controllers (FLCs) has been proposed as a promising approach, as they can perform adequately in dynamically unstructured environments that include large amount of uncertainties. In this paper, an Anti-Lock Braking System (ABS) is controlled by an interval type-2 fuzzy logic controller. The control algorithm is used on a quarter vehicle model and it is seen through simulation studies that it results in a good performance with fast convergence. These results are experimentally validated on a laboratory setup.

A type-2 fuzzy neural system (T2FNS) is proposed in this paper for process control. The structure of the system is presented and the rules for updating its parameters are derived using the gradient algorithm. The effectiveness of the proposed approach is evaluated on a laboratory setup that regulates the speed of a DC motor and the experimental results are compared with those obtained with the use of a type-1 fuzzy neural system (T1FNS). It is seen that T2FNS results in reduced oscillations around the set point in the presence of load disturbances.

In this paper the development of an adaptive neuro-fuzzy architecture for the speed control of a servo system with nonlinear load is presented. The synthesis of the structure is described and a learning algorithm for the neuro-fuzzy control system is derived. The supervised learning algorithm is used to train the unknown coefficients of the system, and then the fuzzy rules of the neuro-fuzzy system are generated. A number of simulation studies are carried out, and the results are compared with those obtained with a PI controller tuned using desired time response characteristics. These and the experimental studies presented show that the neuro-fuzzy control system has a better control performance than the conventional PI controller.

The paper has the goal of comparing the performance of four different approaches to fuzzy modeling, using parameterized conjunctions, a novel concept named Constrained Fuzzy Sets (CFSs), CFSs with parameterized conjunctions, and unnormalized interval type-2 Takagi Sugeno Kang (IT2 TSK). The theoretical and mathematical backgrounds of the four approaches are briefly described and their performances are compared in approximating a nonlinear function.