A dynamic control method for mode switching in a reversible solid oxide battery system

Abstract

This invention belongs to the field of energy conversion and storage, specifically relating to a dynamic control method for mode switching of a reversible solid oxide battery system (rSOC). The method includes: dividing the global nonlinear I-P response process model of the rSOC system into two I-P linear sub-models for SOE and SOFC modes; dividing the scheduling power trajectory corresponding to the mode switching scheduling command of the rSOC system into two sub-trajectories for SOE and SOFC modes; predicting the input current according to the switching scheduling sequence, based on each scheduling power in the sub-trajectory of each mode, using the corresponding I-P linear sub-model; determining the optimal operating point based on each scheduling power and prior knowledge; inputting the input current and optimal operating point corresponding to each scheduling power into the battery system, collecting the output power, and re-predicting the input current corresponding to that scheduling power using closed-loop control, thereby achieving input control of the reversible solid oxide battery system. This invention can accurately and quickly respond to mode switching scheduling commands.

Description

Technical Field

[0001] This invention belongs to the field of energy conversion and storage in renewable energy, and more specifically, relates to a dynamic control method for mode switching of a reversible solid oxide battery system. Background Technology

[0002] Stable and inexpensive renewable energy sources are mainly located in remote areas, far from city centers and energy-intensive industrial regions. Therefore, efficient and transportable energy carriers are essential for establishing a link between energy supply and consumption. Hydrogen has the highest calorific value of all fossil and chemical fuels except nuclear fuel, and electrochemical hydrogen production is a key means to fully utilize renewable energy and overcome its disadvantages, such as intermittency and geographical limitations. Unlike alkaline electrolytic cells and proton exchange membrane electrolytic cells, reversible solid oxide cells (rSOCs) have advantages such as high efficiency, low cost, and good scalability, making them an important pathway to coordinating renewable energy generation, storage, conversion, transmission, and end-user consumption.

[0003] Connecting rSOC systems to fluctuating renewable energy microgrids (REMs) requires excellent load cycling and electrolyzer / fuel cell switching performance. Furthermore, frequent dynamic processes have been reported to accelerate the degradation of critical battery components, resulting in rSOCs operating in such environments having a significantly shorter lifespan than expected. Operating such rSOC power plants within renewable energy microgrids for energy conversion presupposes internal material and component safety, primarily considering temperature constraints and stable fuel supply. The ultimate goal is for the rSOC system to quickly and accurately track dispatch commands. All these requirements and challenges have spurred research into control strategies for rSOC systems geared towards electro-hydrogen conversion.

[0004] Currently, domestic research on rSOC systems is still in its initial stages, and no patents or other literature reports have been found regarding the design of mode switching control strategies for rSOC systems. The mode switching phase of an rSOC system is the most complex dynamic process, facing complex thermo-electric dynamic coupling phenomena. On the one hand, the electrochemical reaction in the solid oxide electrolyzer (SOE) mode is endothermic, while the electrochemical reaction is exothermic when switching to the solid oxide fuel cell (SOFC) mode. This contrasting thermal properties pose a significant challenge to the battery materials. Furthermore, the cold box in the system is near room temperature, while the hot box reaches temperatures as high as approximately 850°C, placing extremely high demands on the heat exchange stability of auxiliary components within the system. Moreover, the fuel supply to the system during mode switching is time-delayed; the rapid electrochemical reaction within the stack and the relatively slow gas supply can lead to the stagnation of the electrochemical reaction, endangering stack safety. No control strategy scheme for an rSOC system that comprehensively considers these factors has yet been proposed. Therefore, proposing a mode switching control scheme that accurately and rapidly responds to mode switching scheduling commands has significant engineering value. Summary of the Invention

[0005] To address the shortcomings and improvement needs of existing technologies, this invention provides a dynamic control method for mode switching in a reversible solid oxide battery system. The purpose is to propose a mode switching control scheme that provides a precise and rapid response to mode switching scheduling commands.

[0006] To achieve the above objectives, according to one aspect of the present invention, a dynamic control method for mode switching of a reversible solid oxide battery system is provided, comprising:

[0007] Based on the mode switching scheduling command of the reversible solid oxide battery system to be switched, the scheduling power trajectory corresponding to the mode switching scheduling command is divided into the scheduling power sub-trajectory in SOE mode and the scheduling power sub-trajectory in SOFC mode.

