Model Predictive Method Research Articles

Suboptimal midcourse guidance design using generalized model predictive spread control

A new generalized model predictive spread control technique is presented for the midcourse guidance of interceptors that are designed to intercept high-speed ballistic missile targets. Because of using the basis functions, this new technique is further computationally efficient over the model predictive static programming technique. Also, the smoothness of the control variable is guaranteed for the smooth basis functions. For demonstrating the performance of the proposed technique, an interceptor midcourse guidance problem with an angle constraint is formulated and solved to intercept an incoming ballistic missile target successfully. Additionally, the results are compared with those of the midcourse guidance design using the model predictive static programming technique. A comparative study of the new technique has also been conducted with the quasi-spectral model predictive static programming technique proposed earlier in the literature. It has been observed that the orthogonality of the basis functions is a necessary assumption and without that, the quasi-spectral model predictive static programming technique is not a near-optimal technique. By using the new technique based on Legendre basis functions, the solution converges to the model predictive static programming method solution by increasing the number of basis functions with less computational load.

Research on SOC balance strategy of STEKF battery based on MMC-BESS model

Under the condition of the battery current fluctuation and the uncertain initial value of the battery SOC in the MMC(Modular Multilevel Converter) battery energy storage system, there are relatively big errors in the prediction of SOC estimation values and there are poor results of the battery SOC balance, by using the ampere-hour integration method and the open-circuit voltage method. The method of model prediction was adopted to study the SOC balance strategy. Firstly, the power transfer function between each port is derived and the second-order Thevenin equivalent nonlinear model of the battery is established. On this foundation, the more accurate estimation results of the battery SOC are obtained employing the strong tracking extended Kalman filter algorithm, under the condition of the large fluctuation of the battery current and the uncertainty initial value of the battery SOC in the MMC-BESS model. In the end of the paper, the validity of the method introduced in the paper way was confirmed according to the simulation results of the simulation calculation based on the power transfer relationship between the battery SOC and each port, using the three-level SOC balance strategy to balance the battery SOC.

Robust proportional incremental nonlinear dynamic inversion control of a flying-wing tailsitter

Tailsitter unmanned aerial vehicles are utilized extensively nowadays since they merge advantages of both fixed-wing unmanned aerial vehicles and rotary-wing unmanned aerial vehicles. However, their attitude control suffers from unknown nonlinearities and disturbances due to the wide flight envelope. To solve the problems, a robust attitude controller based on a newly designed flying-wing tailsitter is proposed in this paper. By employing the angular acceleration feedback to compensate unmodeled dynamics, the proportional incremental nonlinear dynamic inversion control law is first developed. The proportional incremental nonlinear dynamic inversion strengthens the conventional nominal gain incremental nonlinear dynamic inversion with a proportional term to reflect the change of the angular acceleration more directly. Therefore, the tailsitter has a quicker response and performs better in suppressing model uncertainties and external disturbances. Since the angular acceleration is difficult to measure in practice, an angular acceleration estimation method is then established to provide accurate signals for the proportional incremental nonlinear dynamic inversion. The signals are derived as complementary results of model prediction method and direct differential method. Theoretical analysis and systematic simulations are conducted to corroborate the effectiveness of the developed estimation method as well as the robustness of the proposed controller.

A Highly Reliable Propulsion System with Onboard Uninterruptible Power Supply for Train Application: Topology and Control

Providing uninterrupted electricity service aboard the urban trains is of vital importance not only for reliable signaling and accurate traffic management but also for ensuring the safety of passengers and supplying emergency equipment such as lighting and signage systems. Hence, to alleviate power shortages caused by power transmission failures while the uninterruptible power supplies installed in the railway stations are not available, this paper suggests an innovative traction drive topology which is equipped by an onboard hybrid energy storage system for railway vehicles. Besides, to limit currents magnitudes and voltages variations of the feeder during train acceleration and to recuperate braking energy during train deceleration, an energy management strategy is presented. Moreover, a new optimal model predictive method is developed to control the currents of converters and storages as well as the speeds of the two open-end-windings permanent-magnet-synchronous-machines in the intended modular drive, under their constraints. Although to improve control dynamic performance, the control laws are designed as a set of piecewise affine functions from the control signals based on an offline procedure, the controller can still withstand real-time non-measurable disturbances. The effectiveness of proposed multifunctional propulsion topology and the feasibility of the designed controller are demonstrated by simulation and experimental results.

