METHOD AND DEVICE FOR COMPUTER-AID CONTROL AND / OR REGULATION OF THE OPERATION OF A POWER GENERATION SYSTEM, COMBINED POWER PLANT

Description

[0001] The present invention relates to a method and a device for computer-aided control and / or regulation of the operation of an energy generation system and a combined cycle power plant.

[0002] A combined cycle power plant (CCPP) is generally understood to be a power plant or virtual power plant that combines different energy source processes, for example, a combination of a Joule cycle or Brayton cycle with a Clausius-Rankine cycle in a combined cycle gas turbine (CCGT) power plant. Examples of such CCPP power plants include the well-known combined cycle gas turbine (CCGT) power plant or a combined heat and power (CHP) plant, where, in addition to generating electricity, residual energy resulting from the electricity generation process is also utilized. In English-speaking countries, these are referred to as Combined Cycle Power Plants (CCPPs) or Combined Cycle Gas Turbines (CCGTs).

[0003] A specific application of a combined cycle power plant is used in seawater desalination plants. Such plants generally employ a combined cycle power plant, in which the heat from electricity generation is used to desalinate seawater, for example, for drinking water production or for the production of process or cooling water. The present invention and the underlying problem are explained below with reference to such combined cycle power plants for seawater desalination, without, however, limiting the invention to this application.

[0004] Depending on requirements, combined cycle power plants for the purpose of seawater desalination can be operated in such a way that either the generation of electrical energy or the generation of desalinated water is the priority purpose, whereby the priority can be influenced, for example, by the operation of the auxiliary firing, the steam extraction and / or the amount of steam at the turbine.

[0005] WO 2012 / 072352 A1 describes a method for operating and controlling a gas turbine. EP 3 022 611 B1 further describes a method for the computer-aided control and regulation of a power generation plant. WO 2006 / 079483 A1 describes a system and method for operational planning, process monitoring, simulation, and optimization of a combined power generation and water desalination plant. The system comprises various modules containing components for modeling, simulation, and optimization. EP 2 657 468 A2 discloses a system for optimizing the power generation of a power plant equipped with a thermal generator. The system includes facilities for monitoring and simulating the power plant's performance in order to optimize costs. US 2007 / 240648 A1 describes systems and methods for optimizing the emissions and performance of a boiler plant powered by fossil fuels.US 2015 / 184549 A1 describes a method for optimizing the combined operation of a power plant network. EP 2 910 743 A1 describes a method for optimizing a combined cycle power plant using a thermodynamic model and simulations based on it with various operating configurations. DE 10 2007 061604 A1 describes a method for optimizing a steam power plant through simulations based on an operating model.

[0006] The object of the present invention is to improve the operation of a combined cycle power plant, in particular to enable more flexible operation of such a plant. This can be achieved by utilizing existing production flexibility to allow for more efficient operation.

[0007] According to the invention, at least one of these problems is solved by a method with the features of claim 1 and / or by a device with the features of claim 12 and / or by a combined power plant with the features of claim 13.

[0008] Accordingly, the following are planned: A method for computer-aided control and / or regulation of the operation of an energy generation system, in particular a combined power plant for energy generation and seawater desalination, comprising the steps of: creating a simulation model of the energy generation system; providing an efficiency calculation module which is designed to calculate the expected thermal and electrical efficiency of the energy generation system based on the created simulation model and using specified operating parameters;Applying a learning and / or optimization method to the provided efficiency calculation module using previously acquired known operating data of the energy generation system and / or based on simulation data obtained through the simulation model, such that the operation of the energy generation system is optimized with respect to at least one thermal and / or electrical operating parameter according to a target specification, wherein the application is carried out using one or more decision trees and / or based on particle swarm optimization; creating a control strategy for the energy generation system based on the trained efficiency calculation module;and operation of the energy generation system using the created control strategy. A device for computer-aided control and / or regulation of the operation of an energy generation system, in particular the operation of a combined heat and power plant for seawater desalination for the simultaneous generation of electricity and for operating a desalination plant with steam, wherein the device comprises: a simulation device configured to create a simulation model of the energy generation system; an efficiency calculation module configured to calculate the expected thermal and electrical efficiency of the energy generation system based on the created simulation model and using predefined operating parameters of the energy generation system;An AI device trained to apply a learning and / or optimization method to the efficiency calculation module using known operational data and / or simulation data, such that the operation of the energy generation system is optimized with respect to at least one thermal and / or electrical operating parameter, wherein the learning and / or optimization method is performed using one or more decision trees and / or based on particle swarm optimization; a control device trained to create a control strategy for the energy generation system based on the trained efficiency calculation module and to operate the energy generation system using the created control strategy; a combined cycle power plant comprising an energy generation unit and a computer-aided control and / or regulation unit for the operation of the combined cycle power plant.

