Predictive Feedforward Control Research Articles

Relative contribution of sensory and motor deficits on grip force control in patients with chronic stroke

ObjectiveThis study aimed to characterize grasping behavior in static (weight-dependent modulation and stability of control) and dynamic (predictive control) aspects specifically focusing on the relative contribution of sensory and motor deficits to grip force control in patients with chronic stroke. MethodsTwenty-four chronic stroke patients performed three manipulative tasks: five trials of 5-s grasp-lift-holding, 30-s static holding, and vertical dynamic/cyclic oscillation of holding the object. ResultsExerted static grip force on the paretic side exhibited statistically greater than that on the non-paretic side. Spearman's rank correlation coefficient revealed that the contribution to static grip force control was larger in sensory deficits than in motor deficits. In addition, the sensory deficit is related to the reduced coupling between grip force and load force, suggesting difficulty in predictive control due to the absence of sensory feedback. ConclusionsGiven that grip force control involves predictive feedforward and online feedback control, the evaluation of grip force might be an important and feasible evaluation manner for the assessment of sensorimotor control in patients post-stroke. SignificanceDetailed evaluation of grip force control would help to understand the mechanisms underlying hand dysfunction in stroke patients.

Predictive Feedforward Control Based on Speed Error for Speed Ripple Suppression in Compressor Motors

The compressor motor of an air conditioner is subjected to the a fluctuating load torque during the suction and discharge of refrigerant, resulting in a speed ripple in the machine angular cycle. This paper proposes a new predictive feedforward control system that predicts and suppresses the speed ripple cuased by the periodic disturbance of compressor motor by using error data of speed PI controller. The speed ripple suppression performance of proposed control method is experimentally confirmed by applying it to achieve speed control of an air conditioner compressor motor.

Predictive Control of a Heaving Compensation System Based on Machine Learning Prediction Algorithm

Floating structures have become a major part of offshore structure communities as offshore engineering moves from shallow waters to deeper ones. Floating installation ships or platforms are widely used in these engineering operations. Unexpected wave-induced motions affect floating structures, especially in harsh sea conditions. Horizontal motions on the sea surface can be offset by a dynamic positioning system, and heave motions can be controlled by a heave compensation system. Active heave compensation (AHC) systems are applied to control vertical heave motions and improve safety and efficiency. Predictive control based on machine learning prediction algorithms further improves the performance of active heave compensation control systems. This study proposes a predictive control strategy for an active heave compensation system with a machine learning prediction algorithm to minimise the heave motion of crane payload. A predictive active compensation model is presented to verify the proposed predictive control strategy, and proportion–integration–differentiation control with predictive control is adopted. The reliability of back propagation neural network (BPNN) and long short-term memory recurrent neural network (LSTM RNN) prediction algorithms is proven. The influence of the predictive error on compensation performance is analysed by comparing predictive feedforward cases with actual-data feedforward cases. Predictive feedforward control with regular and irregular wave conditions is discussed, and the possible strategies are examined. After implementing the proposed predictive control strategy based on a machine learning algorithm in an active heave compensation system, the heave motion of the payload is reduced considerably. This investigation is expected to contribute to the motion control strategy of floating structures.

The influence of stimulus and behavioral histories on predictive control of smooth pursuit eye movements

The smooth pursuit system has the ability to perform predictive feedforward control of eye movements. This study attempted to examine how stimulus and behavioral histories of past trials affect the control of predictive pursuit of target motion with randomized velocities. We used sequential ramp stimuli where the rightward velocity was fixed at 16 deg/s while the leftward velocity was either fixed (predictable) at one of seven velocities (4, 8, 12, 16, 20, 24, or 28 deg/s) or randomized (unpredictable). As a result, predictive pursuit responses were observed not only in the predictable condition but also in the unpredictable condition. Linear mixed-effects (LME) models showed that both stimulus and behavioral histories of the previous two or three trials influenced the predictive pursuit responses in the unpredictable condition. Intriguingly, the goodness of fit of the LME model was improved when both historical effects were fitted simultaneously rather than when each type of historical data was fitted alone. Our results suggest that predictive pursuit systems allow us to track randomized target motion using weighted averaging of the information of target velocity (stimulus) and motor output (behavior) in past time sequences.