[0008] According to the switching scheduling sequence corresponding to the mode switching scheduling command, the input current is predicted sequentially based on the scheduling power in the linear IP sub-model of each mode using the corresponding mode's IP linear sub-model. Simultaneously with each prediction of the input current, the optimal operating point is determined based on the corresponding scheduling power and prior knowledge. The input current and the optimal operating point corresponding to each scheduling power are input into the reversible solid oxide battery system, the output power is collected, and the input current corresponding to the scheduling power is re-predicted in a closed-loop control manner to realize input control of the reversible solid oxide battery system and complete the dynamic control of mode switching of the reversible solid oxide battery system.

[0009] The IP linear sub-models for each mode are obtained in advance by dividing the global nonlinear IP response process model of the reversible solid oxide battery system into two IP linear sub-models, SOE mode and SOFC mode, at the zero current point or hot standby point.

[0010] The beneficial effects of this invention are as follows: This invention proposes to pre-divide the global nonlinear IP response process model of the reversible solid oxide battery system into two IP linear sub-models at the zero current point or hot standby point: SOE mode and SOFC mode. During control execution, the scheduling power trajectory corresponding to the mode switching scheduling command is divided into scheduling power sub-trajectories in SOE mode and SOFC mode. Furthermore, the corresponding IP linear sub-model is used to predict the input current of the output battery system for each mode's scheduling power sub-trajectory. In other words, this invention divides the input current controller into two independent parts. Based on the corresponding IP linear sub-models of their respective configuration modes, feedback closed-loop control of the input current is performed. Since the global nonlinear IP response process model is nonlinear, existing mode-switching control schemes suffer from difficulties in modeling and identification, making it difficult to obtain an accurate global model. This invention divides the global nonlinear IP response process model into two linear IP sub-models—SOE mode and SOFC mode—at the zero-current point or hot standby point, and independently predicts the dispatched power trajectory, effectively overcoming this problem. Furthermore, while predicting the input current, this invention also predicts the optimal operating point (fuel fraction X) based on prior knowledge for each dispatched power. fuel (Fuel utilization rate FU, excess air ratio AR). In other words, this invention obtains all input quantities of the entire battery system through feedback control and feedforward control. From one perspective, this invention's solution achieves all operational variables corresponding to each scheduling power during mode switching in a decoupled manner, avoiding the problem of blindly increasing the complexity of the predictive control model due to multi-dimensional inputs. It has a smaller computational load and can improve the accuracy of input quantities in response to scheduling commands. From another perspective, this invention is also a cooperative control method, specifically multi-path and cooperative control, ensuring the temperature and fuel safety of the rSOC system during mode switching, while also considering the speed of load tracking during mode switching.

[0011] Furthermore, the IP linear sub-model corresponding to each mode is constructed as follows:

[0012] Multiple random non-periodic currents of the reversible solid oxide battery system in this mode and the steady-state output power corresponding to each input current are collected offline in advance. The IP linear sub-model corresponding to this mode is constructed through model identification.

[0013] A further beneficial effect of the present invention is that the IP linear sub-model is pre-built using offline data, which can ensure the real-time performance and accuracy of model predictive control.

[0014] Furthermore, the prior knowledge is a mapping model between output power and the corresponding optimal operating point, which is constructed as follows:

[0015] The optimal operating point set for each output power within the operating space of the reversible solid oxide battery system is obtained, including each output power point in SOE mode and SOFC mode; wherein, the optimal operating point is obtained by discretizing, traversing, and optimizing to determine the optimal fuel quantity, fuel fraction, and air quantity corresponding to each output power.

[0016] A further beneficial effect of the present invention is that prior knowledge is pre-constructed through offline data, which can ensure the real-time performance and accuracy of model predictive control.

[0017] The present invention also provides a dynamic control system for mode switching of a reversible solid oxide battery system, comprising:

[0018] The scheduling command processing module is used to divide the scheduling power trajectory corresponding to the mode switching scheduling command of the reversible solid oxide battery system to be switched into a scheduling power sub-trajectory under SOE mode and a scheduling power sub-trajectory under SOFC mode according to the mode switching scheduling command; and input each scheduling power in the scheduling power sub-trajectory under each mode into the feedforward controller and the corresponding mode sub-controller in the dual model prediction controller in sequence according to the switching scheduling timing corresponding to the mode switching scheduling command.

[0019] The dual-model predictive controller includes sub-controllers configured with IP linear sub-models in SOE mode and SOFC mode. These sub-controllers are used to obtain the optimal current input based on the received scheduling power through rolling optimization and feedback correction, and then input it into the reversible solid oxide battery system. During this process, the output power of the reversible solid oxide battery system is collected to form a feedback closed-loop control, which further corrects the input current, enabling the rSOC system to quickly reach the target power. The IP linear sub-models for each mode are obtained in advance by dividing the global nonlinear IP response process model of the reversible solid oxide battery system into two linear IP sub-models for SOE mode and SOFC mode at the zero current point or hot standby point.