[0009] The underlying idea of ​​the present invention is to extend the operation of an energy generation system, such as a combined cycle power plant for simultaneous energy generation and seawater desalination, in such a way that, depending on demand, either the generation of electrical energy, the production of desalinated water, or a predetermined combination thereof takes precedence. First, the efficiency of the energy generation system is calculated based on a simulation model and current, predicted, or simulated operating parameters. According to the invention, this operating efficiency is then optimized using a learning and / or optimization method (reinforcement learning control strategy) and with the use of known operating data and simulation data, which serve as training data.In this way, unlike known, more rigidly operated energy generation systems, the energy generation system can be operated flexibly based on changing operating modes and / or conditions. The learning and optimization method according to the invention provides a control strategy that allows the combined power plant to be operated optimally with regard to a target specification, even in light of changing specifications or a changing demand for desalinated water and / or electrical energy.

[0010] The learning and optimization method according to the invention is based on one or more estimation functions, which are based on models, such as thermodynamic or electrical simulation models, and whose basic forms may be known or individually specified, estimated, or approximated. To train the estimation function(s), the machine learning of the control unit can be based on determined, measured, or estimated operating data or on simulation data, such as from approximation models, of the combined cycle power plant and its components, wherein this data can reflect the current, past, and / or future operating mode of the combined cycle power plant and its components.

[0011] The learning and optimization method according to the invention can further be based on the output products and associated parameters of the combined cycle power plant obtained under the considered operating mode. Relevant parameters in this context can include the generated steam flow, which may be usable for further industrial applications and / or desalination, as well as the generated electrical and thermal power, and the efficiency of the power plant component(s) in this operating mode and under the respective input parameters (environmental parameters for power plant operation).

[0012] The learning and optimization method according to the invention can further include a variable learning method for generating a control strategy. This control strategy can use the estimating function(s) trained from machine learning as input(s) to generate and influence a demand for electrical power and steam quantity, as well as efficiency and output, by controlling the power plant operation, whereby control and / or input parameters of the estimating function(s) can be set and controlled in the most efficient way possible. This most efficient way can correspond to power generation and / or a reduction in fuel consumption, e.g., gas, whereby predetermined requirements for power, electricity generation, or other parameters can still be achieved.

[0013] The control strategy can be generated using the learning and optimization method and the simulation model according to the invention to optimize power plant operation. Through optimization using the control strategy, operating modes for the power plant can be determined that meet a specified electrical power demand and can deliver a steam flow or quantity within a required or necessary range, where this range can cover the current steam demand. Such a control strategy can advantageously offer the most energy-efficient control settings to achieve specified or requested quantities of electricity and / or steam, with the estimation function(s) representing the corresponding efficiency values ​​or parameters.These efficiency parameters or efficiency values ​​can then be used by the optimization function of the control strategy to determine (or find) local operating modes that can correspond to the most efficient way of operating the power plant, taking into account different steam generation or steam transport conditions.

[0014] The specific operating mode can result in a surplus or deficit of desalinated water compared to a benchmark value, such as that of an average plant. Depending on the existing surplus or deficit of desalinated water in the desalination plant's storage tanks, the control strategy can be adjusted to guide the control strategy, via a learning and optimization process, towards efficient generation parameters for electricity, heat, or water (steam). Input parameters and power plant operating parameters can thus be adapted to this target, for example, to generate more or less steam at low pressure and to increase or decrease water desalination accordingly, while still advantageously achieving the target for electricity generation.For example, optimization can occur even if the power plant is not currently operating in one of the possible and known optimized operating modes. If a current steam demand does not meet such a local optimal operating mode, the control strategy can detect and take this into account and then, for a specific period, vary or modify the target values ​​or specifications for electrical and / or steam demand so that they correspond to and / or meet a current or achievable electrical operating point / range (setpoint), combined with a reduced operating point / range (setpoint) for steam generation, whereby the electrical operating point / range (setpoint) and / or the operating point / range (setpoint) for steam generation can exhibit or generate higher efficiency (with regard to fuel consumption).After a certain period, the demand for steam can increase and / or the operating point / range (setpoint) for steam generation can be raised above an initial operating point / range (setpoint), thus increasing production beyond an initial level. This new operating point / range (setpoint) for steam generation can, in turn, exhibit higher operational efficiency than the original power plant operation. In this way, the electrical demand can always be met, while the steam demand can be met on average over a specific period.

