Actual System Dynamics Research Articles

Parameter Identification of Recurrent Fuzzy Systems With Fuzzy Finite-State Automata Representation

This paper presents the identification of nonlinear dynamical systems by recurrent fuzzy system (RFS) models. Two types of RFS models are discussed: the Takagi-Sugeno-Kang (TSK) type and the linguistic or Mamdani type. Both models are equivalent and the latter model may be represented by a fuzzy finite-state automaton (FFA). An identification procedure is proposed based on a standard general purpose genetic algorithm (GA). First, the TSK rule parameters are estimated and, in a second step, the TSK model is converted into an equivalent linguistic model. The parameter identification is evaluated in some benchmark problems for nonlinear system identification described in literature. The results show that RFS models achieve good numerical performance while keeping the interpretability of the actual system dynamics.

Variable-Gain Controllers for Nonlinear Systems Using the T–S Fuzzy Model

This correspondence proposes two novel control schemes with variable state-feedback gain to stabilize a Takagi-Sugeno (T-S) fuzzy system. The T-S fuzzy model is expressed as a linear plant with nonlinear disturbance terms in both schemes. In controller I, the T-S fuzzy model is expressed as a linear plant around a nominal plant arbitrarily selected from the set of linear subsystems that the T-S fuzzy model consists of. The variable gain then becomes a function of a gain parameter that is computed to neutralize the effect of disturbance term, which is, in essence, the deviation of the actual system dynamics from the nominal plant as the system traverses a specific trajectory. This controller is shown to stabilize the T-S fuzzy model. In controller II, individual linear subsystems are locally stabilized. Fuzzy blending of individual control actions is shown to make the T-S fuzzy system Lyapunov stable. Although applicability of both control schemes depends on the norm bound of unmatched state disturbance, this constraint is relaxed further in controller II. The efficacy of controllers I and II has been tested on two nonlinear systems.

Evaluation of multiple management objectives for Northeast Atlantic flatfish stocks: sustainability vs. stability of yield

Abstract This paper describes a simulation study that evaluated the ICES scientific advisory process used to recommend total allowable catches (TACs) for flatfish stocks. Particular emphasis is given to examining the effects on stock biomass, yield and stability of constraining interannual variation in TACs. A “management strategy evaluation” approach is used where an operating model is used to represent the underlying reality, and pseudo data are generated for use within a management procedure. The management procedure comprises a stock assessment that uses data to estimate parameters of interest and a decision rule to derive TAC recommendations for the following year. Bounds on TAC of between 20% and 40% have little effect on yields or stability, while a 10% bound on TAC can affect the ability to achieve management targets and result in low-frequency cycling in the stock. In the short term, performance is highly dependent on current stock status but bounds have less effect if the stock is close to equilibrium for a target fishing mortality (F). In addition, it was shown that current ICES biomass and fishing mortality reference points are not always consistent, and several are clearly inappropriate. Importantly, including realistic sources and levels of uncertainty can result in far from optimal management outcomes based on the current procedures. Results also conflicted with expert opinion, in suggesting that management based on a fixed F regime could result in relatively stable yields despite fluctuations in year-class strength and that the management feedback process itself is implicated in causing fluctuations in the system due to significant time-lags in this process. We therefore emphasize that providing more precise population estimates or developing harvest control rules alone will not necessarily help in achieving management objectives, rather management procedures that are robust to uncertainty and tuned to meet management objectives need to be developed. Operating models in these simulations were constrained to be based on existing ICES methods and perceptions of stock dynamics, but we recommend that, in future, operating models that represent the best available understanding of the actual system dynamics be used to evaluate models and rules considered for application.

An evaluation of the implicit management procedure used for some ICES roundfish stocks

Abstract This paper describes a simulation study that evaluated the performance of the scientific advisory process used by ICES to recommend total allowable catches (TACs) for roundfish stocks. A “management strategy evaluation” approach is used, involving development of an operating model to represent the underlying reality, and an observation model to generate pseudo data that are then used within a management procedure. The management procedure comprises an assessment that uses data to estimate parameters of interest and a decision rule to derive TAC recommendations for the following year. There are two important results: including realistic sources and levels of uncertainty can result in far from optimal management outcomes based on the current procedures; and current ICES biomass and fishing mortality reference points are not always consistent, and several are clearly inappropriate. This is because the types of projection used by ICES do not incorporate important lags between assessing stock status and implementing management measures, and they also ignore important sources of uncertainty about the actual dynamics, as well as our ability to collect data and implement management regulations (i.e. model, measurement, and implementation error, respectively). The simulation approach also showed that better management is not necessarily going to be achieved by improving the assessment, because even with a perfect assessment (where the simulated working group knew stock status perfectly), stocks may crash at fishing levels that standard stochastic projections would suggest were safe. It is proposed that, in future, operating models that represent the best available understanding of the actual system dynamics be used to evaluate models and rules considered for application. These operating models should capture the plausible range of characteristics of the underlying dynamics, but not necessarily model their full complexity. In general, they will be more complex than those used by assessment working groups, so developing management procedures that are robust to a broad range of uncertainty. However, the models and rules used as part of the management procedure should be simpler than those used at present.

Failure Detection and Identification in the Control of Large Space Structures

A method is presented for the detection and identification of component failures in structural control systems by considering stochastic noises and uncertainty in the system order. The proposed method is useful for bias–type failures of sensors and actuators and consists of two parts; the generalized likelihood ratio (GLR) method for detecting the faulty component and estimating the time of failure, and the partitioned linear optimal filter algorithm for estimating the bias and state separately. Utility of the combined algorithm for fault detection and recovery is illustrated with the example of a large space structure control system. The method works well for both of the sensor and actuator failures when the actual system dynamics and the filter model are perfectly matched. However, the capability of the basic GLR detector diminishes in the presence of unmodeled structural modes. An approach for alleviating this performance degradation is proposed in this paper. The improved detection algorithm utilizes the multiple sensors distributed at different positions on the structure to yield a new observation in such a way as to eliminate the observation spillover. Verification of the method is shown by an illustrative example.

Developing a policy simulator of the Israeli housing system: A progress report

Since mid-1972, members of the World Institute's Housing Systems Project have been working on a comprehensive analysis of Israeli housing and of some selected problems relating to it. As part of the project, a simulation model is under development. The target is to produce an operational man/machine interaction system focusing on both short and longer range policy issues relating to Israeli housing. As of this writing (August 1973), we have produced a manual version (i.e. non-computerized) of the housing system of Israel to serve as a “first-generation representation”; and have developed a fully-computerized representation aimed at replicating actual housing system dynamics in Israel between 1961 and 1972 (the two latest census years, for which comprehensive data exists). Work is now underway on several adaptations of the simulation, with special interest focusing on human-machine interplay through the use of remote consoles linked to the master computer routines.

A new method for reduction of dynamic systems†

A new method has been presented for the determination of a low-order model approximating a high-order system. It is based on the use of the matrix pseudo-inverse to estimate the parameters of the model which minimize the sum of the squares of the errors between the response of the actual system and that of the model at the sampling instants. One of the advantages of this method is that the description of the actual system dynamics need not be known, but only the measurements of the input-output data are required. As the algorithms are iterative, computation is fairly straightforward, and the requirement for storage of input-output data depends only on the order of the assumed model, not on the number of iterations in the interval considered for minimization. An example of the application of the method for the determination of an approximate second-order model of a seventh-order system has been given, and compared with the reduced model obtained using another method.