Method of aero-engine on-line optimization and multivariable control based on model prediction

A design method of aero-engine on-line optimization and multivariable control based on model prediction, a control system structure mainly consisting of two parts:

• the first part is a prediction model acquisition layer that continuously establishes a small deviation linear model of an aero-engine near different steady state points based on the actual operating state of the aero-engine in each control cycle and external environment parameters and that supplies model parameters to a controller;
• and the second part is a control law optimizing and decision-making layer that consists of a model prediction controller and an external output feedback and determines the control output of the controller by solving a linear optimization problem with constraint and closed-loop feedback effect, wherein the design method comprises the following specific steps:
• S1. continuously establishing a small deviation linear model of an aero-engine taking the current actual operating state of the aero-engine and the external environment as steady state points in real time;
• the actual input of the aero-engine comprising a fuel flow of a main combustor, a fuel flow of an afterburner, a cross-sectional area of a nozzle, a fan guide vane angle, a compressor guide vane angle and other aero-engine controllable parameters;
• the external environment parameters comprising a flight altitude, a flight Mach number and other environmental parameters;
• obtaining the small deviation linear model of the aero-engine through a method of power extracting and expressing in the form of state space;
• S2. normalizing the small deviation linear model of the aero-engine obtained in step S1, i.e., converting absolute increments of an input variable, a state variable and an output variable into relative increments about the steady state points through linear transformation;
• discretizing the normalized model so that the model parameters are convenient for use in digital control;
• a sampling cycle of discretization being the same as the control cycle;
• S3. designing a model prediction multivariable controller;
• data stored in the controller comprising:
• linear model parameters of aero-engine normalization and discretization, a discretization model of an actuator and constraint conditions of the control system, wherein the control structure of the control system, the actuator and the aero-engine are two subsystems which are connected in series;
• the aero-engine and the actuator are combined in series in each control cycle into a state space expression of an entire system, and are extended into an augmented state space expression;
• for the on-line optimization part, constructing a target function of index minimization of the difference between predicted output and actual measured output, the change range of the input and other factors, and taking the limit of the actual input and output of the control system as the optimization constraint condition to online solve the linear optimization problem with constraints to obtain the control output of the next step;
• S4. applying the controller output obtained in step S3 to a controlled system and comparing with the actual output of the aero-engine obtained by a sensor to obtain a control error as the input of the model prediction controller;
• and guiding the controller to make a decision on the controller output of the next step.