[0015] As will be explained later, a switch can be made from an operating scenario with a simple electrical power requirement and low-pressure steam generation to another operating scenario. This second scenario may feature a slightly higher or lower low-pressure steam production, but with greater power plant efficiency. In this way, a shortage or surplus of water to be generated can be accumulated over a specific period, during which time the plant can always operate with higher efficiency than originally specified.

[0016] Through the buffering capacity of the desalination plant's desalination tanks, this variation in steam production and the associated water production can be carried out, provided that specified value ranges or limit ranges for existing or generated water quantities are still reached or maintained.

[0017] Advantageous designs and further developments result from the further sub-claims as well as from the description with reference to the figures in the drawing.

[0018] According to a preferred embodiment, in the process step of creation, the simulation model is created on the basis of known, temporally successive states and properties of the energy generation system, in particular the thermal and / or electrical operating mode and the efficiency.

[0019] It is particularly advantageous to perform some or all process steps of provisioning, applying, creating, and / or operating iteratively, especially after predefined time intervals and / or predefined operational events. In particular, this allows the simulation model and / or the efficiency calculation module to be continuously and / or iteratively compared and / or trained against the actual operation of the energy generation system. In the case of a combined cycle power plant, a specific level of steam generation and / or generated electrical and thermal output can be considered and compared with a target value. This advantageously allows the efficiency of the power plant operation and its respective operating modes to be increased and adapted to target values.This can correspond to a constantly or regularly repeated application of combined cycle optimization and / or load point shifting, for example towards more efficient operating conditions while maintaining the minimum electrical power.

[0020] The target specification, for example in the case of a combined cycle power plant, could include a predetermined electrical output and / or a predetermined steam quantity. However, other operating parameters are conceivable, such as steam pressure, temperature, etc.

[0021] According to a preferred embodiment, the application step comprises a machine learning training procedure using an artificial intelligence (AI) device, which in particular includes a trainable artificial neural network and / or a deep learning unit. The learning (or training) is based, for example, on training data of known action sequences with known states of operation of the technical system resulting from the action sequences.

[0022] According to the invention, the application step comprises an optimization method using one or more decision trees and / or based on particle swarm optimization to determine an operating strategy. Thus, for a given operating strategy, a given electrical power and steam flow can be generated through optimization, whereby the efficiency can be advantageously increased by the optimization.

[0023] According to a preferred embodiment, the simulation model comprises a physical model, an analytical model and / or a data-driven model for one or more components of the energy generation system.

[0024] According to a preferred embodiment, the energy generation system is a combined heat and power plant for seawater desalination. In this case, the simulation model takes into account, for example, a power plant operating parameter and / or an environmental parameter. By selecting the appropriate power plant operating parameter and / or environmental parameter, the estimation of the power plant's operating efficiency can be improved and defined in greater detail, which can lead to an increase in the accuracy of the efficiency calculation.

[0025] A power plant operating parameter can include at least one of the following pieces of information about the combined cycle power plant: Operating principle of a gas compressor; operating principle of a gas preheater; operating principle of a gas turbine; operating principle of a heat exchanger steam generator; operating principle of a steam turbine; operating principle of pumping systems; operating principle of a desalination tank.

[0026] An environmental parameter can include at least one of the following pieces of information about the combined heat and power plant: an ambient air temperature; an ambient air humidity; an ambient air pressure; an exhaust gas temperature; a steam temperature; an HRSG auxiliary firing; a steam flow of a gas turbine.

[0027] The aforementioned parameters and power plant components allow for the creation and use of an assessment model, for example, for the efficiency calculation module and / or the control strategy. In conjunction with this, or additionally based on further approaches (such as known models for the operating mode of the power plant components), models for electrical and / or thermal energy efficiency can be created and used, which can take into account the operating modes of parts or the entire operating cycle of the combined cycle power plant. These operating modes can consider all relevant power plant components and / or power plant operating parameters and / or environmental parameters.