Model Predictive Feedforward Control for High-Dynamic Speed Control of Combustion Engine Test Beds

Hardware-in-the-Loop tests have become a key factor in decreasing development time of various drive systems. In this context, high performance test beds, where feedback control plays a major role, are necessary to meet the demanding requirements such as real driving emissions in the automotive industry. Particularly for combustion engine test beds, periodic combustion strokes cause large oscillations in different measurement signals on engine test beds, which causes severe problems since conventional controllers try to compensate for these periodic disturbances. The control settings are a trade-off between fast reference tracking and undesirably strong disturbance rejection, which may compromise test results. An ordinary two-degree-of-freedom (2-DOF) control with flatness-based feedforward control produces relief in separating reference and disturbance behavior but cannot handle input constraints. Thus, a model predictive feedforward controller (MPFFC) in combination with the 2-DOF strategy is introduced for high-dynamic speed control in this contribution. The MPFFC generates optimal input and output trajectories, which respect constraints of the system. Fast reference tracking is achieved without the risk of falsification by high feedback. Comparisons between the proposed method, a simple feedback and a standard 2-DOF controller with a conventional flatness-based feedforward control are performed on a state-of-the-art 4-cylinder engine test bed. The experiments constitute a rare implementation of a MPFFC on a real system and demonstrate its capabilities of separating reference and disturbance response as well as taking advantage of certain situations to even further improve reference tracking speed.

Feedforward Model Predictive Control of Fuel-Air Ratio for Lean-Burn Spark-Ignition Gasoline Engines of Passenger Cars

Precise fuel-air ratio (FAR) control on transient conditions is one of the key technologies for lean-burn spark-ignition (SI) engines. To improve fuel economy and reduce emissions, lean-burn engines must regulate their FAR immediately under different operating conditions. The precision of FAR control is limited by the large time-varying feedback delay which is caused by mixture combustion, exhaust gas transport, and lean $N{O_{x}}$ trap (LNT) module. Hence, a FAR predictive controller is proposed to track the desired FAR value precisely in the control framework of feedforward and feedback control. First of all, a feedforward predictive model for the FAR in the cylinder and a feedback predictive model for the FAR in the exhaust pipe with a time-delay characteristic are built respectively. Next, the FAR tracking requirement and the physical actuator constraints are transformed into the optimization objective function. Finally, a feedback/embedded feedforward predictive controller is designed, and the optimization problems are solved online to obtain the fuel injection quality. The simulation results based on GT-POWER illustrate that the proposed predictive control algorithm with feedforward predictive control can track the dynamic FAR precisely in a wide range. Meanwhile, the feedback controller decreases the effects of time delay and parameters uncertainty on the system dynamics.

Characteristics of the gait adaptation process due to split-belt treadmill walking under a wide range of right-left speed ratios in humans.

The adaptability of human bipedal locomotion has been studied using split-belt treadmill walking. Most of previous studies utilized experimental protocol under remarkably different split ratios (e.g. 1:2, 1:3, or 1:4). While, there is limited research with regard to adaptive process under the small speed ratios. It is important to know the nature of adaptive process under ratio smaller than 1:2, because systematic evaluation of the gait adaptation under small to moderate split ratios would enable us to examine relative contribution of two forms of adaptation (reactive feedback and predictive feedforward control) on gait adaptation. We therefore examined a gait behavior due to on split-belt treadmill adaptation under five belt speed difference conditions (from 1:1.2 to 1:2). Gait parameters related to reactive control (stance time) showed quick adjustments immediately after imposing the split-belt walking in all five speed ratios. Meanwhile, parameters related to predictive control (step length and anterior force) showed a clear pattern of adaptation and subsequent aftereffects except for the 1:1.2 adaptation. Additionally, the 1:1.2 ratio was distinguished from other ratios by cluster analysis based on the relationship between the size of adaptation and the aftereffect. Our findings indicate that the reactive feedback control was involved in all the speed ratios tested and that the extent of reaction was proportionally dependent on the speed ratio of the split-belt. On the contrary, predictive feedforward control was necessary when the ratio of the split-belt was greater. These results enable us to consider how a given split-belt training condition would affect the relative contribution of the two strategies on gait adaptation, which must be considered when developing rehabilitation interventions for stroke patients.

