مجله کنترل
سال هشتم شماره 1 (بهار 1393)

In this paper، design of an optimal control strategy for the powertrain of a parallel hydraulic hybrid bus is proposed. The powertrain includes an internal combustion engine as the first power generation source and a hydraulic pump/motor as the second one. Design procedure of a proper control strategy for the hybrid powertrains is extremely dependent on the speed trend of the driving cycle. This functionality is such that the control decision is affected by the future trend of the driving cycle speed، too. Here، a dynamic programming algorithm is used for generating the optimal control strategy in a special driving cycle. The disadvantage of the designed control strategy is that it is fully dependent to the future information of the driving cycle. This problem would be eliminated by using an intelligent control strategy. The control strategies including an identification unit for the driving cycle are named as the intelligent. An appropriate method to design the intelligent control strategy is using the online models of the optimal control strategies for some standard driving cycles in different time periods of a special driving cycle. In this paper، a set of models contains several neural networks is applied to generate the online models of the pre-developed optimal control strategies. The generated models are used as the online optimal control strategies on the hydraulic hybrid bus. Finally، the results of the bus simulation using the online optimal control strategy and a rule-based one are compared for assessment of the proposed design. It can be seen that the fuel consumption of the bus is reduced by using the online optimal control strategy.

In this paper، we present a novel continuous reinforcement learning approach. The proposed approach، called «Fuzzy Least Squares Policy Iteration (FLSPI)»، is obtained from combination of «Least Squares Policy Iteration (LSPI)» and a zero order Takagi Sugeno fuzzy system. We define state-action basis function based on fuzzy system so that LSPI conditions are satisfied. It is proven that there is an error bound for difference of the exact state-action value function and approximated state-action value function obtained by FLSPI. Simulation results show that learning speed and operation quality for FLSPI are higher than two previous critic-only fuzzy reinforcement learning approaches i. e. fuzzy Q-learning and fuzzy Sarsa learning. Another advantage of this approach is needlessness to learning rate determination.

This paper proposes a new method for the identification of switched linear systems. The proposed method includes two main steps of mapping and clustering. At the first step، a mapping is developed from the space of input-output data into the parameter space. A lot of linear equation sets، composed of equal number of equations and unknowns، are solved in this part. At the next step، submodel parameters are derived by clustering the parameters obtained in previous step، into several groups. Since the clustering step is carried out in the parameter space، the proposed method makes no distinction between the identification of switched linear systems and piecewise linear systems and the identification of submodel parameters is done independently of the estimation of switching signal. Numerical examples show the effectiveness of the proposed method in the identification of switched linear systems.

Today all critical infrastructure and industrial systems apply network-based automation and control systems to monitor and control their processes. Safe، effective and efficient management، coordination and operation of these units are possible through these control systems. In other words، automation and control systems are considered as the brain and nervous systems of critical infrastructure and industrial systems. Using computer and information technology to enhance the quality، performance and reliability of control systems، caused them facing unexpected threats; Cyber attacks are the most important ones. Following the Stuxnet attack، Industrial Cyber Security (ICS) has become a serious challenge for control engineering studies. Over the last few years، many researches have been conducted in the field of ICS. In this paper، we survey the literature of this area from the perspective of control engineering to present an overview of this issue and informing experts of this field with the related importance and research opportunities.

This paper deals with identification of a group of multi-input static systems with noisy inputs and output. This group includes static models y=∑_ (i=1) ^n▒〖a_i u_i 〗 where u_is are contaminated by noise. It is assumed that we know which inputs are the contaminated ones. To reach the goal، the PEM which is an efficient classical system identification method is implemented. Then، by combining it with the IV method which is commonly used for identification of the intended group of systems، a new method is proposed. We call the new method as compensated PEM. Since both PEM and IV methods are recursively implementable، the proposed method could be implemented recursively as well. It is proved that the new method results in an unbiased estimation. Simulation results are provided to verify the given theorems and compare the proposed method with its competitors in mean and variance of the estimation. Finally a practical application of the method is investigated.

Optimal control of an observer trying to localize a stationary target is presented. The objective is to generate the control signal in order to reduce the position estimate error. The trace of estimation error covariance matrix at final time is selected as the optimality criterion. This problem is solved by presenting a control policy which relates the observer course to the target bearing. In addition to generating paths in real-time which are very close to the optimal path، the advantage of the proposed method is its ability to consider the observer’s field of view. In comparison to a similar method، the performance of the proposed method is shown using various simulations. This method is also applied to the control of a mobile robot.