[0028] The simulation models and efficiency calculation modules for electrical and / or thermal energy efficiency can be checked for their current accuracy, which can be a step in machine learning or reinforcement learning. This allows for an estimation of how much the quantities generated by the power plant operation (amount of electricity, electrical power, steam flow, amount of high- and low-pressure steam, steam temperature, and others) deviate from a target or expected value. From this, a degree of deviation can be determined, for example, for the operating strategy and its characteristic parameters. For instance, when determining the degree of deviation, variations in the efficiency of fuel utilization for the power plant, such as natural gas, can be considered for current operation and when switching to a different operating mode.This allows for an optimized distribution of power between different units or areas of the combined power plant.

[0029] In addition to or as an alternative to the environmental parameters mentioned so far, further control parameters of the combined cycle power plant and its components can be taken into account, for example exhaust gas temperature of a gas turbine, so-called "wet compression" as a power increase method, steam temperature, parameters of the auxiliary firing, attemperance or jet injection of the high-pressure steam, extraction flow at the steam turbine, as well as the use of the high-pressure to intermediate pressure (HP / IP) and high-pressure to low pressure (HP / LP) diverting stations.

[0030] According to a preferred embodiment, the control strategy is modified by the preceding learning and / or optimization procedure such that at least one optimized operating mode is determined in which a minimum value for electrical power is generated and the generated steam flow is maintained within a tolerance interval, wherein the tolerance interval fulfills a predetermined demand criterion. The tolerance interval can then increase the probability that a more efficient operation can be found while adhering to the demand criteria and / or that a target value for at least one generated parameter can be met.

[0031] According to a preferred embodiment, during operation of the combined cycle power plant, the operating parameters are checked to determine whether the steam flow currently generated in the plant exceeds the steam flow specified by the operating strategy. If an exceedance is detected, it can be provided that, while maintaining the specified electrical output, additional desalination takes place to at least partially compensate for a previous or anticipated period of lower desalination efficiency. Operation with higher steam production can be temporarily followed later by operation with lower steam production (or vice versa) if both modes of operation (or at least the combined operation) achieve higher efficiency than if the plant were operated precisely according to a specified target over a certain period.

[0032] According to a preferred embodiment, the operating strategy is scaled to future operating modes of the energy generation system. Scaling advantageously allows for responses to anticipated changes in operating or environmental conditions, as well as future demand. This involves identifying variations in electrical demand (e.g., electricity or energy demand), steam demand, and the demand for desalinated water. By scaling the target values ​​or specifications for electrical output and / or steam generation, the power plant operation can be adjusted to achieve the desired or necessary electrical output or steam generation at a given efficiency. Alternatively, or even instead, an expected demand profile (for electrical output and steam generation) can be met as efficiently as possible for steam generation overall (for steam; for electrical output, however, at any given time).

[0033] The learning and optimization approach according to the invention can correspond to a switch from an operating scenario with a simple electrical power requirement and low-pressure steam generation to a further operating scenario. This further operating scenario can exhibit a slightly higher or lower low-pressure steam production, but with higher power plant operation efficiency. In this way, a shortage or surplus of water to be generated can be accumulated over a specific period, whereby operation with increased efficiency than originally specified can always occur during this period.

[0034] After a certain period of time, or when a critical threshold is reached—for example, when a permissible maximum or minimum water level in the tanks is reached—operations can be switched to the other, more advantageous, or to a further operating scenario. This further scenario may also exhibit higher efficiency than the original requirement, but it does not allow for the consumption of the shortfall / surplus. In this way, a predetermined quantity of water can be produced over a specific period, while simultaneously generating the predetermined electrical power—precisely the predetermined electrical power—and consistently achieving higher efficiency than if the originally predetermined operating mode were followed.

[0035] Compared to known approaches, at least one or more simulation models and / or at least one or more efficiency calculation modules can be used to develop a suitable operating strategy that directly controls power plant operation. This operating strategy can also distribute operations across different units or areas of the power plant. Control parameters can be used to optimize the operation of the power plant or its components, for example, to improve electricity and steam generation with minimal fuel consumption. Numerous environmental parameters can also be considered, including the degradation (aging or wear and tear) of power plant components.For example, a range of values ​​for a power plant operation output of ±10 MW and / or for a steam flow of ±10 kg / s can be considered for operation, or switching between these values ​​is possible, or other values ​​may be considered.

