Method for model-predictive control of a fuel-air mixture of a system and an associated system

Description

[0001] The invention relates to a method for model-predictive control of a fuel-air mixture of a system and an associated system, which is preferably a gas boiler.

[0002] In devices known in the prior art for controlling a fuel-air mixture in a heating boiler, whereby heating boilers designed as gas boilers usually involve a gas-air mixture, a differential pressure sensor is usually provided which determines the pressure difference or the differential pressure between a pressure at a measuring point upstream of a main flow throttle of the gas boiler and a (reference) pressure at a reference point.

[0003] Furthermore, a digital controller is usually designed to adjust an actuator based on the measured pressure so that a desired setpoint value of the pressure is reached or maintained as far as possible without control deviation and without overshoot.

[0004] The actuator used is, for example, a valve that can be described as a control or gas valve, the valve position of which can be adjusted via a stepper motor, so that the flow through the control valve can be controlled step by step by the stepper motor.

[0005] In order to keep the measured pressure as the actual value within the target value, the current technology often involves repeatedly activating the actuator in rapid succession and over a long period of time, which can lead to significant wear on the actuator. This wear necessitates regular maintenance and, if necessary, replacement of the actuator and / or its components.

[0006] Further systems or gas boilers as well as methods for controlling a fuel-air mixture of a system are also known, for example, from the document WO 99 / 63273 A1, whereby the documents WO 2013 / 117516 A1 and WO 2011 / 120689 A1 also disclose aspects relevant to the technical background of the invention.

[0007] The invention is therefore based on the object of overcoming the aforementioned disadvantages and of providing a method for controlling a fuel-air mixture by means of which the service life of the actuator can be increased.

[0008] This object is achieved by the combination of features according to patent claim 1 and by the combination of features according to patent claim 8.

[0009] According to the invention, a method is therefore proposed for controlling a fuel-air mixture of a system, wherein this system is in particular a heating boiler and preferably a gas boiler. The system comprises a mixing device for mixing a fuel with air to form the fuel-air mixture, an actuator arranged upstream of the mixing device in the flow direction of the fuel and controlled by a manipulated variable for controlling a fuel mass flow of the fuel flowing into the mixing device, and a differential pressure sensor for detecting a differential pressure between a pressure of the fuel related to the flow direction of the fuel upstream of the mixing device and downstream of the actuator compared to a reference pressure in the flow direction of the air upstream of the mixing device as an actual value. The differential pressure can also be referred to as offset pressure or pressure difference.Furthermore, the system preferably comprises a main flow throttle which is arranged fluidically between the actuator and the mixing device, wherein the differential pressure sensor preferably detects the differential pressure between a pressure of the fuel upstream of the main flow throttle and the reference pressure. Due to the system, the actual value detected by the differential pressure sensor does not change immediately when the position of the actuator is changed by a change in the manipulated variable, which correspondingly leads to a change in the mass flow of the fuel flowing into the mixing device, but rather after a dead time and with an amplification factor which depends on the manipulated variable causing the change. A change in position is understood here in particular to mean a change in a flow position or a change in a mass flow through the actuator. Accordingly, the system behavior when the manipulated variable orwhen the position of the actuator changes, it can be described by the dead time and the gain factor. According to the method according to the invention, in a first method phase for identifying the system behavior, the manipulated variable for controlling the actuator is determined using a standard controller in order to adjust the actual value on average to a target value. Standard controllers are understood to be linear controllers in particular, and more particularly P, I, PI, PD and PID controllers. The manipulated variable and the actual value can oscillate around the target value with an amplitude and a frequency, particularly during the first method phase, due to the dead time and the gain factor. A curve of the actual value and a curve of the manipulated variable over time are recorded during the first method phase to identify the system behavior, and from these the gain factor is determined as a function of the manipulated variable and the dead time.After determining the dead time and the gain factor in the first process phase, the manipulated variable is determined in a second process phase for model-predictive adaptive control of the system using a model-based controller, in particular a Smith predictor. The model-based controller takes into account the previously determined gain factor and the dead time, which is preferably incorporated into the determination of the manipulated variable by the Smith predictor. The model-based controller is designed to adjust the actual value to the target value in such a way that the manipulated variable needs to be changed less frequently and to a lesser extent in the second process phase than in the first process phase.

