Method for determining an optimized predictive operating strategy of a fuel cell system of a fuel cell vehicle

Abstract

The invention presented relates to a method for determining an optimized predictive operating strategy of a fuel cell system (10) of a fuel cell vehicle (20), the method comprising the steps: - Determination of several operating parameters of the fuel cell system (10) within an adaptive prediction horizon (30) lasting over several driving cycles, - Derivation of target conditions for a next stop / standstill based on the determined operating parameters from the adaptive prediction horizon (30), - transmission of the derived target conditions to an already running optimized operating strategy of a shrinking prediction horizon (50) within a current driving cycle, - Determination of control variables taking into account the already running optimized operating strategy of the shrinking prediction horizon (50) within the current driving cycle, and - Adjusting the operating mode of the fuel cell system (10) based on the determined control variables by a control device (60).

Description

[0001] The invention relates to a method for determining an optimized predictive operating strategy of a fuel cell system of a fuel cell vehicle according to claim 1. The invention further relates to a fuel cell system and a fuel cell vehicle. State of the art

[0002] In vehicles with fuel cell systems (FCS) as a drive system, which are referred to as Fuel Cell Vehicles (FCVs), the oxidizing agent oxygen from the ambient air is usually used to react with hydrogen in the fuel cell to form water or water vapor and thus to deliver electrical power through electrochemical conversion.

[0003] In fuel cell vehicles, the aim is always to achieve an energy-optimized operating state. For example, hydrogen consumption must be optimized. However, aging and degradation processes can also occur in fuel cell vehicles, particularly in the fuel cell system, which must be taken into account. The efficiency and service life of the fuel cell system therefore depend on the combination of various operating parameters, which can be determined by an operating strategy. The operating strategy therefore represents the logical and chronological sequence of all operating states of the fuel cell system, taking the operating parameters into account. A control device of the fuel cell system determines the operating strategy, which determines how the fuel cell system must be adjusted to achieve an energy-optimized operating state of the fuel cell vehicle.

[0004] A variety of operating strategies are known from the state of the art. However, these are not optimized because they are based on already known characteristic maps. Such operating strategies are not adaptive and have the disadvantage of being directed only at a specific operating target or driving cycle. Disclosure of the invention

[0005] According to a first aspect of the invention, a method for determining an optimized predictive operating strategy of a fuel cell system of a fuel cell vehicle is presented. The method comprises the following steps: - Determination of several operating parameters of the fuel cell system within an adaptive prediction horizon lasting over several driving cycles, - Derivation of target conditions for a next stop / standstill based on the determined operating parameters from the adaptive prediction horizon, - Transmission of the derived target conditions to an already running optimized operating strategy of a shrinking prediction horizon within a current driving cycle, - Determination of control variables taking into account the already running optimized operating strategy of the shrinking prediction horizon within the current driving cycle, and - Adjusting the operating mode of the fuel cell system based on the determined control variables by a control device.

[0006] The method according to the invention is characterized by the fact that multiple predictions allow multiple operating targets to be optimized simultaneously, thus ensuring an overall improved, energy-optimized operating state of the fuel cell vehicle. This, in turn, leads to a significant reduction in the TCO (Total Cost of Ownership).

[0007] In particular, the method according to the invention can enable better predictive real-time optimization and thus better dynamics of the fuel cell system, so that required performance trajectories, in particular power trajectories, can be easily achieved. Furthermore, the method according to the invention can minimize fuel consumption (hydrogen), the aging or degradation of the fuel cell stack, the aging or degradation of the high-voltage battery, and the aging or degradation of many other components, such as the air compressor.

[0008] Further features and advantages of the invention are set forth below.

[0009] For example, the method may provide for optimization over the prediction horizon, which extends over several driving cycles, using a cost function, also known as a quality function, that simultaneously includes one or more operating objectives. In each case, several driving cycles or driving tasks are considered.

[0010] A cost function that addresses, for example, the operating target of H2 consumption can look like this for the prediction horizon lasting over several driving cycles: min{J=∫tbegtendmfuel(act)+∑n=1Nmfuel[n]dt}

[0011] Here, n represents the number of trips or driving cycles, mfuel(act) represents the fuel consumption for the remaining current trip, and mfuel(n) represents the fuel consumption for the next trip(s) or for the next partial trip. The boundaries of the prediction horizon are defined by tbeg and tend.

