Energy scheduling optimization method for energy storage tunnel

Abstract

The invention provides an energy scheduling optimization method for an energy storage tunnel, which comprises the following steps: collecting real-time operation data of all energy storage units in the energy storage tunnel, evaluating the health state of each energy storage unit based on a multi-dimensional health monitoring model, and identifying potential faults through a time sequence analysis and anomaly detection mechanism; constructing a multi-level distributed scheduling architecture; based on the historical load data, the external environment factor and the state of the energy storage unit, predicting a load demand through a weighted moving average model, and dynamically adjusting a prediction result of the load demand in combination with an adjustment factor; according to the prediction error of the load demand and the health state threshold value of the energy storage unit, designing an anomaly detection mechanism to judge whether to trigger fault detection, and after the fault detection is triggered, dynamically adjusting the electric quantity of the energy storage unit or switching a standby unit to realize automatic repair of the system; and integrating the health state, the electric quantity state and the load prediction result of the energy storage unit, and dynamically optimizing a global charging and discharging strategy by taking minimization of energy loss and health deterioration as a target.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy scheduling optimization, and particularly relates to an energy scheduling optimization method for energy storage tunnels. Background Art

[0002] With the large-scale application of renewable energy and the increasing volatility of the power grid load, the demand for energy storage systems is growing. Especially in the regulation of fluctuating energy sources such as wind energy and solar energy, energy storage technology plays a crucial role. As a key device for electric energy storage and scheduling, energy storage tunnels have been widely used in fields such as power grid load regulation, smooth access of renewable energy, and emergency backup power supplies. However, the energy storage tunnel system faces a series of challenges that need to be addressed urgently in practical applications.

[0003] First of all, traditional energy storage tunnel systems mostly rely on centralized scheduling methods, that is, all energy storage units are uniformly scheduled through a centralized control system. Although this method can ensure the coordination of the system to a certain extent, there are also several significant problems: on the one hand, the centralized scheduling has poor adaptability to dynamic load changes, which is prone to cause scheduling delays in the system. Especially when the load suddenly changes or the performance of the energy storage unit deteriorates, the system scheduling reaction lags behind and cannot respond to the changes in the system state in a timely manner; on the other hand, with the increase in the number of energy storage units, the system complexity of centralized control rises sharply, resulting in an increase in computing resources and communication burdens, further affecting the response speed and energy efficiency of the system.

[0004] Secondly, existing energy storage scheduling systems often lack effective fault detection and emergency response mechanisms. Once a fault occurs in an energy storage unit, traditional systems usually rely on manual intervention or simple preset rules for emergency handling, lacking flexible and intelligent automatic repair capabilities. This makes it possible that when a fault occurs in an energy storage unit, the system may not be able to recover in time, resulting in a decrease in the overall efficiency of the system, and even causing power supply interruptions or instability.

[0005] Furthermore, the scheduling optimization problem of energy storage tunnels also involves how to optimize the energy distribution under various complex constraints. Traditional scheduling algorithms, especially those based on heuristic methods or simple linear models, are prone to fall into local optimal solutions when facing a large number of energy storage units, resulting in energy efficiency losses. At the same time, due to the dynamic changes in the power grid load and the state of energy storage units, traditional methods lack sufficient adaptability and cannot achieve real-time scheduling optimization in the true sense.

[0006] Based on these problems, existing energy storage tunnel systems have defects such as low energy scheduling efficiency, untimely fault recovery, and poor system adaptability. In order to address these challenges and improve the energy efficiency, stability, and reliability of energy storage tunnels, an advanced method that can adapt to dynamic load changes, has high scheduling capabilities, and supports self-repair of faults is urgently needed. Summary of the Invention

[0007] The object of the present invention is to design an energy scheduling optimization method for energy storage tunnels, which overcomes the problems existing in the existing energy storage tunnel scheduling methods, such as load response lag, difficult fault repair, low optimization efficiency, etc., and also significantly improves the stability, reliability and energy efficiency of the system.

[0008] To achieve the above object, the present invention provides an energy scheduling optimization method for energy storage tunnels, and the method includes the following steps:

[0009] S1. Collect the real-time operation data of all energy storage units in the energy storage tunnel, evaluate the health status of each energy storage unit based on a multi-dimensional health monitoring model, and identify potential faults through time series analysis and anomaly detection mechanism; the real-time operation data includes voltage, current, temperature, power and charge-discharge times;

[0010] S2. Construct a multi-level distributed scheduling architecture, including a three-level scheduling architecture of a global scheduling layer, a regional scheduling layer and a local scheduling layer, where:

[0011] The global scheduling layer conducts global energy planning according to the system load demand;

[0012] The regional scheduling layer coordinates the load balance of energy storage units within the region according to the global scheduling instruction;