適応出力予測器による予測フィードフォワードを併用したASPR性に基づく出力フィードバック制御

適応出力予測器による予測フィードフォワードを併用したASPR性に基づく出力フィードバック制御

Investigating repetition and change in musical rhythm by functional MRI

Groove-based rhythm is a basic and much appreciated feature of Western popular music. It is commonly associated with dance, movement and pleasure and is characterized by the repetition of a basic rhythmic pattern. At various points in the musical course, drum breaks occur, representing a change compared to the repeated pattern of the groove. In the present experiment, we investigated the brain response to such drum breaks in a repetitive groove. Participants were scanned with functional magnetic resonance imaging (fMRI) while listening to a previously unheard naturalistic groove with drum breaks at uneven intervals. The rhythmic pattern and the timing of its different parts as performed were the only aspects that changed from the repetitive sections to the breaks. Differences in blood oxygen level-dependent activation were analyzed. In contrast to the repetitive parts, the drum breaks activated the left cerebellum, the right inferior frontal gyrus (RIFG), and the superior temporal gyri (STG) bilaterally. A tapping test using the same stimulus showed an increase in the standard deviation of inter-tap-intervals in the breaks versus the repetitive parts, indicating extra challenges for auditory-motor integration in the drum breaks. Both the RIFG and STG have been associated with structural irregularity and increase in musical-syntactical complexity in several earlier studies, whereas the left cerebellum is known to play a part in timing. Together these areas may be recruited in the breaks due to a prediction error process whereby the internal model is being updated. This concurs with previous research suggesting a network for predictive feed-forward control that comprises the cerebellum and the cortical areas that were activated in the breaks.

Predictive control of ankle stiffness at heel contact is a key element of locomotor adaptation during split-belt treadmill walking in humans

Split-belt treadmill walking has been extensively utilized as a useful model to reveal the adaptability of human bipedal locomotion. While previous studies have clearly identified different types of locomotor adaptation, such as reactive and predictive adjustments, details of how the gait pattern would be adjusted are not fully understood. To gain further knowledge of the strategies underlying split-belt treadmill adaptation, we examined the three-dimensional ground reaction forces (GRF) and lower limb muscle activities during and after split-belt treadmill walking in 22 healthy subjects. The results demonstrated that the anterior component of the GRF (braking force) showed a clear pattern of adaptation and subsequent aftereffects. The muscle activity in the tibialis anterior muscle during the early stance phase was associated with the change of braking force. In contrast, the posterior component of GRF (propulsive force) showed a consistent increase/decrease in the fast/slow leg during the adaptation period and was not followed by subsequent aftereffects. The muscle activity in the gastrocnemius muscle during the stance phase gradually decreased during the adaptation phase and then showed a compensatory reaction during the washout phase. The results indicate that predictive feedforward control is required to set the optimal ankle stiffness in preparation for the impact at the heel contact and passive feedback control is used for the production of reflexively induced propulsive force at the end of the stance phase during split-belt treadmill adaptation. The present study provides information about the detailed mechanisms underlying split-belt adaptation and should be useful for the construction of specific rehabilitation protocols.

Neuromuscular mechanisms and neural strategies in the control of time-varying muscle contractions

The organization of the neural input to motoneurons that underlies time-varying muscle force is assumed to depend on muscle transfer characteristics and neural strategies or control modes utilizing sensory signals. We jointly addressed these interlinked, but previously studied individually and partially, issues for sinusoidal (range 0.5-5.0 Hz) force-tracking contractions of a human finger muscle. Using spectral and correlation analyses of target signal, force signal, and motor unit (MU) discharges, we studied 1) patterns of such discharges, allowing inferences on the motoneuronal input; 2) transformation of MU population activity (EMG) into quasi-sinusoidal force; and 3) relation of force oscillation to target, carrying information on the input's organization. A broad view of force control mechanisms and strategies emerged. Specifically, synchronized MU and EMG modulations, reflecting a frequency-modulated motoneuronal input, accompanied the force variations. Gain and delay drops between EMG modulation and force oscillation, critical for the appropriate organization of this input, occurred with increasing target frequency. According to our analyses, gain compensation was achieved primarily through rhythmical activation/deactivation of higher-threshold MUs and secondarily through the adaptation of the input's strength expected during tracking tasks. However, the input's timing was not adapted to delay behaviors and seemed to depend on the control modes employed. Thus, for low-frequency targets, the force oscillation was highly coherent with, but led, a target, this timing error being compatible with predictive feedforward control partly based on the target's derivatives. In contrast, the force oscillation was weakly coherent, but in phase, with high-frequency targets, suggesting control mainly based on a target's rhythm.