[0020] The feedforward controller is used to determine the optimal operating point based on the received scheduling power and prior knowledge, and input it into the reversible solid oxide battery system.

[0021] The beneficial effects of this invention are as follows: This invention proposes a cooperative control system based on a combination of dual model predictive control (DM-MPC) and a feedforward controller based on the optimal operating point. One mode corresponds to one DM-MPC, and each DM-MPC feedback controls the input current of the rSOC system; the feedforward controller controls the gas flow rate (fuel quantity, air quantity) entering the stack according to the dispatched power. The control process involves all operating variables (current, fuel fraction, fuel utilization rate, air excess ratio) in a decoupled manner, achieving control over all operating variables corresponding to each dispatched power during mode switching. This results in a smaller computational load and improves the accuracy of input quantities in response to dispatch commands. Furthermore, it avoids the problem of blindly increasing the complexity of the predictive control model due to multi-dimensional inputs, ensuring the temperature and fuel safety of the rSOC system during mode switching, while also considering the speed of load tracking during mode switching.

[0022] Furthermore, the IP linear sub-model corresponding to each mode is constructed as follows:

[0023] Multiple random aperiodic currents and the steady-state output power corresponding to each input current of the reversible solid oxide battery system in this mode are pre-collected offline. An IP linear sub-model corresponding to this mode is constructed through model identification. A further beneficial effect of this invention is that the IP linear sub-model, constructed in advance using offline data, ensures the real-time performance and accuracy of model predictive control.

[0024] Furthermore, the prior knowledge is a mapping model between output power and the corresponding optimal operating point, which is constructed as follows:

[0025] The optimal operating point set for each output power within the operating space of the reversible solid oxide battery system is obtained, including each output power point in SOE mode and SOFC mode. The optimal operating point is obtained by discretizing, traversing, and optimizing to determine the optimal fuel quantity, fuel fraction, and air quantity corresponding to each output power. A further beneficial effect of this invention is that prior knowledge is pre-constructed using offline data, ensuring the real-time performance and accuracy of model predictive control.

[0026] The present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed by a processor, it controls the device where the storage medium is located to perform a dynamic control method for mode switching of a reversible solid oxide battery system as described above.

[0027] In summary, the above-described technical solutions conceived in this invention can achieve the following beneficial effects:

[0028] This invention proposes a collaborative control scheme based on a combination of dual model predictive control (DM-MPC) and a feedforward controller based on the optimal operating point. One mode corresponds to one DM-MPC, and each DM-MPC feedback controls the input current of the rSOC system. The feedforward controller controls the gas flow rate (fuel quantity, air quantity) entering the stack according to the dispatched power. The control process involves all operating variables (current, fuel fraction, fuel utilization rate, air excess ratio) in a decoupled manner. Existing mode-switching control schemes suffer from nonlinear predictive control optimization problems that are nonconvex, computationally intensive, and prone to getting trapped in local optima. This invention divides the dispatched power trajectory corresponding to the dispatch command according to the mode, and each mode is independently predictively controlled. This avoids the problem of blindly increasing the complexity of the predictive control model due to multi-dimensional inputs, resulting in less computation. It also avoids the difficulties in modeling and identifying the nonlinear model of global load tracking in the rSOC system, enabling rapid and accurate acquisition of the battery system's input quantities. Furthermore, this invention is a collaborative control method, which ensures temperature and fuel safety of the rSOC system during mode switching while also considering the speed of load tracking during mode switching. Attached Figure Description

[0029] Figure 1 A block diagram of a dynamic control method for mode switching of a reversible solid oxide battery system provided in an embodiment of the present invention;

[0030] Figure 2 This is a schematic diagram illustrating the working principle of the feedback control module provided in an embodiment of the present invention.

[0031] Figure 3 IP response diagram of a reversible solid oxide battery system switching from SOE mode to SOFC mode, provided in an embodiment of the present invention;

[0032] Figure 4 for Figure 3 The graphs showing the changes in the maximum cell temperature Max.T and the maximum cell temperature gradient Max.Tgrad over time;

[0033] Figure 5 for Figure 3 The graph showing the change of H2 mole fraction over time at the corresponding battery terminal node 5;

[0034] Figure 6 IP response diagram of a reversible solid oxide battery system switching from SOFC mode to SOE mode, provided in an embodiment of the present invention;

[0035] Figure 7 for Figure 6 The graphs showing the changes in the maximum cell temperature Max.T and the maximum cell temperature gradient Max.Tgrad over time;

[0036] Figure 8 for Figure 6 The graph shows the change of H2 mole fraction over time at the corresponding battery terminal node 5. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0038] Example 1

[0039] A dynamic control method for mode switching in a reversible solid oxide battery system, such as Figure 1 As shown, it includes:

[0040] Based on the mode switching scheduling command of the reversible solid oxide battery system to be switched, the scheduling power trajectory corresponding to the mode switching scheduling command is divided into the scheduling power sub-trajectory in SOE mode and the scheduling power sub-trajectory in SOFC mode.