A model predictive scheme to integral controller design for uncertain LTI systems with nonzero reference

A model predictive method is developed to tune an integral controller for uncertain systems subjected to constrained input signals. For this purpose, a stabilizing integral controller is firstly designed for linear time-invariant (LTI) systems with polytopic uncertainty. The integral controller gains can be determined via a feasible solution of a linear matrix inequality (LMI). Then, a predictive control is incorporated into the integral controller synthesis through an optimization problem subjected to some LMI constraints. The suggested control is successfully applied to a typical uncertain system and an uncertain chemical reactor. The effectiveness of the proposed technique will be shown in comparison with other control methods.

Multi-UAV Formation Control Based on a Novel Back-Stepping Approach

This paper addressed a well-documented open problem regarding multiple unmanned aerial vehicles (multi-UAVs) formation keeping. A cooperative guidance control method based on the back-stepping approach is proposed to rapidly form the desired formation and reach the multi-UAV steady state. The UAV formation system consists of four UAVs, forming a regular triangle formation. One of the UAVs is a virtual leader and is located at the center of the triangle; the other three UAVs are located at the vertices of the triangle. The forward speed of leader is used as the forward direction of the formation, and the followers follow the leader in a formation flight. The error dynamics model of each follower is established by using the leader guidance mechanism, and the communication mode between any two UAVs is established based on graph theory. In addition, the guidance control law is obtained via the back-stepping approach. We reasonably construct the Lyapunov function to prove the stability of the proposed cooperative guidance control law in formation aggregation and keeping. Moreover, the proposed method is compared with the model prediction method (MPC) and the Lyapunov method to further verify the effectiveness of the proposed method. The simulation results show that each UAV can converge to the desired motion trajectory and fly in the desired formation with fast convergence speed and small steady-state error.

Research on distributed photovoltaic power generation prediction based on grey model for energy Internet of city

Power load forecasting is one of the basic tasks of power system dispatching coordination and rational arrangement of energy production. Reasonable power load forecasting can ensure safe, continuous and stable energy supply, and provide reasonable guarantee for the safe and economic operation of power system. Based on Beijing’s new energy and renewable energy online monitoring system, this paper collects power generation data, environmental data and equipment working condition data, and uses gray model prediction method to predict the load of XXXXX photovoltaic power station, and obtain more accurate power load data, which is complex fluctuation. The photovoltaic power generation project provides the basis for post-generation planning.

Bayesian model based on Markov chain Monte Carlo for identifying mine water sources in Submarine Gold Mining

Based on the Bayesian theory, the Bayesian water source discrimination model and model prediction method were established by using the adaptive differential evolution Metropolis algorithm of Markov chain Monte Carlo simulation as the parameter posterior distribution sampling calculation method. Taking the mine water samples of Sanshandao Gold Mine as an example, according to the chemical composition analysis of the water samples from different monitoring points, six indexes of Mg2+, Na+ + K+, Ca2+, SO42−, Cl− and HCO3− were selected as the discrimination factors. According to the water characteristics of each aquifer and the actual needs of discrimination, the water inrush sources in the mining area were divided into four categories: sea water is class I, Quaternary pore water is class II, bedrock fissure water is class III, and fresh water is class IV. Taking 40 typical water inrush samples as training samples, using SPSS statistics and Bayes method based on Markov Chain Monte Carlo to build Bayes discriminant analysis model, the posterior distribution of algorithm estimation based on water sample information was obtained, and the analysis method of mine water damage source was obtained. The results showed that the Bayes discriminant model based on Markov Chain Monte Carlo has high accuracy, good applicability and discriminant ability, and has certain guiding significance for the prevention and control of water damage and the related field work.