Adaptive Predict Control of Nonlinear System Based on a Modified Neural Network

Based on a modified neural network with a new structure and a fast algorithm, an approach on adaptive predictive control is represented in this paper. Advantage of neural network modeling accuracy and good control performance of the optimal predictive control are combined. In the meantime, we can take the nonlinear network model output as a kind of measured disturbance, which is overcome in predictive feed-forward control. So the approach achieves satisfied control effectiveness.

Lack of Predictive Control in Lifting Series of Virtual Objects by Individuals With Diplegic Cerebral Palsy

To date, research on the motor control of hand function in cerebral palsy has focused on children with hemiplegia, although many persons with diplegic cerebral palsy (dCP) have asymmetrically decreased hand function. We explored the predictive capabilities of the motor system in a simple motor task of lifting a series of virtual objects for five persons with spastic dCP and five age-matched controls. When a person lifts an object, s/he uses an expectation of the weight of the object to generate a motor command. We asked the study subjects to lift a series of increasing weights and determined whether they extrapolated from past experience to predict the next weight in the series, even though that weight had never been experienced. Planning of precision grasp was assessed by measurement of the grip force at the beginning of the lifting task and by estimating the motor command. Execution of precision grasp was assessed by measurement of the time interval between the onset of grip and the onset of movement. We found that persons with dCP demonstrated a lack of predictive feed-forward control in their lifting movements: they exhibited a significantly longer time between onset of grip and onset of movement than the control subjects and they did not predict the weight of the next object in the lifting task. In addition, for subjects with dCP, the time between the onset of grip and the onset of movement of the dominant hand correlated strongly with the outcome of a hand function test. We postulate that a higher-order motor planning deficit in addition to execution deficit are evident in the subjects with spastic diplegic.

Cerebellum Predicts the Future Motor State

Feed forward control and estimates of the future state of the motor system are critical for fast and coordinated movements. One framework for generating these predictive signals is based on the central nervous system implementing internal models. Internal models provide for representations of the input-output properties of the motor apparatus or their inverses. It has been widely hypothesized that the cerebellum acquires and stores internal models of the motor apparatus. The results of psychophysical, functional imaging and transcranial magnetic stimulation studies in normal subjects support this hypothesis. Also, the deficits in patients with cerebellar dysfunction can be attributed to a failure of predictive feed forward control and/or to accurately estimate the consequences of motor commands. Furthermore, the computation performed by the cerebellar-like electrosensory lobes in several groups of fishes is to predict the sensory consequences of motor commands. However, only a few electrophysiological investigations have directly tested whether neurons in the cerebellar cortex have the requisite signals compatible with either an inverse or forward internal model. Our studies in the monkey performing manual pursuit tracking demonstrate that the simple spike discharge of Purkinje cells does not have the dynamics-related signals required to be the output of an inverse dynamics model. However, Purkinje cell firing has several of the characteristics of a forward internal model of the arm. A synthesis of the evidence suggests that the cerebellum is involved in integrating the current state of the motor system with internally generated motor commands to predict the future state.

Grip force control of predictable external loads

The grip force used to grasp and hold an object is modulated synchronously and precisely with a self-produced load indicating predictive feedforward control. It is unclear whether an externally produced load can be anticipated with similar feedforward-timing and precision if it can be predicted, e.g., because it has a periodic time course. In the present study we tested eight healthy subjects during the compensation of an externally produced sinusoidal load with cycle duration 1.5 s and more than 700 repetitions during two successive sessions. Performance parameters characterizing the timing and precision of the grip force-load coupling were analyzed across the sessions and compared with a retention measurement on the following day and with an experimental condition when the same loads were self-produced. The time lag between the grip force and the load decreased from values greater than zero to values close to zero during the practice sessions indicating a change from a more reactive to a predictive response. In contrast, the precision and economy of the coupling showed no improvement. Performance on the second day was similar to initial performance, only some retention of feedforward timing was obvious. Precision and economy of grip force control during self-produced loading was clearly superior to external loading even after extended practice. Our findings confirm that periodic external loads are controlled by predictive feedforward mechanism after sufficient experience. However, performance was not stable and did not reach the level of self-generated loading. These results are interpreted as reflecting the significance of an efferent copy of the motor command in sensorimotor processing that may be associated with a distinct neuronal representation.