[0010] By identifying the system behavior determined by the dead time and the gain factor in the first process phase and taking the system behavior into account in the second process phase, the actuator must be controlled less frequently in the second process phase and less strongly when controlled than in the first process phase and than in conventional controls or controls based exclusively on standard controllers, so that the wear of the actuator is reduced with the method according to the invention.

[0011] The proposed method for model predictive control enables a robust and at the same time high-performance control of the differential pressure, whereby by appropriately taking into account a change in the system behavior, it can also be referred to as an adaptive and self-learning control.

[0012] The system's gain depends on a multitude of factors and is therefore difficult or impossible to predict. The system's transfer function, which describes the gain, can be influenced, for example, by the heating output of the boiler or the position of the stepper motor or valve, the gas type, the gas supply pressure, and / or the control valve or its valve characteristic. Nevertheless, the gain can be determined relatively easily if the measured value of the differential pressure sensor is observed as the output when the manipulated variable changes as the input.

[0013] The dead time in the system is dominated by delays in data processing by the controller or control device, and by delays in data processing by the differential pressure sensor. Delays or dead times related to fluid mechanics processes are negligible in comparison. The resulting dead time can be considered essentially constant.

[0014] In order to ensure that a change in the system behavior is also detected and taken into account in the second process phase, an advantageous further development provides that during the second process phase a deviation of the actual value from the target value is determined and if the deviation and / or an average value of the deviation exceeds a predetermined limit value, the dead time and the gain factor are determined according to the first process phase.

[0015] Also to detect and account for a change in system behavior during the second process phase, a further embodiment provides for a control variable profile (over time) to be recorded during the second process phase, and for the number and magnitude of control variable changes to be determined. If the number and / or magnitude of control variable changes exceed a respective predetermined limit, the dead time and gain factor are recalculated according to the first process phase.

[0016] It may be provided that the procedure remains in the first procedural phase for a predetermined period of time before moving to the second procedural phase.

[0017] Alternatively or additionally, it can also be provided that in the first process phase a predetermined manipulated variable change takes place according to a predetermined time course and / or the system designed as a heating boiler follows a certain heating output curve in order to be able to determine the system behavior in a predetermined manner.

[0018] According to an advantageous further development, the dead time can be determined from a delay between a change in the manipulated variable and a change in the actual value caused by the change in the manipulated variable.

[0019] The gain in the system can be determined with little numerical effort from the signal curves (actual value of the pressure or the pressure difference or manipulated variable) of the oscillating controller in the first process phase.

[0020] Since the amplitudes of these signals are comparatively small for most systems, a gain determined in this way can be considered representative of the system.

[0021] Preferably, the gain factor and / or the dead time are / is determined by a Wiener filter. The Wiener filter is designed to determine the gain factor and / or the dead time based on the manipulated variable and the actual value.

[0022] Especially when using a simplified digital Wiener filter, the gain results from the quotient of cross-correlation and auto-correlation of the signals over time. In this case, a period of 2 to 3 seconds is sufficient. The cross-correlation of the data can also be used to determine or validate the dead time in the system.

[0023] The model-based controller, which is used in the second process phase to determine the manipulated variable, preferably comprises the Smith predictor and a standard controller. The manipulated variable is determined by superimposing the Smith predictor with the standard controller, so that the dead time is compensated by the Smith predictor and the gain by the standard controller.

[0024] A further aspect of the invention relates to a system, which is in particular a gas boiler. The system comprises a mixing device for mixing a fuel with air to form the fuel-air mixture, an actuator arranged upstream of the mixing device in the fuel flow direction and controlled by a manipulated variable for regulating a fuel mass flow of the fuel flowing into the mixing device, and a differential pressure sensor for detecting a differential pressure between a fuel pressure related to the fuel flow direction upstream of the mixing device and downstream of the actuator compared to a reference pressure in the air flow direction upstream of the mixing device as an actual value.Furthermore, the system preferably comprises a main flow restrictor, which is fluidically arranged between the actuator and the mixing device, wherein the differential pressure sensor preferably detects the differential pressure between a fuel pressure upstream of the main flow restrictor and the reference pressure. Furthermore, the system is provided with a control device, which is signal-connected to the actuator and the differential pressure sensor and is designed to carry out a method according to the invention.