[0041] According to the switching scheduling sequence corresponding to the mode switching scheduling command, the input current is predicted sequentially based on the scheduling power in the linear IP sub-model of each mode using the corresponding mode's IP linear sub-model. Simultaneously with each prediction of the input current, the optimal operating point is determined based on the corresponding scheduling power and prior knowledge. The input current and the optimal operating point corresponding to each scheduling power are input into the reversible solid oxide battery system, the output power is collected, and the input current corresponding to the scheduling power is re-predicted in a closed-loop control manner to realize input control of the reversible solid oxide battery system and complete the dynamic control of mode switching of the reversible solid oxide battery system.

[0042] The IP linear sub-models for each mode are obtained in advance by dividing the global nonlinear IP response process model of the reversible solid oxide battery system into two linear IP sub-models, SOE mode and SOFC mode, at the zero current point or hot standby point.

[0043] This embodiment proposes a cooperative control strategy. Considering the temperature constraints and fuel safety of the rSOC system during mode switching, current, reactant mole fraction, fuel utilization rate, and air excess ratio are incorporated into the control variables. The technical solution adopted by this embodiment to solve its technical problem is as follows: First, it includes a feedforward controller based on the Optimal Operating Point (OOP) and two DM-MPCs. The feedforward controller adjusts the input parameters (fuel fraction, fuel utilization rate, and air excess ratio) of the battery system according to the scheduled power to control the amount of fuel and air entering the battery system. The two current feedback controllers are composed of DM-MPCs. The cooperative control strategy can be found in [reference needed]. Figure 1 .

[0044] Among them, the aforementioned closed-loop control can be achieved through Figure 2 The process shown is implemented.

[0045] As a preferred implementation scheme, the IP linear sub-model corresponding to each mode can be constructed as follows:

[0046] Multiple random non-periodic currents and the steady-state output power corresponding to each input current of the reversible solid oxide battery system in this mode are collected offline in advance. The corresponding IP linear sub-model is constructed through model identification.

[0047] Mode switching in an rSOC system is a continuous process with significant load fluctuations. The system operates under a wide range of manipulated variables, exhibiting pronounced nonlinear characteristics. This makes it difficult to describe the system's global response characteristics using a simple model (transfer function or state-space equation). Furthermore, the system's input parameters affect its internal security and transient response performance; considering multi-dimensional inputs inevitably increases the complexity of controller design. The key to the method in this embodiment lies in designing two predictive control modules, each independently based on its configured IP linear sub-model, to perform feedback closed-loop control of the input current, such as... Figure 1 It consists of a dual model predictive control structure. Based on the operating mode of the rSOC system, the globally nonlinear load tracking (IP) model is divided into two linear sub-models, namely the IP response models in SOE mode and SOFC mode. These two models are obtained through model identification.

[0048] The stochastic aperiodic currents and corresponding steady-state power of the two modes were used as the input and output for model identification. The IP data was identified as transfer functions using Matlab's System Identification Toolbox™. To improve the accuracy of the identified models, the IP data was normalized before identification. Finally, the identified linear sub-models of the IP for the SOE and SOFC modes were obtained:

[0049]

[0050]

[0051] In the formula, U(s) and Y(s) represent the input and output of the Laplace transform, respectively.

[0052] The accuracy of the identification model is generally tested by the goodness of fit (Fit) and the mean squared error (MSE), which are defined as follows:

[0053]

[0054] In the formula, y represents the output of the system model; This represents the output of the identification model; N represents the total amount of data; and j represents the data index.

[0055] Table 1 shows the results of the test of the identification model. The data shows that the transfer function well characterizes the system response and can be used for subsequent controller design.

[0056] Table 1 Evaluation metrics for the identification model

[0057]

[0058] Furthermore, closed-loop control can be achieved through... Figure 2 The process is shown below. Model predictive control (MPC), as a proactive process control strategy, achieves global optimal control by dynamically calculating local optima. In each control cycle, MPC continuously solves the optimization problem ψ based on n-step predictions of future system power.

