Electrified railway hybrid energy storage elastic adaptive time domain energy management method considering extreme conditions

Abstract

The application discloses an electrified railway hybrid energy storage elastic adaptive time domain energy management method considering extreme conditions. First, the traction load, photovoltaic output, energy storage state of charge and power grid operation information are collected, and abnormal load and extreme weather event detection is carried out based on statistical threshold and change rate; second, a double-mode adaptive prediction time domain adjustment mechanism is constructed, which shrinks the prediction time domain in high-frequency fluctuation and expands the prediction time domain according to the severity when the extreme event is triggered; then, a mixed integer linear programming model containing a dynamic state of charge (SOC) safety margin is established, the battery and super capacitor charging and discharging are coordinated, and a safe energy storage defense line is constructed in the conventional state and the deep discharge capacity is released in the crisis state; finally, a rolling optimization output energy storage control instruction is adopted. The method can improve the energy supply continuity, risk resistance, recovery elasticity and operation economy of the traction power supply system under complex disturbance.

Description

Technical Field

[0001] This invention relates to the fields of rail transit microgrids, new energy and energy storage optimization scheduling technology, and more specifically, to a flexible adaptive time-domain energy management method for hybrid energy storage in electrified railways that takes into account extreme conditions, particularly an energy management method based on event-driven adaptive time-domain model predictive control (ED-AH-MPC). Background Technology

[0002] Rail transit, exemplified by electrified railways, is key to achieving low-carbon development in the transportation sector. With the advancement of integrated transportation and energy development, the introduction of renewable energy sources such as photovoltaics, hybrid energy storage systems (HESS) composed of supercapacitors and batteries, and railway power regulators (RPCs) to construct smart railway microgrids in traction power supply systems (TPSS) have become important solutions for improving energy utilization efficiency and power quality. The introduction of hybrid energy storage systems not only enhances the absorption of regenerative braking energy but also reduces overall operating costs.

[0003] However, with the increasing frequency of low-probability, high-impact (HILP) events such as extreme weather, the operational flexibility of traction power supply systems faces severe challenges. Extreme conditions can lead to a precipitous drop in photovoltaic output and a sharp surge in traction load. Traditional microgrid energy management methods based on model predictive control (MPC) typically employ a fixed prediction time domain. When facing real-world operating conditions with high randomness and abrupt changes, a fixed time domain inherently suffers from a "horizon paradox": if a shorter prediction time domain is used, the system has better computational speed and can reduce prediction errors, but it suffers from a serious "shortsightedness" problem, failing to anticipate and store energy for upcoming extreme weather or sustained load peaks; if a longer prediction time domain is used, although it provides a global perspective, it often leads to severe accumulation of prediction errors and a huge computational burden when encountering high-frequency load fluctuations, resulting in suboptimal dispatch commands.

[0004] Furthermore, due to the inherent characteristics of railway microgrid systems—low inertia and limited available energy—traditional fixed static state of charge (SOC) safety margin mechanisms and protection logic are too rigid. They cannot achieve dynamic and flexible responses to varying degrees of disturbance severity, and are prone to prematurely depleting reserve energy during small fluctuations or failing to unlock deep discharge limits in time during severe crises. Therefore, there is an urgent need for a highly flexible system energy management method that can automatically scale the optimization perspective based on the fluctuation conditions and external crisis events faced by the system, and dynamically adjust the system's safety boundaries. Summary of the Invention

[0005] To address the problem that existing microgrid energy management strategies struggle to balance long-term safety and short-term accuracy in the face of extreme weather and abnormal load shocks, the present invention aims to provide a resilient adaptive time-domain energy management method for hybrid energy storage in electrified railways under extreme conditions, thereby improving the system's self-immunity and resilience.