A Novel Current Controller for Grid-Connected Voltage-Source-Inverters

In this article, a new open-loop current controller based on model prediction method for grid-connected voltage-source-inverters (VSIs) is proposed, which consists of two proportional factors and a delay part (Proportional-Proportional-Delay, PPD). First, the working principle of the PPD controller is explained from the combined step response perspective based on a L–R load, and parameter constraints of PPD controller are derived and discussed from the time domain and differential-integral transform perspectives, respectively. After that, parameter design rules of PPD controller are given by considering the constraint of switching frequency. Due to the sensitivity of the model prediction method to nonideal factors, such as voltage drops of semiconductor devices, sampling, and driving delays, etc., influences of these nonideal factors on performance of PPD controller are discussed in this article, and corresponding compensation strategies are given as well. Finally, the effectiveness of proposed PPD controller and its compensation strategy against nonideal factors are verified by experimental results from a 3 kW prototype in laboratory.

Data-Driven Model Predictive Monitoring for Dynamic Processes

Process monitoring plays an important role in maintaining favorable process operation conditions and is gaining increasing attention in both academic community and industrial applications. This paper proposes a data-driven model predictive fault detection method to achieve efficient monitoring of dynamic processes. First, a measurement sample is projected into a dominant latent variable subspace that captures main variance of the process data and a residual subspace. Then the dominant latent variable subspace is further decomposed as a dynamic feature subspace and a static feature subspace. A fault detection residual is generated in each subspace, and corresponding monitoring statistic is established. By using the model predictive monitoring scheme, not only the status of a process but also the type of a detected fault, namely a dynamic feature fault or a static feature fault, can be identified. Effectiveness of the proposed data-driven model predictive monitoring scheme is tested on a lab-scale distillation process.

Forecast of short-term wind power based on a novel hybrid method

Short-term wind power forecasting plays an important role in power generation, because it prevents the power system operation from its uncertain and intermittent nature. This article proposes a novel method for short-term wind power forecasting, which combines the wavelet transform, particle swarm optimization dynamic gray model and Lyapunov exponent prediction method. First, the approach decomposes the wind power curve into the high-frequency and low-frequency curves by wavelet transform, which represent the detail and tendency signals, respectively. Then, we use the proposed particle swarm optimization dynamic gray model to forecast the low-frequency curve with its smooth and periodic outline. Moreover, Lyapunov exponent prediction method is used to predict high-frequency curves, which possess the chaos characteristics. Finally, we obtain the wind power forecasting result from the combination of the predicted low and high frequencies. The experiment of four seasons in an US wind farm validates that the proposed method is effective in solving the short-term wind power forecasting problem. The obtained results, discussed comprehensively, show that the hybrid method has better prediction accuracy than the other methods, such as artificial neural network, persistence, and autoregressive integrated moving average model, with the lowest average mean absolute percentage error is 8.07% and the average root mean square error is 0.8164 over four seasons.

A new classified mixed model predictor

We develop a new method for classified mixed model prediction (CMMP). The original CMMP method (Jiang et al., 2018b) does not incorporate covariate information in matching the class between the new observations and the training data. As a result, the method may not outperform the mixed model prediction (MMP) method in terms of predictive performance. The new CMMP method that we develop utilizes covariate information, and therefore is more accurate in terms of the matching. We show that the new CMMP method outperforms the MMP in terms of the predictive performance. Furthermore, we develop a second-order unbiased estimator of the mean squared prediction error (MSPE) for the new CMMP, which was previously not available for the original CMMP. Theoretical and empirical properties of the proposed new CMMP method as well as the MSPE estimator are studied. A real data application is considered.

Short Term Prediction of Photovoltaic Power Based on FCM and CG-DBN Combination

Affected by many factors, the photovoltaic output power is characterized by nonlinearity, volatility and instability. Therefore, short-term forecasting models are required to have multiple inputs, levels, and categories. In order to solve the above problems and improve the accuracy of predictions, this paper proposes a combined model prediction method based on similar-day clustering and the use of Conjugate Gradient (CG) to improve Deep Belief Network (DBN). The initial method uses fuzzy C-Means Clustering Algorithm (FCM) to perform similar-day clustering on the original data according to the degree of membership. The CG-DBN prediction model is then designed according to the category, with the model ultimately being used to perform the short-term prediction of the PV output power. The proposed scheme uses data from Zhejiang Longyou power station for experimental analysis and verification, and the results were compared with the back propagation neural networks model, Support Vector Machine (SVM) model, and traditional deep belief network. The model’s predicted results are compared. Finally, it is concluded that, in the short-term PV power load forecasting, the prediction performance of the FCM and CG-DBN combination forecast model is better than the above three models and has strong feasibility in short-term PV power forecasting.