High-stability temperature control for ST-7/LISA Pathfinder gravitational reference sensor ground verification testing

This article demonstrates experimental results of a thermal control system developed for ST-7 gravitational reference sensor (GRS) ground verification testing which provides thermal stability δT < 1 mK/√Hz to f < 0.1 mHz, and which by extension is suitable for in-flight thermal control of the LISA spacecraft to compensate solar irradiate 1/f fluctuations. Although for ground testing these specifications can be met fairly readily with sufficient insulation and thermal mass, in contrast, for spacecraft the very limited thermal mass calls for an active control system which can simultaneously meet disturbance rejection and stability requirements in the presence of long time delay; a considerable design challenge. Simple control laws presently provide ∼ 1mK/√Hz for >24 hours. Continuing development of a model predictive feedforward control algorithm will extend performance to <1 mK/√Hz at f < 0.01 mHz and possibly lower, extending LISA coverage of super massive black hole mergers.

PREDICTIVE ADAPTIVE FEEDFORWARD CONTROL OF A TIME SCALED SOLAR PLANT

This paper concerns the control of a solar energy collector field using predictive feedforward adaptive control techniques based on multiple identifiers. The ACUREX field used in these work is described by a partial differential equation (PDE). The plant is characterized by: non linearity, fast accessible disturbances and time varying dynamics. The dynamic dependency on flow is overcome by time-scaling. The result of this transformation is a discrete linear model with a Finite Impulse Response (FIR) transfer function. This means that the optimal predictive controller is given by a feedforward block. Simulation results on a detailed plant physical model are presented in order to illustrate the method.

Predictive Feedforward Control for a Hydroelectric Plant

In this brief, it is shown that predictive feedforward control can be used to help a hydroelectric plant achieve its target for delivered power while operating in grid (power system) frequency control mode. Due to the fundamental limitation on how fast the water in the supply tunnel can be accelerated by gravity, a hydraulic turbine is always subject to some delay between opening or closing its guide vane and the consequent change in power. Here, it is proposed that the effect of this delay can be alleviated by predicting 3-4 s ahead what power will be required, based on grid frequency measurements, and using this as a feedforward signal to supplement the normal PID feedback controller. Results are presented to show the improved accuracy of power delivery achieved with this approach.

EMG responses to an unexpected load in fast movements are delayed with an increase in the expected movement time.

When moving an object, the motor system estimates the dynamic properties of the object and then controls the movement using a combination of predictive feedforward control and proprioceptive feedback. In this study, we examined how the feedforward and proprioceptive feedback processes depend on the expected movement task. Subjects made fast elbow flexion movements from an initial position to a target. The experimental protocol included movements made over a short and a long distance against an expected light or heavy inertial load. In each task in a few randomly chosen trials, a motor applied an unexpected viscous load that produced a velocity error, defined as the difference between the expected and unexpected velocities, and electromyographic (EMG) responses. The EMG responses appeared not earlier than 170-250 ms from the agonist EMG onset. Our main finding is that the onset of the EMG responses was correlated with the expected time of peak velocity, which increased for longer distances and larger loads. An analysis of the latency of the EMG responses with respect to the velocity error suggested that the EMG responses were due to segmental reflexes. We conclude that segmental reflex gains are centrally modulated with the time course dependent on the expected movement task. According to this view, the control of fast point-to-point movement is feedforward from the agonist EMG onset until the expected time of peak velocity after which the segmental reflex feedback is briefly facilitated.

Application of Compensation Control in FCC Main Fractionator

AbstractA compensation control system is developed and applied to the main fractionator of a fluid catalytic cracking unit. A compensation model is used for the outlet temperature of light cycle oil (LCO) accounting for the top pressure change of the column. A corrected temperature is used as a indication of the pour point of LCO, and the top pressure of the column is used as a feedforward signal to maintain the pour point by means of a predictive feedforward control. The control strategy has proved successful in the FUJIAN refinery of the China Petrochemical Inc. which is used as an example of the system.