[0006] This invention proposes a resilient event-driven adaptive temporal energy management method, which mainly includes the following steps:

[0007] (1) Operation status monitoring and multi-source integrated event detection: Real-time collection of grid status, traction load power, photovoltaic power generation power and energy storage charge status; based on statistical rules and rate of change characteristics, perform event scanning to determine whether high-frequency fluctuations, extreme load surges or extreme weather-induced source-load cliff drops are triggered.

[0008] (2) Dual-mode adaptive optimization time domain calculation: By parallel calculation of the comprehensive volatility index of traction load and photovoltaic power output fluctuation, the volatility-based "contraction mode" or the event detection-based "expansion mode" is driven to adaptively calculate the optimal prediction time domain length that best fits the current moment.

[0009] (3) Construct an energy storage scheduling optimization model with dynamic safety margin: take minimizing the grid-connected power purchase cost, energy storage loss and related penalty terms in a specific prediction time domain as the objective function, set equipment physical constraints and power balance constraints, and design an energy storage state of charge (SOC) safety margin management mechanism with "energy storage-release" dual-state switching.

[0010] (4) Introducing rolling optimization solution and issuing instructions: In the current prediction time domain, the above model is solved periodically using the mixed integer linear programming (MILP) method, and the optimal hybrid energy storage action instruction of the current time step is issued and executed.

[0011] The beneficial effects of this invention are as follows:

[0012] 1. An innovative event-driven dual-mode time-domain adaptive mechanism (ED-AH-MPC) was proposed, which solved the "time-domain paradox" problem. It actively shrinks the time domain to filter noise and accumulate errors during high-frequency oscillations and actively extends the time domain to capture extreme interference before a crisis occurs, which greatly improves the system resilience under extreme events.

[0013] 2. A dynamic SOC safety margin mechanism is introduced to build a protective barrier when facing high uncertainty in daily life, raising the lower limit of availability; when a certain extreme event occurs, the barrier is automatically broken to unlock deep discharge capacity, effectively reducing the duration of low SOC risk of the system and improving the system's elastic energy supply level. Attached Figure Description

[0014] Figure 1 This is the main flowchart of the adaptive time-domain energy management method for hybrid energy storage in electrified railways based on the present invention.

[0015] Figure 2 This is a physical architecture diagram of a traction power supply microgrid system with hybrid energy storage according to an embodiment of the present invention;

[0016] Figure 3 This is a diagram illustrating the detection, identification, and early warning results under an abnormal and sudden surge in traction load in an embodiment of the present invention.

[0017] Figure 4 This is a simulation and evolution diagram of the photovoltaic power output drop caused by extreme weather in an embodiment of the present invention.

[0018] Figure 5 This is a comparison chart of the system comprehensive response curves of the adaptive method proposed in this invention under extreme weather events;

[0019] Figure 6 This diagram compares the duration of the low-SOC risk caused by the adaptive method proposed in this invention with that of the traditional fixed time-domain predictive control strategy under extreme weather conditions. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to specific steps.

[0021] Step 1: Real-time monitoring of operational status and detection of extreme events. Establish a multi-source event detection mechanism to scan the data stream of the track microgrid.

[0022] Calculate the mean traction load distribution with standard deviation and in the auxiliary time window ( Perform radar-style look-ahead scanning to define trigger signals for abnormal peak load surge events. The judgment logic is: when the condition is satisfied When an abnormal event signal is triggered, that is... It is not triggered under other circumstances, i.e., $F_{load}(t)=0$.

[0023] Monitor changes in the moving average of photovoltaic power sequences, and flag extreme weather events when forecasted photovoltaic power falls below a critical threshold. That is, when When the trigger value is 1, the trigger value is 1.

[0024] Step 2: Calculation of the Dual-Mode Adaptive Horizon prediction mechanism. The optimal model prediction time domain for the next decision step is calculated. .

[0025] First, quantify the combined volatility index of traction load and photovoltaic fluctuations. :

[0026] Calculate traction load fluctuation rate and photovoltaic volatility ;

[0027] The composite volatility index takes the maximum of the two: .