Controller Design for Electrical Drives by Deep Reinforcement Learning: A Proof of Concept

This article presents an approach to the controller design for electrical drives, which makes use of methods of deep reinforcement learning. Conventional control methods dominated the field for a long time, since they usually lead to control solutions with very robust and steady results. Yet, it often can be found that the overall control performance heavily correlates with the experience and education of the developing engineer. Moreover, conventional methods strongly depend on the available knowledge of the control system (e.g., plant model accuracy), which often causes the necessity for thorough identification methods. Real-time capability issues are also a present problem of sophisticated control approaches, such as model-predictive methods. Especially, in the domain of electrical drive train control, solving elaborate online optimization problems may be critical when very small plant time constants have to be considered. The methods of deep reinforcement learning will not only enable to acquire a suitable controller structure, but, moreover, the procedure will tune itself, which will allow for a more abstract level of investigation. This article presents a first proof of concept by means of controlling the phase currents of a permanent magnet synchronous motor in a field-oriented framework. The results found are promising and motivate further research in this field.

Deep Visual MPC-Policy Learning for Navigation

Humans can routinely follow a trajectory defined by a list of images/landmarks. However, traditional robot navigation methods require accurate mapping of the environment, localization, and planning. Moreover, these methods are sensitive to subtle changes in the environment. In this letter, we propose PoliNet, a deep visual model predictive control-policy learning method that can perform visual navigation while avoiding collisions with unseen objects on the navigation path. PoliNet takes in as input a visual trajectory and 360 $^{\circ }$ images from robot's current view and outputs velocity commands for a planning horizon of $N$ steps that optimally balance between trajectory following and obstacle avoidance. PoliNet is trained using a differentiable neural image predictive model and a traversability estimation model in an model predictive control setup, with minimal human supervision. PoliNet can be applied to visual trajectory in new scenes without retraining. We show experimentally that the robot can follow a visual trajectory even if it does not start from the exact same position and in the presence of previously unseen obstacles. We validated our algorithm with tests both in a realistic simulation environment and in the real world outperforming state-of-the-art baselines under similar conditions in success rate, coverage rate of the trajectory, and with lower computational load. We also show that we can generate visual trajectory in simulation and execute the corresponding path in the real environment.

Coal seam development characteristics and distribution predictions in marginal sea basins: Oligocene Yacheng Formation coal measures, Qiongdongnan Basin, northern region of the South China Sea

Coal measures located in marginal sea basins are important hydrocarbon source rocks. For the purpose of effectively guiding future oil and gas exploration, the characteristics and distribution patterns of coal seams in coal measures of a marginal sea basin are systematically outlined. Coal measures in marginal sea basins can have large thicknesses, but the individual coal seams can be very thin and lack lateral continuity. In the study area, the organic micro-components of the coal are dominated by vitrinite, with very low amounts of inertinite and liptinite. The amount of inorganic microscopic components is large, but with limited drilling results, few cores and thin coal seams, which are easily overlooked during logging activities, a comprehensive analysis of the logging data may improve efficiency in coal-seam identification and thickness determination. The development and distribution of coal seams in marginal sea basins are controlled by various factors, including (1) paleoclimatic and paleobotanic conditions, which could fundamentally limit coal formation, (2) coal-forming sedimentary environmental conditions that may limit the scope of coal-seam development from a macroscopic perspective, and (3) paleotectonic and paleotopographic conditions that define the coal-forming structures. Therefore, the descending and rising cycles of base-levels, along with changes in the growth rates of the accommodation spaces, can be used to determine the horizons that are potentially favourable for coal formation and can also indicate the migration trends of coal-forming environments on the structural plane. Seismic wave impedance inversion methods could be utilised for semi-quantitative assistance for prediction of coal seams. In summary, for models of coal-seam development in marginal sea basins, the grades should be divided according to reliability, and the different reliability levels should be predicted separately.The characteristics of coal seams developed in marginal sea basins are described.The macerals of coals developed in marginal sea basins have been ascertained.A development model and distribution prediction method for coal seams are assessed according to the control factors.A model for the prediction of coal-seam distribution is presented.

