Page 261 - Mathematical Models and Algorithms for Power System Optimization
P. 261

Optimization Method for Load Frequency Feed Forward Control 253


                                               Disturbance

                               Compensation  controler  Power system  Local estimator estimates
                                                                the state of units


                                         Forecast the load
                                           disturbance      Central estimator estimates
                                                                the state of unit
                                                     Fig. 7.17
                                Schematic diagram for simulation test of tracking control.
               The concrete simulation process is, (1) the initial disturbance is given by the identified load
               disturbance model, (2) the actual state of the simulation is calculated according to the principle of
               electrical distance distribution, (3) Δω j is taken as the actual value the measurement errors in
               Gaussiandistributionasameasurementvalue,(4)theunitstateatthattimeisestimatedbythelocal
                                 ^
               estimator, (5) the ΔP Tj and Δ^ ω j estimated by the local estimator is taken as the input to the
               center estimator, (6) the state variables of the equivalent generator is estimated and used to make
               load forecasting. Take the load forecasting value ΔP L as the next step of disturbance value,
               and repeat the above process, during which the compensation controller shall be used to
               compensate disturbance and control the system state, thereby forming the whole process of
               tracking control.


               7.7 Transformation Methods of Linear Models


               The accurate mathematical descriptions for many industrial processes are unknown. The task
               of the modern identification method is to build the suitable or approximate mathematical
               models for these processes from the measurement data, which are often in discrete forms.
               Though the performance of a stochastic process can be well represented by discrete models,
               the continuous models are sometimes convenient for the system analysis and control. It is
               well known that the time interval of identified discrete models is fixed, which depends on the
               fixed sampling interval to a large extent in the measuring device. However, in the real time
               estimation and control of power system, from time to time it is desirable to change such time
               intervals to achieve the ideal control performance. For example, 1-s and 4-s sampling time
               intervals are allowed during estimation and control.

               When addressing the load frequency control problem of the power system, this chapter applies
               a series of system theories including identification, estimation, and control, and develop
               many specified types of models, such as state equation, transfer function, difference equation,
               and differential equation. Sometimes the model forms need to be transformed to each other for
               the special purpose. This section proposes some new methods to solve those problems. For the
               conversion problem between models with different sampling intervals, this section proposes
   256   257   258   259   260   261   262   263   264   265   266