[0059]

[0060]

[0061] Δu(k)=C T (A T A+λI) -1 A T (W-Y0);

[0062] In the formula, W=[ω(k+1),ω(k+2),...,ω(k+n)] T λ(j) represents the reference power trajectory; λ(j) represents the control weight coefficient; n and m represent the prediction interval and control interval, respectively.

[0063] The internal model of MPC is the aforementioned constructed linear sub-model of IP. In each control cycle, only the first control variable is applied to the rSOC system, and the future output of the system is predicted according to the following formula:

[0064]

[0065] In the formula, and Y p0 These represent the output at p future time points under the influence of the control variable, and the output at p future time points without the influence of the control variable.

[0066] When the control action is applied at time k, the power output within the prediction interval is calculated by the internal model embedded in the MPC. At this time, the error between the measured value and the predicted value serves as a feedback correction loop. The error at future times is calculated using a weighted method to compensate for the error in the model prediction. The corrected prediction vector is as follows:

[0067]

[0068] In the formula, H represents the error correction matrix. Starting from time k, the initial prediction value is rolled forward over time. The initial prediction can be expressed as follows:

[0069]

[0070] In the formula, S represents the shift matrix.

[0071] The switching time between single-pole double-throw (SPO) control models depends on the zero-power point during mode switching. For example, when the total power consumption P in SOE mode... SOE In t i If the time is set to 0, then control... Figure 1 The single-pole switch in the middle jumps to the other side, indicating that the invention is a mode switching control strategy for overcurrent zero point.

[0072] As a preferred implementation scheme, the prior knowledge is a mapping model between output power and the corresponding optimal operating point, which is constructed as follows:

[0073] The optimal operating point set for each output power in the operating space of the reversible solid oxide battery system is obtained, including each output power point in SOE mode and SOFC mode. The optimal operating point is obtained by discretizing, traversing, and optimizing to determine the optimal fuel quantity, fuel fraction, and air quantity corresponding to each output power.

[0074] Optimal operating points (OOPs) are defined as the set of operating points that maximize system efficiency while satisfying internal temperature constraints (maximum temperature Max.T and maximum temperature gradient Max.Tgard constraints) and external load power / hydrogen production rate requirements. Mode switching based on the optimal steady-state performance of the rSOC system's OOPs is beneficial for control strategy design. Specifically, the system's input parameters—current I, reactant mole fraction X—are used in this process. fuelFuel efficiency (FU) and excess air ratio (AR) are discrete within their operating range, with a discrete accuracy of ΔI = 2A and ΔX. fuel =0.05, ΔFU=0.025, ΔAR=0.25. Using a tree diagram, the system power and efficiency under each set of operating parameters are obtained. Based on the set temperature constraints, the maximum temperature limit Max.T and the maximum temperature gradient Max.Tgard for each cell are respectively:

[0075] 873K≤Max.T≤1173K;

[0076] Max.Tgrad≤10K / cm;

[0077] Within the set of operating points that satisfy temperature constraints, find the set of operating points X that maximizes the efficiency of the rSOC system under each operating condition. i ={I i ,X fuel,i ,FU i AR i ,},X i This refers to the optimal operating point under the corresponding operating condition P. In other words, under operating condition X i The steady-state electrical characteristics of the system are:

[0078] E(X i )=(P SOE ,P SOFC U S ,η SOE / SOFC );

[0079] Let P SOE ,P SOFC,net =P sys This means that within the thermal safety range, the system's H2 production rate or net power generation for external demand can be used to determine the optimal operating point under these conditions, based on P. sys Sure:

[0080]

[0081] like Figure 1 The feedforward controller is based on the system's scheduling power P set Select the corresponding operation point X for the specific working condition in the OOPs knowledge base. i 3 ={X fuel,i ,FU i AR i ,} is used as the input parameter for manipulation into the system.

[0082] The definition of optimal efficiency is expressed as follows:

[0083] The efficiency definitions for SOE and SOFC modes are as follows:

[0084]

[0085] In the formula, and These represent the production and consumption of H2, respectively. The lower heating value of hydrogen is expressed as 241.83 kJ / mol. -1 ;P SOE and P SOFC These represent the total power in SOE mode and the net power generated in SOFC mode, respectively.

[0086] Example 2

[0087] A dynamic control system for mode switching of a reversible solid oxide battery system, comprising:

[0088] The scheduling command processing module is used to divide the scheduling power trajectory corresponding to the mode switching scheduling command of the reversible solid oxide battery system to be switched into a scheduling power sub-trajectory under SOE mode and a scheduling power sub-trajectory under SOFC mode according to the mode switching scheduling command; and according to the switching scheduling sequence corresponding to the mode switching scheduling command, sequentially input each scheduling power in the scheduling power sub-trajectory of each mode into the feedforward controller and the corresponding mode sub-controller in the dual model prediction controller.