[0028] Based on this volatility, the market enters the "contraction and benchmark determination phase." When excessively high volatility is detected... When this occurs, the "shrinkage mode" is triggered, defining an intermediate time domain. To filter out predicted noise.

[0029] Then determine if there are extreme events in the detection window. or This triggers the "Radar Extended Mode," adding event compensation time-domain increments. :

[0030]

[0031] in, To quantify the severity of the crisis, and These are the basic expansion coefficient and the cooperative fluctuation adjustment coefficient, respectively. The final prediction time domain length is then limited to:

[0032]

[0033] This method ensures that the energy storage system has a sufficiently long control field of view to perform deep pre-charging in critical situations.

[0034] Step 3: Establish a resilient model for prediction and a mixed-integer optimization problem (containing dynamic energy storage safety margin constraints).

[0035] In the adaptively determined future time period Internally, the objective function is to maximize economic efficiency and operational stability. The specific objective function is... This includes the cost of purchasing electricity from the main grid, the cost of energy storage degradation and loss, and the costs of corresponding peak shaving and valley filling mechanisms, whether as penalties or rewards.

[0036]

[0037] In addition to satisfying the basic power conservation constraints of the microgrid system, the constraints under the above objectives also include:

[0038]

[0039] The key is to establish a dynamic system state of charge (SOC) safety margin mechanism constraint based on a "storage-release" dual-state approach, breaking through the traditional rigid protection boundary:

[0040] Mode A: When no extreme events are triggered under normal conditions (hold mode), the lower limit of the deep discharge threshold of the battery system is increased based on the comprehensive volatility and prediction error coefficient.

[0041]

[0042] This model, as a preparatory phase, can proactively store usable energy for future unobserved uncertainties (i.e., reserve a "yellow defense line").

[0043] Mode B: When an extreme weather or high load alarm signal is triggered (release mode), the protection threshold is automatically released, allowing the protection to drop to extreme boundaries (e.g., It performs deep crisis power supply discharge to release all the energy of the system.

[0044] Step 4: Rolling solution and execution of energy management instructions.

[0045] The updated initial system state is fed into the adaptive predictive control model constructed above, and a commercial optimization algorithm (such as a mixed-integer programming solver) is used to solve the problem. The solution length is... The internal variable column for step size only includes the first time step control plan (i.e., the execution window) in the generation strategy. The corresponding operation is delegated to the bidirectional energy storage converter composed of supercapacitors and batteries. The clock step advances to the next time segment, receives the actual measurement feedback from the system again, performs a new round of comprehensive mutation and fluctuation characteristic calculation, triggers time-domain scaling reconstruction and rolling optimization process, and achieves flexible closed-loop control of extreme fluctuations of electrified railways.

Claims

1. A flexible adaptive time-domain energy management method for hybrid energy storage in electrified railways considering extreme conditions, characterized in that, Includes the following steps: Real-time acquisition of microgrid status data for electrified railways; execution of multi-source integrated event detection to determine whether a high-frequency fluctuation event or an extreme and severe impact event has been triggered. Based on the comprehensive volatility index calculated in real time according to the operating status and the event detection results, the prediction time domain length most suitable for the current moment is calculated through a dual-mode adaptive time domain adjustment mechanism. Within the prediction time domain, a hybrid energy storage scheduling optimization model is established based on the physical constraints of the internal equipment of the system and the dynamic state of charge safety margin mechanism with "energy storage-release" dual-state switching. Using the mixed integer linear programming method, the hybrid energy storage scheduling optimization model is solved periodically and continuously within the prediction time domain to obtain the control strategy sequence, and the optimal hybrid energy storage power allocation command for the current time step is issued to the hybrid energy storage system for execution.