[0036] The simulation models and / or efficiency calculation modules can be custom-built or predefined. They can be accessible and trained online. Furthermore, power plant operation can be varied to allow for flexibility in water and / or steam production, while maintaining variable efficiency and ensuring that electrical power output still meets a target value.

[0037] The simulation models and / or efficiency calculation modules can be used to leverage the efficiency of one or more units of a combined cycle power plant to determine a more sensitive distribution of electrical and steam output across various available units, particularly in the case of large combined cycle power plants with numerous units. This allows for improved power plant operation, which can be more effective than the previously common offline operation using simulations of power plant operation, which required implementation of control curves for operating parameters after the simulation. The optimization determined according to the invention can consider a significantly larger number of operating parameters and can do so interactively and in real time with the power plant operation.

[0038] According to a preferred embodiment of the combined power plant, the power generation device is configured to generate electrical energy and thermal energy for operating a desalination plant, and / or the combined power plant includes a steam generation device configured to simultaneously generate electrical energy and steam for operating a desalination plant.

[0039] The above embodiments and further developments can be combined with one another as appropriate. Further possible embodiments, further developments, and implementations of the invention also include combinations of features of the invention described previously or subsequently with regard to the exemplary embodiments, even if not explicitly mentioned. In particular, the person skilled in the art will also add individual aspects as improvements or additions to the respective basic form of the present invention.

[0040] The present invention will be explained in more detail below with reference to the exemplary embodiments shown in the schematic figures of the drawings. These show: Fig. 1 a schematic representation of an energy generation system designed as a combined power plant according to an embodiment of the present invention; Fig. 2 a block diagram to illustrate the functioning of the energy generation system Fig. 1 ; and Fig. 3 a flowchart for the inventive method for computer-aided control and / or regulation of the operation of an energy generation system.

[0041] The accompanying drawings are intended to provide a further understanding of the embodiments of the invention. They illustrate embodiments and, in conjunction with the description, serve to explain the principles and concepts of the invention. Other embodiments and many of the advantages mentioned will become apparent with reference to the drawings. The elements of the drawings are not necessarily shown to scale.

[0042] In the figures of the drawing, identical, functionally equivalent and similarly acting elements, features and components - unless otherwise stated - are each provided with the same reference symbols.

[0043] The Fig. 1 shows a schematic representation of an energy generation system designed as a combined power plant according to an embodiment of the present invention.

[0044] The energy generation system according to the invention comprises a control unit KE, a combined power plant KW and a desalination plant EA.

[0045] The control unit KE can be coupled to a combined cycle power plant KW via a control interface or be a component of the combined cycle power plant KW. According to the invention, the control unit KE is designed to perform a learning and / or optimization process relating to the control strategy for the combined cycle power plant KW, in order to generate a predetermined electrical output and a predetermined steam flow.

[0046] The control unit KE, used to control the operation of the combined cycle power plant KW for generating electricity EL and operating a desalination plant EA with steam D, is configured to determine at least one estimation function for the thermal and electrical operation and efficiency of the combined cycle power plant. This estimation function is trained, for example, using machine learning (or another learning or optimization method) based on operational or measurement data and / or simulation data. Based on this estimation function, a control strategy for the optimized operation of the combined cycle power plant is then developed using reinforcement learning. This control strategy is applied to the operation of the combined cycle power plant to generate a specified or predetermined electrical output and / or a predetermined quantity of steam for operating the desalination plant EA, as specified or predetermined by the operator of the combined cycle power plant.

[0047] Fig. 2 shows a block diagram to illustrate the functioning of the energy generation system. Fig. 1 .

[0048] The Figur 2 Figure 1 shows a block diagram connecting an input of the fuel gas FG, for example at a pressure of 10 bar, and an output of the production products, in particular electrical power and steam at low pressure LP, high pressure HP, and intermediate pressure IP. From the generated quantities for electrical power, the efficiency of the power plant operation according to the invention can be determined with respect to the required quantity of fuel gas FG, and this can be optimized to obtain higher quantities of steam at HP, IP, or LP while meeting achievable targets.