[0012] Another preferred cost function, which takes into account in particular the costs incurred over driving cycles and stop or standstill phases, can, for example, look as follows for the prediction horizon lasting over several driving cycles: min{J=∫tbegtendCoststotal(act)+∑n=1NCoststotal[n]dt}

[0013] Preferably, further operating objectives, such as the aging or degradation of the components to be optimized, can also be taken into account in the optimization function, so that the optimization function for the prediction horizon lasting over several driving cycles can, for example, look like this: min{J=∫tbegtendmfuel(act)+Ddegr(act)∑n=1Nmfuel[n]+∑n=1NDdegr[n]dt}

[0014] According to the presented method, however, it can also be provided that the optimization over the prediction horizon lasting over several driving cycles is determined by an energy function instead of the cost function.

[0015] The energy function for the prediction horizon lasting over several driving cycles can, for example, look like this: min{J=∫tbegtendEdrive(act)+Estandstill(next)+∑n=1N(Edrive[n]+Estandstill[n+1])dt}

[0016] Edrive(act) corresponds to the energy requirement for the remainder of the current journey and Estandstill(next) corresponds to the energy requirement for the next stop or standstill phase(s) of the next journey(s) or for the next partial journey.

[0017] In other words, multiple operating parameters can be considered within the prediction horizon, which extends over several driving cycles. The operating objectives can include, among other things, driving cycles, stop or standstill phases, and changes during these phases.

[0018] Other operating objectives that can be taken into account in the optimization function within the prediction horizon lasting over several driving cycles are, for example, the driving and elevation profile of a route in subsequent driving cycles, the driving orders, the change in the vehicle load or vehicle mass, the change in the battery state of charge (SOC), the hydrogen tank content as well as the filling station network and the filling station availability.

[0019] From the optimization function derived from the prediction horizon, which lasts over several driving cycles, target conditions, especially optimal target conditions, for the next stop / standstill can be derived. For example, it is possible to determine in which phase of the stop / standstill the battery state of charge (SOC) should be set for the next stop / standstill.

[0020] Such target conditions can then be transmitted or specified to an already running optimized operating strategy of a prediction horizon within a current driving cycle.

[0021] The prediction horizon, which lasts over several driving cycles, is adaptive and dependent on the specified operating objectives. The prediction horizon within the current driving cycle is shrinking and therefore faster.

[0022] In other words, in order to achieve an energy-optimal operating state of the fuel cell system, operating goals can be considered both predictively in the adaptive prediction horizon lasting over several driving cycles and in the prediction horizon shrinking within the current driving cycle, whereby the influences of the stop or standstill phases as well as the changes in these phases are also taken into account.

[0023] This means that the method according to the invention can be combined in particular with MPC (model predicted control), MHE (moving horizon estimation), RL (reinforcement learning) and other AI / ML methods.

[0024] The method according to the invention is also suitable for commercial vehicles, such as long-haul trucks. The method according to the invention is also suitable for mobile or stationary systems with different fuel cell technologies, such as PEM and SOFC, which operate in a start-stop mode or are subject to cyclic operation.

[0025] According to a second aspect of the invention, the invention also provides a fuel cell system for a fuel cell vehicle with a control device for adjusting the operating mode of the fuel cell system.

[0026] Another object of the invention is a fuel cell vehicle with such a fuel cell system.

[0027] Further advantages, features, and details of the invention will become apparent from the following description, in which exemplary embodiments of the invention are described in detail with reference to the drawings. The features mentioned in the claims and in the description may be essential to the invention individually or in any combination.

[0028] The invention is explained in more detail below with reference to the accompanying drawings, which schematically show: Fig. 1 shows a block diagram illustrating the optimization method according to the invention; Fig. 2 a flowchart of the method steps according to the invention for adjusting the operating mode of a fuel cell system; and Fig. 3 a fuel cell vehicle according to the invention with a fuel cell system and a control device.

[0029] Fig. Figure 1 shows a block diagram illustrating the concept of the optimization method according to the invention. A large number of operating objectives are analyzed over a longer prediction horizon 30, which can last for several driving cycles zi. The operating objectives to be considered can include, for example, the driving and elevation profile of a route in subsequent driving cycles, the driving tasks, changes in the vehicle load, etc.

[0030] Vehicle mass, change in battery state of charge (SOC), hydrogen tank content, and the filling station network and availability.

[0031] In addition, stop or standstill phases 40 (Ih=0 to Ih=1, Ih=1 to Ih=2, Ih=2 to Ih=3, Ih=3 to Ih=4, etc.) and the changes in these phases can be taken into account. A change in the stop or standstill phases 40 can, for example, be a change in the vehicle load / mass and the resulting change in the energy or power requirement.

[0032] The prediction horizon 30 is adaptive and dependent on the operating objectives. The prediction horizon 30 depends in particular on the length of the driving cycles or stop or standstill phases 40. For very short distances, several consecutive driving routes can be considered. For very long distances, only a partial route can be considered.