Adaptive energy management strategy for hybrid batteries/supercapacitors electrical vehicle based on model prediction control

AbstractIn this paper, in order to extend battery lifespan and lift power performance in hybrid batteries/supercapacitors electrical vehicles, a new energy management strategy is proposed based on model prediction control and adaptive method. Firstly, models of batteries and supercapacitors are built, which are then adapted and simplified to develop state space expression. Secondly, a series of reference values of battery power and supercapacitor state of charge for model prediction methods are properly calculated by the adaptive method. Thereafter, an energy management strategy based on the model prediction control method is designed to allocate the output power of batteries and supercapacitors within constraints, which guarantees batteries lifespan and the power performance of vehicle. Finally, simulation and experiment results are provided to evaluate battery lifespan and power performance of vehicles under HWFET and UDDS road conditions. The results obtained show that the proposed strategy, compared with the former methods, reduce average power and power variation of batteries, and effectively utilize supercapacitors depending on the power demand, which can extend battery lifespan and lift the power performance of vehicle.

Hybrid Model Predictive Control Strategy of Supercapacitor Energy Storage System Based on Double Active Bridge

In order to solve the problem of which the dynamic response of a supercapacitor (SC) is limited due to the mismatch dynamic characteristics between the DC/DC converter and supercapacitor in an energy storage system, this paper proposes a hybrid model predictive control strategy based on a dual active bridge (DAB). The hybrid model predictive control model considers the supercapacitor and DAB in a unified way, including the equivalent series resistance and capacitance parameters of the SC. The method can obtain a large charging and discharging current of the SC, thereby not only improving the overall response speed of the system, but also expanding the actual capacity utilization range of the SC. The simulation results show that compared with the model prediction method of the dual active bridge converter, the proposed control method can effectively improve the overall response speed of the system, which can be improved by at least 0.4 ms. In addition, the proposed method increases the actual upper limit of the SC voltage, reduces the actual lower limit of the SC voltage, and then expands the actual capacity utilization range of the SC by 18.63%. The proposed method has good application prospects in improving the dynamic response performance of energy storage systems.

A model predictive control-based hybrid MPPT method for boost converters

ABSTRACTIn this study, Perturb and Observe Maximum Power Point Tracking algorithm and Model Predictive Control (MPC) algorithm are combined to improve the dynamic performance of the control structure of boost converters used in renewable energy systems. For this purpose, a digital control method that does not need any comparator is developed. The maximum power point current value is determined from the Maximum Power Point Tracking (MPPT) algorithm and it is used as the reference for current control operation in the MPC algorithm. Dynamic capability and tracking performance of the proposed control method is validated by simulation and experimentally. The dynamic behaviour of the control algorithm is analysed by carrying out step change tests. Results show that the control method has excellent dynamic performance thanks to its feature of reference tracking in only one switching period. Furthermore, overshoots and undershoots are successfully eliminated.

Model Predictive Trajectory Planning for Automated Driving

In order to enable automated driving systems on the road, several key challenges need to be solved. One of these issues is real-time maneuver decision and trajectory planning. This paper introduces a general framework for maneuver and trajectory planning with model predictive methods. It discusses several representations of this framework in distinct complexity levels. In general, sophisticated models require to nonquadratic objective functions with nonlinear constraints, leading to increased computational complexities during calculation. Yet, oversimplified models can neither cope with vehicle dynamics in critical maneuvers nor do they represent complex traffic scenes with several maneuver options appropriately. A scheme to partition the trajectory space into homotopy regions is proposed. In each homotopy class, linearization about a trajectory from this class is applied. Demonstrations by simulation and with experimental vehicles show the capability of the proposed method in selecting optimal maneuvers and trajectories. This is even valid during extreme maneuvers, such as in last moment collision avoidance.