[0049] A first gas turbine GT11 and a second gas turbine GT12 can each be supplied with the fuel gas FG, for example, after it has been compressed by a gas compressor FG-V to a pressure of approximately 35 bar. The operation of the gas turbines can be described using a known or custom-designed gas turbine model (GT model), taking into account relevant operating data and parameters, for example, to determine estimation functions. The excess gas and / or exhaust gas from the two gas turbines can each be directed to a first heat exchanger steam generator HRSG11 and a second heat exchanger steam generator HRSG12. This section can be described using a model for the heat exchanger steam generators (HRSG model).

[0050] The heat exchange steam generators can be operated with feedwater from an associated pump and / or from a condenser in the power plant and generate high-pressure (HP) steam. A steam inlet (HP inlet) for the high-pressure steam can direct the steam to a steam turbine (ST) and then to an expansion system from high pressure (HP) to intermediate pressure (IP) and / or from high pressure (HP) to low pressure (LP). The steam turbine (ST) can, in turn, contribute to electrical power generation, in addition to that of the gas turbines, or supply steam at intermediate pressure (IP) and / or low pressure (LP). Furthermore, a portion of the steam output from the steam turbine (ST) can be directed to the condenser, which can then generate feedwater. A portion of the low-pressure (LP) output can also be condensed directly and then returned to the heat exchange steam generators.

[0051] The output products can then be compared with the demand for electrical power and intermediate-pressure steam, with a forecast for electrical power demand and a forecast for steam demand, and a deviation can be calculated. From this, an adjustable variation of the operating parameters can be generated, and a prioritization signal for electrical power and / or steam can be determined and transmitted to the heat exchanger steam generators and / or turbines to modify steam generation / power generation. Furthermore, to generate low-pressure steam (LP) from the steam turbine (ST), the extraction of residual steam can be adjusted according to the deviation and prioritization values.For the operation of the steam turbine and for the evaluation of the products, a steam turbine model and a steam prioritization model can be applied, in which appropriate system parameters can be taken into account to create appropriate estimation functions for the operation of these power plant modules.

[0052] Fig. 3 shows a flowchart for the inventive method for computer-aided control and / or regulation of the operation of an energy generation system.

[0053] In the inventive method for computer-aided control and / or regulation of the operation of an energy generation system, an initial process step S1 involves creating a simulation model of the energy generation system, the combined heat and power plant (CHP), or parts or components thereof. Depending on its complexity, this simulation model includes a more or less accurate representation of the energy generation system, the combined heat and power plant (CHP), or parts or components thereof, and thus, in a sense, an estimation function for its thermal and electrical operation.

[0054] In a subsequent second process step S2, an efficiency calculation module is provided. This module is designed to calculate the expected thermal and electrical efficiency of the energy generation system based on the created simulation model and using predefined operating parameters.

[0055] In a subsequent process step S3, the efficiency calculation module is trained S2. This training step S3 is performed by applying a learning and / or optimization method to the provided efficiency calculation module using known operating data of the energy generation system and / or simulation data obtained from the simulation model. In this way, the operation of the energy generation system can be optimized with respect to at least one thermal and / or electrical operating parameter. The optimization is preferably performed based on a target specification, for example, from the operator of the energy generation system, the combined cycle power plant, and / or the seawater desalination plant. Machine learning, for example, using a trained artificial neural network (ANN), is preferably suitable for the learning and / or optimization method.However, any other artificial intelligence (AI) methods would be conceivable, such as deep learning or reinforcement learning. Other learning and / or optimization methods, such as the use of suitable decision trees, would also be conceivable.

[0056] Based on the trained or optimized efficiency calculation module, an optimized control or operating strategy for operating the energy generation system is determined in a further process step S4.

[0057] Finally, in step S5, the energy generation system or combined heat and power plant is operated using the determined optimized control or operating strategy. In this way, a predetermined electrical output and / or a predetermined quantity of steam can be generated as needed and individually adapted to operate the desalination plant.