2. The method for elastic adaptive time-domain energy management of hybrid energy storage in electrified railways considering extreme conditions, as described in claim 1, is characterized in that... The specific implementation process of performing multi-source integrated event detection includes: calculating the historical average traction load distribution for abnormal traction load surge events. with standard deviation And perform look-ahead scanning within the auxiliary time window, when the conditions are met. At that time, trigger the load event signal. Otherwise, the trigger value is 0; for photovoltaic power sag events caused by extreme weather, the moving average value of the photovoltaic power sequence is monitored in real time. When full At that time, the signal for triggering extreme weather events was obtained. Otherwise, the trigger value is 0, where As a benchmark photovoltaic power generation capacity, The threshold for determining the depth of power drop during extreme weather events.

3. The method for elastic adaptive time-domain energy management of hybrid energy storage in electrified railways considering extreme conditions, as described in claim 1, is characterized in that... The comprehensive volatility index $v(t)$, calculated in real time based on operating status, is calculated as follows: the traction load volatility is calculated separately. With photovoltaic power generation volatility The calculation formula is: as well as The composite volatility index can be obtained by taking the maximum of the two values. ,in This represents the first-order difference operation of a sequence. This indicates the standard deviation of time series data. A small constant protection term is used to prevent the denominator from becoming zero.

4. A method for elastic adaptive time-domain energy management of hybrid energy storage for electrified railways considering extreme conditions, as described in claims 1 and 3, is characterized in that... The implementation process of the dual-mode adaptive time-domain adjustment mechanism includes: first, determining whether a volatility-based "contraction mode" is triggered; and then, when the calculated composite volatility index... Greater than the preset high-frequency fluctuation threshold At that time, the benchmark optimization time domain The proportional decay is converted into a contracted intermediate time domain. First, it is used to filter prediction noise and limit error accumulation; second, it determines whether an event-based "extended mode" is triggered. If any extreme event trigger signal in the event detection is activated, then the intermediate time domain additional event joint compensation time domain increment is applied. Finally, the optimal prediction time domain length for the current moment is determined to be... , ,in This is the maximum permissible time domain boundary set by the system.

5. The method for elastic adaptive time-domain energy management of hybrid energy storage in electrified railways considering extreme conditions, as described in claim 4, is characterized in that... The event joint compensation time domain increment The calculation process integrates the severity of extreme events with compensation for high-frequency fluctuation characteristics; the specific mathematical formula is as follows: ,in In order to activate the pre-quantified severity index of the emergency, The empirical constant of the basic expansion coefficient, This is the comprehensive adjustment coefficient for coordinated fluctuations.

6. The method for elastic adaptive time-domain energy management of hybrid energy storage in electrified railways considering extreme conditions, as described in claim 1, is characterized in that... The objective function of the hybrid energy storage scheduling optimization model To minimize the overall operating cost within the forecast time domain, specifically including the main grid electricity purchase cost. Degradation loss cost of hybrid energy storage systems And operating penalties and reward costs The formula is expressed as The model also sets requirements to satisfy the system's basic active power conservation constraints. ,in For photovoltaic power, For the power purchase capacity, and These are the total energy storage discharge and total charging power, respectively. To shed load power, This refers to the power of abandoned light.

7. The method for elastic adaptive time-domain energy management of hybrid energy storage in electrified railways considering extreme conditions, as described in claim 1, is characterized in that... The dynamic state-of-charge safety margin mechanism with "energy storage-release" dual-state switching operates as follows: when no extreme event trigger signal is detected, the system enters the first operating state, i.e., the hold mode, at which point the overall volatility is considered. Prediction error compared to model estimates Dynamically raise the lower limit of deep discharge protection for energy storage batteries to build a flexible defense; set the dynamic lower limit threshold as follows: When a trigger signal is detected indicating either a sudden abnormal load or worsening extreme weather, the system switches to the second operating state, i.e., the release mode. At this time, the dynamic lower limit threshold will be automatically and forcibly released, and the system will switch back to the original physical boundary lower limit of the equipment. This allows energy storage systems to extend their discharge capacity to overcome transient energy crises.

Images