[0013] The local scheduling layer dynamically adjusts the charge-discharge strategy of a single energy storage unit in combination with the health status and real-time power status of the energy storage unit;

[0014] S3. Based on historical load data, external environmental factors and the status of energy storage units, predict the load demand through a weighted moving average model, and dynamically adjust the prediction result of the load demand in combination with a regulation factor;

[0015] S4. Design an anomaly detection mechanism according to the prediction error of the load demand and the health status threshold of the energy storage unit to judge whether to trigger fault detection, and after triggering the fault detection, realize automatic system repair by dynamically adjusting the power of the energy storage unit or switching to a standby unit;

[0016] S5. Synthesize the health status, power status of the energy storage unit and the load prediction result, and dynamically optimize the global charge-discharge strategy with the goal of minimizing energy loss and health degradation.

[0017] Further, the multi-dimensional health monitoring model is used to monitor the health status of the energy storage unit; the health status of the energy storage unit is obtained by performing regression analysis on the real-time operation data based on an empirical model of historical data and the system knowledge base;

[0018] In step S1, for the operation data of each energy storage unit, the wavelet transform method is used to extract the key features in the time series data; based on the key features of the operation data of each energy storage unit, a weighted fusion method is used to synthesize multi-dimensional health assessment data, and an adaptive threshold method is used for anomaly detection.

[0019] Furthermore, the charge and discharge strategy of the local scheduling layer is dynamically adjusted through the health status weighted formula:

[0020]

[0021] Where is the charge and discharge state of energy storage unit i at time t; ΔSOC i (t) is the charge and discharge amount of energy storage unit i at time t; HS i is the health state of energy storage unit i; γ i is the regularization coefficient, which is used to adjust the influence intensity of the health state on the charge and discharge strategy.

[0022] Furthermore, the purpose of the regional scheduling layer is to coordinate the scheduling of multiple energy storage units within the region to achieve load balance and pre-adjust the local load fluctuations:

[0023]

[0024] Where w i is the weight of energy storage unit i, which is related to the health state HS i ; SOC target is the regional scheduling target charge state; is the energy change amount of energy storage unit i at time t; λ1 and λ2 are regularization coefficients, which respectively control the influence weights of energy loss and health state.

[0025] Furthermore, the global scheduling layer is used to minimize the overall energy loss of the system, the deterioration of the health state of the energy storage unit, and achieve the optimization of energy scheduling:

[0026]

[0027] Where α k is the weight of the kth region; SOC target is the global target charge state; λ1 and λ2 are regularization coefficients, which are used to balance the influence of energy loss and health loss; is the state of energy storage unit i at time t.

[0028] Furthermore, the anomaly detection mechanism is realized through dynamic threshold judgment:

[0029] Obtain the prediction error of the load demand. If the prediction error exceeds the threshold based on the external environmental factor, anomaly detection will be triggered; the external environmental factor is determined based on the fluctuation of the load demand due to temperature.

[0030] Furthermore, the optimization objective function of the global charge-discharge strategy is:

[0031]

[0032] where η i is the efficiency coefficient of the energy storage unit, β i is the error adjustment coefficient, and ΔD adj (t + k) is the load prediction error adjustment value.

[0033] Furthermore, the automatic repair includes at least one of the following methods:

[0034] (a) Repair the power through the charging adjustment amount to repair the power;

[0035] (b) Repair the health state according to the health state recovery amount ΔHS1(t + k) = ζ·[HS target - HS1(t + k)];

[0036] (c) Switch to the standby energy storage unit, and its scheduling amount is calculated by backup (+) = backup ·[^(+)-(+)]

[0037] The beneficial technical effects of the present invention are at least as follows:

[0038] By introducing a multi-level distributed scheduling architecture, the present invention breaks through the limitations of traditional centralized scheduling methods. In this architecture, the system is divided into three levels: top-level global scheduling, middle-level regional coordination, and bottom-level energy storage unit control. Each level is independently responsible for different scheduling tasks, avoiding the computational bottleneck and communication burden of centralized scheduling. This hierarchical scheduling architecture enables the system to more flexibly respond to load fluctuations and changes in the state of energy storage units. At the top level, the global scheduling system can perform global energy planning and optimize the energy flow between the power grid load and the energy storage unit; the middle-level regional coordination layer adjusts the energy distribution of local energy storage units according to the top-level scheduling instructions to achieve load balancing within the region; the bottom-level control system directly regulates the charge-discharge process of each energy storage unit to ensure that each energy storage unit operates according to the optimal strategy. This solution solves the defect that traditional centralized scheduling cannot quickly respond to dynamic load changes and significantly improves the response speed and energy efficiency of the system.