Claims

1. Method for the computer-aided open-loop and / or closed-loop control of the operation of an energy generation system, in particular of a combined power plant for energy generation and seawater desalination, having the steps: building (S1) a simulation model of the energy generation system; providing (S2) an efficiency calculation module which is designed to calculate the expected thermal and electrical efficiency of the energy generation system on the basis of the built simulation model and using predetermined operating parameters; applying (S3) a learning and / or optimization method to the provided efficiency calculation module using known operating data of the energy generation system and / or on the basis of simulation data obtained by the simulation model in such a way that the operation of the energy generation system is optimized in terms of at least one thermal and / or electrical operating parameter in accordance with a target value, wherein the application (S3) is performed using one or more decision trees and / or based on a particle swarm optimization; building (S4) an open-loop control strategy for the energy generation system on the basis of the trained efficiency calculation module; and operating (S5) the energy generation system using the built open-loop control strategy.

2. Method according to Claim 1, characterized in that, in the building method step (S1), the simulation model is built on the basis of known consecutive states and properties of the energy generation system.

3. Method according to either of the preceding claims, characterized in that the providing (S2), applying (S3), building (S4) and / or operating (S5) method steps are performed iteratively, in particular after predetermined time intervals and / or predetermined operating events.

4. Method according to one of the preceding claims, characterized in that the target value contains a predetermined electrical output and / or a predetermined steam quantity.

5. Method according to one of the preceding claims, characterized in that the applying step (S3) contains a training method for machine learning using an artificial intelligence (AI) device which comprises in particular a trainable artificial neural network and / or a deep learning unit.

6. Method according to one of the preceding claims, characterized in that the simulation model has a physical model and / or an analytical model and / or a data-driven model for one or more components of the energy generation system.

7. Method according to one of the preceding claims, characterized in that the energy generation system is a combined power plant for seawater desalination, wherein the simulation model takes into account a power plant operating parameter and / or environmental parameter.

8. Method according to Claim 7, characterized in that the power plant operating parameter contains at least one of the following pieces of information about the combined power plant: - mode of operation of a gas compressor; - mode of operation of a gas preheater; - mode of operation of a gas turbine (GT); - mode of operation of a heat recovery steam generator (HRSG); - mode of operation of a steam turbine (ST); - mode of operation of pump units; - mode of operation of a desalination tank.

9. Method according to Claim 7 or 8, characterized in that the environmental parameter contains at least one of the following pieces of information about the combined power plant: - a temperature of ambient air; - a humidity of the ambient air; - a pressure of the ambient air; - an exhaust gas temperature; - a steam temperature; - an HRSG supplementary firing; - a steam flow of a gas turbine (GT).

10. Method according to one of Claims 7 to 9, characterized in that the open-loop control strategy is modified by the above learning and / or optimization method in such a way that at least one optimized operating mode is determined in which a minimum value for the electrical output is generated and the generated steam flow is kept within a tolerance interval and / or in that a check is made of the operating parameters during the operation of the combined power plant as to whether a steam flow presently generated in the combined power plant exceeds the steam flow predetermined by the operating strategy and, in the case of a detected surplus, additional desalination takes place whilst maintaining the predetermined electrical output.

11. Method according to one of the preceding claims, characterized in that the operating strategy is scaled to future operating modes of the energy generation system.

12. Device (KE) for the computer-aided open-loop and / or closed-loop control of the operation of an energy generation system, in particular of the operation of a combined power plant (KW) to simultaneously generate electricity (EL) and to supply a desalination plant (EA) with steam (D), using a method according to one of Claims 1 to 11, wherein the device (KE) has: a simulation device which is designed to build a simulation model of the energy generation system; an efficiency calculation module which is designed to calculate the expected thermal and electrical efficiency of the energy generation system on the basis of the built simulation model and using predetermined operating parameters of the energy generation system; an AI device which is designed to apply a learning and / or optimization method to the efficiency calculation module using known operating data and / or on the basis of simulation data in such a way that the operation of the energy generation system is optimized in terms of at least one thermal and / or electrical operating parameter, wherein the learning and / or optimization method is performed using one or more decision trees and / or based on a particle swarm optimization; an open-loop control device which is designed to build an open-loop control strategy for the energy generation system on the basis of the trained efficiency calculation module and to operate the energy generation system using the built open-loop control strategy.

13. Combined power plant, with a device for energy generation, and with a device (KE) according to Claim 12, for the computer-aided open-loop and / or closed-loop control of the operation of the combined power plant.

14. Combined power plant according to Claim 13, characterized in that the device for energy generation is designed to generate electrical energy and thermal energy for operating a desalination plant, and / or in that the combined power plant comprises a device for steam generation which are designed simultaneously to generate electrical energy and steam for the operation of a desalination plant.