[0025] According to an advantageous development of the system, the actuator is a control valve with a stepper motor, which can be used to adjust the mass flow through the control valve. Furthermore, the manipulated variable is a number of steps of the stepper motor or a value that determines the number of steps of the stepper motor.

[0026] Preferably, the target value is a predetermined value and in particular 0 Pa.

[0027] The features disclosed above can be combined as desired, as long as this is technically possible and they do not contradict each other.

[0028] Other advantageous developments of the invention are characterized in the subclaims or are presented in more detail below, together with the description of the preferred embodiment of the invention, with reference to the figure. It shows: Fig. 1 an exemplary schematic representation of a gas boiler.

[0029] Figure 1 schematically shows a part or section of a gas boiler, wherein a Venturi mixer is shown as the mixing device 4, into which air is sucked from the environment by a fan 5 through an air inlet L at an air pressure p0. In the mixing device 4, the inflowing air and a fuel (gas) flowing in through the fuel supply G are mixed to form a fuel-air mixture.

[0030] The fuel flowing in from the fuel supply G, which is in particular a gas, flows through a safety valve 1, an actuator 2 designed as a control valve, and the main flow throttle 3. The safety valve 1 preferably has a pass-through position and a blocking position in which the flow of fuel through the safety valve 1 is blocked. The actuator 2 is designed to control the volume or mass flow of the fuel, so that the fuel flow through the actuator 2 to the mixing device 4 can be adjusted. By adjusting or regulating the actuator 2, the mixing ratio of the fuel-air mixture can be adjusted. For this purpose, the actuator 2 has a valve 22, the flow position of which can be changed or adjusted by a stepper motor 21, wherein the stepper motor 21 is controlled by a control device 6 with a manipulated variable.

[0031] Furthermore, at least one differential pressure sensor 7 is provided, which is designed to determine the differential pressure between the pressure p2 of the fuel upstream of the main flow throttle 3 and downstream of the actuator 2 and a reference pressure, wherein the reference pressure is preferably the ambient pressure p0 or a pressure p1 of the air in an air-conducting supply line to the mixing device 4. For this purpose, the differential pressure sensor 7 can, for example, have a respective pressure sensor or pressure transducer for detecting a respective pressure p0, p1, p2. Furthermore, further pressure sensors can be provided for detecting the further pressures pg, p3 and p4, which can serve as reference pressure sensors for detecting a reference pressure or for checking the plausibility of the pressures p0, p1, p2. The differential pressure sensor 7 is connected to the control device 6 for signaling purposes.

[0032] The fuel-air mixture is conveyed by the fan 5 to a burner (not shown) of the gas boiler, where the fuel-air mixture is burned.

[0033] According to the method, in a first method phase, it is provided, by way of example, that a controller implemented on the control device 6 controls the differential pressure determined by the differential pressure sensor 7 between the pressure p2 of the fuel upstream of the main flow throttle 3 and a reference pressure p1 as the actual value on average to a desired setpoint value (in particular 0 Pa) by controlling the actuator 2. Since the system is dominated by dead times, the closed control loop, which has the manipulated variable generated by the controller for controlling the stepper motor 21 as the input variable and the differential pressure determined by the differential pressure sensor 7 as the output variable, exhibits boundary-stable behavior. The differential pressure as the actual value therefore oscillates with an amplitude and frequency around the setpoint value, with the average control deviation being approximately 0. This boundary-stable behavior can be used to identify the system or system behavior.The system behavior is characterized by the dead time and by a gain depending on the position of the actuator 2, which are determined accordingly.

[0034] Once the system has been identified, i.e., the dead time and gain have been determined as a function of the position of actuator 2, the control can be adjusted in a second process phase. For this purpose, the controller used in the first process phase and implemented on control device 6 is expanded by a Smith predictor, or the Smith predictor is activated. Alternatively, another controller, implemented, for example, on control device 6, can be activated, which takes over control in the second process phase and also includes a Smith predictor.

[0035] In the second process phase, the control of actuator 2, or the determination of the manipulated variable, takes place by superimposing a standard controller (such as PI or PID) and the Smith predictor. The dead time is compensated by the Smith predictor, and the gain is compensated by the standard controller. A controller superimposing a standard controller and a Smith predictor can also be referred to as a Smith controller and / or model predictive controller.