[0033] In Fig. In Figure 1, the adaptive property of the prediction horizon 30 is represented by thick dashed arrows. For example, it can be seen that the prediction horizon 30 lasts longer in the second and third driving cycles z2 and z3 than in the first and second driving cycles z1 and z2. The prediction horizon 30 in the third and fourth driving cycles z3 and z4 is, in turn, shorter than in the first and second driving cycles z1 and z2.

[0034] From the analysis from the adaptive prediction horizon 30, target conditions are derived, in particular optimal stop / standstill target conditions, which are then transmitted or specified to an already running optimized operating strategy.

[0035] The already running optimized operating strategy takes place within a prediction horizon 50 within a current driving cycle zi and can preferably take into account further operating objectives.

[0036] In contrast to the adaptive prediction horizon 30, which depends on the length of the driving cycles or stop or standstill phases 40 and can therefore last shorter or longer, the prediction horizon 50 is shrinking within the current driving cycle zi.

[0037] In Fig. In Figure 1, the shrinking property of the prediction horizon 50 is represented by thin dashed arrows. For example, it can be seen that the optimization in the first driving cycle z1 is performed using a smaller calculation step size. The prediction horizon 50 within the current driving cycle z1 becomes increasingly shorter until the next stop / standstill Ih=1. The faster and more accurate optimization in the prediction horizon 50 enables predictive real-time optimization of additional operating objectives.

[0038] Fig. 2 shows a flowchart of the method steps according to the invention for adjusting the operating mode of a fuel cell system 10.

[0039] According to a first step S1, operating parameters of the fuel cell system 1 are first determined. For example, operating parameters for hydrogen consumption, a fuel cell stack 90, a high-voltage battery 80, or other components of the fuel cell system 10 can be determined.

[0040] In a subsequent step S2, a quality factor is determined for each determined operating parameter based on various calculations (quality function J?? in the form of a cost function and / or energy function), models and / or sensor values, whereby each quality factor represents an operating target variable.

[0041] Subsequently, in a third step S3, a quality function J is calculated based on the determined quality factors, also taking into account the influences of the stop or standstill phases 40 and the changes during these phases. The optimization is performed in the form of a minimization so that trajectories are determined. As a result, the quality function J provides a minimum value. The quality function J is calculated and optimized for an operating segment, in particular for the adaptive prediction horizon 30 that lasts over several driving cycles. The quality function J can also consider only a partial route.

[0042] In a fourth step S4, the results of the calculated quality function of an already running optimized operating strategy of a shrinking prediction horizon 50 are transmitted within a current driving cycle.

[0043] In a fifth step S5, control variables are determined using the transmitted results of the quality function, taking into account the already running optimized operating strategy of the shrinking prediction horizon 50 within the current driving cycle.

[0044] In the final, sixth step S6, the operating mode of the fuel cell system 10 is adjusted based on the determined control variables. More specifically, the operating mode of the fuel cell system 10 is adjusted based on the result of the predictively optimized quality function via a torque of an electric motor or the drive motor of the fuel cell vehicle, by a power split via a DC / DC converter, and / or a start-stop behavior of the compressor system of the fuel cell system 10.

[0045] In Fig.3 shows a fuel cell vehicle 20 according to the invention with a fuel cell system 10 according to the invention and a control device 60. The control device 60 can, for example, receive information from the environment of the fuel cell vehicle 20 through a sensor system 70 or from the driver of the fuel cell vehicle 20 himself through his actions in the fuel cell vehicle 20. The control device 60 can receive further information via a diagnostic device of the energy system, in particular the fuel cell stack 80, the high-voltage battery 90, or other components of the fuel cell vehicle 20. Based on this information and with the aid of a computer program, the control device 60 can set an energy-optimized operating mode of the fuel cell vehicle 20 via the drive train of the fuel cell vehicle 20. Furthermore, an electric motor 100 and a hydrogen tank 110 are optionally shown.

Claims

[1] Method for determining an optimized predictive operating strategy of a fuel cell system (10) of a fuel cell vehicle (20), the method comprising the steps of: - Determination of several operating parameters of the fuel cell system (10) within an adaptive prediction horizon (30) lasting over several driving cycles, - Derivation of target conditions for a next stop / standstill based on the determined operating parameters from the adaptive prediction horizon (30), - transmission of the derived target conditions to an already running optimized operating strategy of a shrinking prediction horizon (50) within a current driving cycle, - Determination of control variables taking into account the already running optimized operating strategy of the shrinking prediction horizon (50) within the current driving cycle, and - Adjusting the operating mode of the fuel cell system (10) based on the determined control variables by a control device (60). [2] Fuel cell system (10) for a fuel cell vehicle (20) with a control device (60) for adjusting the operating mode of the fuel cell system (10) according to claim 1. [3] Fuel cell vehicle (20) with a fuel cell system (10) according to claim 2.

Images