[0089] The dual-model predictive controller includes sub-controllers configured with IP linear sub-models in SOE and SOFC modes. These sub-controllers are used to obtain the optimal current input based on the received scheduling power through rolling optimization and feedback correction, and then input it into the reversible solid oxide battery system. During this process, the output power of the reversible solid oxide battery system is acquired to form a feedback closed-loop control, which further corrects the input current, enabling the rSOC system to quickly reach the target power. The IP linear sub-models for each mode are obtained in advance by dividing the global nonlinear IP response process model of the reversible solid oxide battery system into two IP linear sub-models for SOE and SOFC modes at the zero current point or hot standby point.

[0090] A feedforward controller is used to determine the optimal operating point based on the received scheduling power and prior knowledge, and then input it into the reversible solid oxide battery system.

[0091] For details on the structure of the dual-model predictive controller, please refer to the appendix. Figure 2 .

[0092] This invention proposes a cooperative control system based on a combination of dual model predictive control (DM-MPC) and a feedforward controller based on the optimal operating point. One mode corresponds to one DM-MPC, and each DM-MPC feedback controls the input current of the rSOC system. The feedforward controller controls the gas flow rate (fuel quantity, air quantity) entering the stack according to the dispatched power. The control process involves all operating variables (current, fuel fraction, fuel utilization rate, air excess ratio) in a decoupled manner, achieving control over all operating variables corresponding to each dispatched power during mode switching. This reduces computational complexity and improves the accuracy of input responses to dispatch commands. Furthermore, it avoids the problem of blindly increasing the complexity of the predictive control model due to multi-dimensional inputs, ensuring temperature and fuel safety of the rSOC system during mode switching, while also considering the speed of load tracking during mode switching.

[0093] As a preferred implementation method, the IP linear sub-model corresponding to each mode is constructed as follows:

[0094] Multiple random non-periodic currents and the steady-state output power corresponding to each input current of the reversible solid oxide battery system in this mode are collected offline in advance. The corresponding IP linear sub-model is constructed through model identification.

[0095] As a preferred implementation method, the prior knowledge is a mapping model between output power and the corresponding optimal operating point, which is constructed as follows:

[0096] The optimal operating point set for each output power in the operating space of the reversible solid oxide battery system is obtained, including each output power point in SOE mode and SOFC mode. The optimal operating point is obtained by discretizing, traversing, and optimizing to determine the optimal fuel quantity, fuel fraction, and air quantity corresponding to each output power.

[0097] The relevant technical solutions are the same as in Embodiment 1, and will not be repeated here.

[0098] To evaluate the performance of the controller proposed in this invention, tests were conducted based on a system model with rSOC, and the results were compared with those of a PID controller with optimized parameters, as detailed below:

[0099] Example 1: A reversible solid oxide battery system switches from SOE mode to SOFC mode.

[0100] During the transition of a reversible solid oxide battery system from SOE mode to SOFC mode, the control method of this invention and a PID controller with optimized parameters are used respectively to obtain the changes in current and power over time, as follows: Figure 3As shown; the changes in the maximum cell temperature Max.T and the maximum cell temperature gradient Max.Tgrad over time are as follows. Figure 4 As shown; the change of H2 mole fraction at battery terminal node 5 over time, as follows: Figure 5 As shown.

[0101] That is, Figures 3-5 This describes the dynamic response process of the rSOC system when switching from SOE mode (electrolyzer) to SOFC mode (fuel cell), determined by the total electrolysis power P. SOE =8kW switching to net discharge power P SOFC,net =5kW. In the figure, DM-MPC represents the scheme of this invention; SOA-PID stands for Searchoptimization algorithms-PID, which represents the reference control scheme (i.e., the PID controller with optimized parameters); Ref represents the reference switching trajectory / scheduling command; Current represents the current; Max.T represents the maximum temperature of the cell; Max.Tgrad represents the maximum temperature gradient of the cell; Molarfractionof H2inNode5 represents the H2 mole fraction at the end node 5 of the cell; the horizontal axis Time represents time;

[0102] Figure 3 The dynamic response processes of DM-MPC and SOA-PID power tracking are shown under the control results proposed in this invention. It can be observed that although both SOE mode power tracking exhibit overshoot, DM-MPC has a lower overshoot and reaches steady state faster. At the switching point of t=350s, after switching to SOFC mode, SOA-PID shows faster load tracking and a larger current change rate, but still exhibits an upward overshoot. During IP tracking, the changes in I and P are positively correlated, with current I having a key impact on power fluctuations. Figure 4 The display shows the temperature dynamics during the switch from SOE to SOFC mode. It can be seen that DM-MPC exhibits superior performance due to its smaller temperature fluctuations compared to SOA-PID. Figure 5 The figure shows the change in H2 mole fraction at node 5 of the battery after switching to SOFC mode (load increased after t = 350 s). The hydrogen concentration at node 5 is related to fuel supply and power fluctuations. The greater the power change (the faster the current change), the more likely it is to cause untimely supply and thus "fuel shortage," endangering the safety of the stack. As shown in the figure, DM-MPC controls the current change rate more slowly, effectively ensuring fuel safety.