[0036] By taking the gain and the dead time into account, oscillation of the actual value around the setpoint value can be minimized, so that actuator 2 needs to be controlled less often and less strongly.

[0037] The invention is not limited to the preferred embodiments described above. Rather, a number of variants are conceivable that utilize the presented solution even in fundamentally different embodiments.

Claims

1. A method for control of a fuel-air mixture of a system, which in particular is a gas heater, wherein the system has a mixing device (4) for mixing a fuel with air in order to form the fuel-air mixture, an actuator (2) arranged in the flow direction of the fuel upstream of the mixing device (4) and driven by a control variable for control of a fuel mass flow and a differential pressure sensor (7) for detecting a differential pressure between a pressure p2 of the fuel upstream of the mixing device (4) and downstream of the actuator (2) with regard to a reference pressure in the flow direction of the air upstream of the mixing device (4) as an actual value, wherein the actual value detected by the differential pressure sensor (7) changes with a change in position of the actuator (2) after a dead time and with a gain factor which depends on the control variable causing the change so that the system behaviour can be described by the dead time and the gain factor, characterised in that the control variable is determined by driving the actuator (2) in a first method phase for identification of the system behaviour with a standard controller in order to align the mean actual value to a target value, wherein the control variable and the actual value can each swing about the target value by the dead time and the gain factor with an amplitude and a frequency, wherein a course of the actual value and a course of the control variable are detected during the first method phase for identification of the system behaviour and the gain factor as a function of the control variable and the dead time are determined therefrom, wherein the control variable, after the dead time and the gain factor have been determined, in a second method phase for model-predictive adaptive control of the system is determined with a model-based controller in particular having a Smith predictor, which takes into account the gain factor and the dead time, in order to align the actual value to the target value so that the control variable has to be changed less often and less severely in the second method phase compared to the first method phase.

2. The method according to claim 1, wherein a deviation of the actual value from the target value is determined during the second method phase, and wherein, if the deviation and / or a mean value of the deviation exceeds a predetermined limit value, the dead time and the gain factor are identified according to the first method phase.

3. The method according to claim 1 or 2, wherein a course of the control variable is detected and a number and height of changes of the control variable are determined during the second method phase, and wherein, if the number and / or the height of the changes of the control variable exceed a respective predetermined limit value, the dead time and the gain factor are identified according to the first method phase.

4. The method according to any one of the preceding claims, wherein the method remains in the first method phase for a predetermined time before a switch to the second method phase.

5. The method according to any one of the preceding claims, wherein the dead time is determined from a delay between a change of the control variable and a change of the actual value caused by the change of the control variable.

6. The method according to any one of the preceding claims, wherein the gain factor and / or the dead time is / are determined by a Wiener filter, wherein the Wiener filter is formed to determine the gain factor and / or the dead time by the control variable and the actual value.

7. The method according to any one of the preceding claims, wherein the model-based controller comprises the Smith predictor and a standard controller, and wherein the control variable is determined from an overlap of the Smith predictor with the standard controller so that the dead time is compensated by the Smith predictor and the gain factor is compensated by the standard controller.

8. A system, in particular a gas heater, wherein the system has a mixing device (4) for mixing a fuel with air in order to form the fuel-air mixture, an actuator (2) arranged in the flow direction of the fuel upstream of the mixing device (4) and driven by a control variable for control of a fuel mass flow and a differential pressure sensor (7) for detecting a differential pressure between a pressure p2 of the fuel upstream of the mixing device (4) and downstream of the actuator (2) with regard to a reference pressure in the flow direction of the air upstream of the mixing device (4) as an actual value, and wherein the system further has a control device (21) which is connected to the actuator and the differential pressure sensor (7) in a technically signal-secure manner, characterised in that the control device (21) is formed to perform a method according to any one of the preceding claims.

9. The system according to claim 8, wherein the actuator (2) is a control valve with a stepper motor (21) by which the mass flow can be adjusted by the control valve, and wherein the control variable is a number of steps of the stepper motor (21).

10. The system according to claim 8 or 9, wherein the target value is a predetermined value, in particular 0 Pa.