[0103] Example 2: Switching a reversible solid oxide battery system from SOFC mode to SOE mode

[0104] During the transition of a reversible solid oxide battery system from SOFC mode to SOE mode, the control method of this invention and a PID controller with optimized parameters are used respectively to obtain the changes in current and power over time, as follows: Figure 6 As shown; the changes in the maximum cell temperature Max.T and the maximum cell temperature gradient Max.Tgrad over time are as follows. Figure 7 As shown; the change of H2 mole fraction at battery terminal node 5 over time, as follows: Figure 8 As shown.

[0105] That is, Figures 6-8 This represents the dynamic response process of the rSOC system when switching from SOFC mode (fuel cell) to SOE mode (electrolyte cell), as shown by the net discharge power P. SOFC,net =5kW switched to total electrolysis power P SOE =8kW. In the figure, Molarfractionof H2Oin Node5 represents the molar fraction of H2O at the end node 5 of the battery.

[0106] Figure 5 The IP curve for load tracking power scheduling is displayed, highlighting the speed of SOA-PID power tracking as fuel safety is no longer a concern during load adjustments. SOA-PID achieves faster load tracking and reaches steady state more quickly even with slight overshoot. Similarly, DM-MPC performs better after switching to SOE mode. Additionally, as... Figure 6 As shown, after switching to SOE mode, the sharp peak in the SOA-PID scheme occurs because the heater turns on at t=350s. This is due to the different overcurrent change rates between nodes. Figure 7 As shown, DM-MPC limits the rate of current change, making it safer than SOA-PID.

[0107] In summary, the cooperative control structure proposed in this invention consists of a feedback loop composed of a dual model predictive controller (DM-MPC) and a feedforward control loop. Specifically, based on the power scheduling command of the rSOC system, the scheduling sequence is switched. A dual model predictive controller structure composed of two IP sub-model predictive controllers (MPCs) acts as the feedback loop to control the current input to the rSOC system according to the scheduled power. The fuel quantity, fuel fraction, and air quantity input to the system are determined by a feedforward controller. The feedforward loop contains prior knowledge based on optimal operating points (OOPs). The knowledge base contains the correspondence between system power / operating conditions and inputs, specifically meaning that at any system power, there exists a set of optimal input operations that maximize the efficiency of the rSOC system.

[0108] The control flow is as follows: The operating mode of the rSOC system is determined based on the power dispatch command. If it is in SOE mode, the corresponding SOE IP linear sub-model within the DM-MPC is invoked, and vice versa. Simultaneously, for any given power / operating condition, the rSOC system has a uniquely determined optimal system input. Similarly, based on the dispatched power, the feedforward controller controls other operational variables input to the rSOC system: fuel quantity, fuel fraction, and air quantity. This constitutes a cooperative control scheme for safe and rapid mode switching of the rSOC system.

[0109] This invention employs a collaborative scheme combining a feedback structure of dual-model predictive control and a feedforward control based on optimal operating points (OOPs) to ensure the thermal and fuel safety of an rSOC system during mode switching and to rapidly reach load power. The advantages of this invention are: 1. It solves the problems existing in predictive control applications of the global nonlinear model of the rSOC system: non-convexity, high computational burden, susceptibility to local convergence, poor real-time performance, and high difficulty in modeling and identification. 2. The OOPs-based feedforward control reduces the complexity of the predictive model by decreasing the input-output dimension of the predictive control model. 3. To verify the beneficial effects of the composite control scheme, a simple bidirectional case of an rSOC system switching between SOE / SOFC modes was designed, and its results were compared with those of a PID controller optimized by a search optimization algorithm (SOA) (SOA-PID). Figures 3 to 8 As shown, the results demonstrate that this invention has excellent global performance in the mode switching control of the rSOC system, effectively ensuring thermal safety and fuel safety, and achieving rapid load tracking.

[0110] Example 3

[0111] A computer-readable storage medium includes a stored computer program, wherein when the computer program is executed by a processor, it controls the device where the storage medium is located to perform a dynamic control method for mode switching of a reversible solid oxide battery system as described in Embodiment 1.

[0112] The relevant technical solutions are the same as in Embodiment 1, and will not be repeated here.

[0113] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A dynamic control method for mode switching in a reversible solid oxide battery system, characterized in that, include: Based on the mode switching scheduling command of the reversible solid oxide battery system to be switched, the scheduling power trajectory corresponding to the mode switching scheduling command is divided into the scheduling power sub-trajectory in SOE mode and the scheduling power sub-trajectory in SOFC mode. According to the switching scheduling sequence corresponding to the mode switching scheduling command, the input current is predicted sequentially based on the scheduling power in the linear IP sub-model of each mode using the corresponding mode's IP linear sub-model. Simultaneously with each prediction of the input current, the optimal operating point is determined based on the corresponding scheduling power and prior knowledge. The input current and the optimal operating point corresponding to each scheduling power are input into the reversible solid oxide battery system, the output power is collected, and the input current corresponding to the scheduling power is re-predicted in a closed-loop control manner to realize input control of the reversible solid oxide battery system and complete the dynamic control of mode switching of the reversible solid oxide battery system. Among them, the IP linear sub-models of each mode are obtained in advance by dividing the global nonlinear IP response process model of the reversible solid oxide battery system into two IP linear sub-models, SOE mode and SOFC mode, at the zero current point or hot standby point. The prior knowledge is a mapping model between output power and the corresponding optimal operating point, which is constructed as follows: The optimal operating point set for each output power within the operating space of the reversible solid oxide battery system is obtained, including each output power point in SOE mode and SOFC mode; wherein, the optimal operating point is obtained by discretizing, traversing, and optimizing to determine the optimal fuel quantity, fuel fraction, and air quantity corresponding to each output power.

2. The dynamic control method for mode switching of a reversible solid oxide battery system according to claim 1, characterized in that, The construction method of the IP linear sub-model corresponding to each mode is as follows: Multiple random non-periodic currents of the reversible solid oxide battery system in this mode and the steady-state output power corresponding to each input current are collected offline in advance. The IP linear sub-model corresponding to this mode is constructed through model identification.

3. A dynamic control system for mode switching in a reversible solid oxide battery system, characterized in that, include: The scheduling command processing module is used to divide the scheduling power trajectory corresponding to the mode switching scheduling command of the reversible solid oxide battery system to be switched into a scheduling power sub-trajectory under SOE mode and a scheduling power sub-trajectory under SOFC mode according to the mode switching scheduling command; and according to the switching scheduling sequence corresponding to the mode switching scheduling command, sequentially input each scheduling power in the scheduling power sub-trajectory of each mode into the feedforward controller and the corresponding mode sub-controller in the dual model prediction controller. The dual-model predictive controller includes sub-controllers configured with IP linear sub-models in SOE mode and SOFC mode. These sub-controllers are used to obtain the optimal current input based on the received scheduling power through a rolling optimization stage and a feedback correction stage, and then input this current into the reversible solid oxide battery system. During this process, the output power of the reversible solid oxide battery system is acquired to form a feedback closed-loop control, further correcting the input current so that the rSOC system can quickly reach the target power. The IP linear sub-models for each mode are pre-obtained by dividing the global nonlinear IP response process model of the reversible solid oxide battery system into two IP linear sub-models for SOE mode and SOFC mode at the zero current point or hot standby point. The feedforward controller is used to determine the optimal operating point based on the received scheduling power and prior knowledge, and input it into the reversible solid oxide battery system.

4. The dynamic control system for mode switching of a reversible solid oxide battery system according to claim 3, characterized in that, The construction method of the IP linear sub-model corresponding to each mode is as follows: Multiple random non-periodic currents of the reversible solid oxide battery system in this mode and the steady-state output power corresponding to each input current are collected offline in advance. The IP linear sub-model corresponding to this mode is constructed through model identification.

5. The dynamic control system for mode switching of a reversible solid oxide battery system according to claim 3, characterized in that, The prior knowledge is a mapping model between output power and the corresponding optimal operating point, which is constructed as follows: The optimal operating point set for each output power within the operating space of the reversible solid oxide battery system is obtained, including each output power point in SOE mode and SOFC mode; wherein, the optimal operating point is obtained by discretizing, traversing, and optimizing to determine the optimal fuel quantity, fuel fraction, and air quantity corresponding to each output power.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed by a processor, it controls the device where the storage medium is located to perform a dynamic control method for mode switching of a reversible solid oxide battery system as described in claim 1 or 2.

Images