[0039] Another innovative point of the present invention is the introduction of a self - repair mechanism, enabling the energy storage tunnel system to automatically detect, isolate, and recover from faults without relying on manual intervention when a fault occurs. By continuously monitoring the health status of each energy storage unit in real - time, when a fault or performance degradation is detected, the system can quickly adjust the scheduling strategy, transfer the load of the faulty unit to other healthy units, and ensure the stable operation of the system during the occurrence of a fault. The self - repair mechanism can not only reduce the energy efficiency loss caused by the failure of a single energy storage unit but also quickly restore the system to the optimal working state after system repair, greatly improving the reliability and fault - tolerance ability of the system.

[0040] To further improve the efficiency of energy scheduling, the present invention also introduces a real - time adaptive scheduling optimization algorithm. This algorithm can dynamically adjust the scheduling strategy according to the real - time status of energy storage units and the power grid load demand. By using advanced prediction models and adaptive algorithms, the system can predict load fluctuations in real - time and adjust the charge - discharge behavior of energy storage units, avoiding the problem of local optimal solutions in traditional methods. This optimization method enables the system to always maintain the best energy scheduling state in the face of dynamic changes, minimizing energy loss and improving the overall operation efficiency to the greatest extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present invention will be further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the following drawings.

[0042] Figure 1 It is a flowchart of the energy scheduling optimization method for the energy storage tunnel of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation to the present invention.

[0044] In one or more embodiments, as Figure 1 shown, an energy scheduling optimization method for an energy storage tunnel is disclosed. The method includes the following steps:

[0045] S1. Collect the real - time operation data of all energy storage units in the energy storage tunnel, evaluate the health status of each energy storage unit based on a multi - dimensional health monitoring model, and identify potential faults through time - series analysis and anomaly detection mechanisms; the real - time operation data includes voltage, current, temperature, power, and charge - discharge times.

[0046] Specifically, in this step, our goal is to collect data from all energy storage units in the energy storage tunnel system in real time, obtain multi-dimensional information such as their operating status, load, and health status, and process and analyze this information to provide data support for subsequent scheduling optimization and fault detection. To achieve this goal, we propose a "multi-dimensional health monitoring and data processing method", which is not just simple data collection, but combines time series analysis and anomaly detection mechanisms, and can accurately reflect the health status and operating conditions of energy storage units.

[0047] Furthermore, first, collect relevant data from all energy storage units (such as batteries, supercapacitors, etc.) in the energy storage tunnel system, mainly including but not limited to information such as battery voltage, current, temperature, power, charge and discharge times, etc. Make a preliminary estimate of the health status (HS) of each energy storage unit:

[0048] HS i = f(V i , I i , T i , P i , C i ) (1)

[0049] where HS i represents the health status of the i-th energy storage unit; V i represents the voltage of the i-th energy storage unit; I i represents the current of the i-th energy storage unit; T i represents the temperature of the i-th energy storage unit; P i represents the power of the i-th energy storage unit; C i represents the charge and discharge times of the i-th energy storage unit.

[0050] The function f(·) in this step is an empirical model based on historical data and the system knowledge base, and can be obtained by performing regression analysis on the performance of energy storage units under different working conditions. The evaluation of the health status is multi-dimensional, and usually uses weighted average or other methods to fuse each index, so as to comprehensively evaluate the health status of the battery unit.

[0051] Furthermore, for the operating data of each energy storage unit, we introduce time series analysis technology, especially the wavelet transform method to extract key features in time series data, such as change trends, sudden events, abnormal fluctuations, etc. In practical applications, the working state of the energy storage system will be affected by many environmental factors (such as temperature changes, power grid fluctuations, etc.), so simple threshold judgment may not be sufficient to detect potential abnormal states. Therefore, we construct a set of wavelet coefficients (W) to extract multi-scale features in the time series:

[0052]

[0053] Among them, W i,j is the j-th wavelet coefficient of the i-th energy storage unit; x(t) is the operating data (such as voltage, current, etc.) of the energy storage unit at time t; ψ i,j (t) is the wavelet basis function, representing the characteristics of the i-th energy storage unit at the j-th time scale. By the change of the wavelet coefficient, we can judge whether there is abnormal fluctuation in the energy storage unit, and then conduct fault warning.

[0054] Furthermore, after obtaining the time series characteristics and health status evaluations of all energy storage units, we use the weighted fusion method to synthesize the multi-dimensional health assessment data, and use the adaptive threshold method to conduct anomaly detection. The health status of each energy storage unit is re-evaluated. If the health status is lower than a certain threshold, it is marked as abnormal, further triggering a fault warning. The specific formula is as follows:

[0055]

[0056] Among them, is the health status of the i-th energy storage unit after fusion; HS i,j is the health status of the i-th energy storage unit in the j-th dimension (such as voltage, current, temperature, etc.); w j is the weight of the j-th dimension, and the weight is obtained through the correlation analysis of historical data. Then, according to the health status evaluation result, the dynamic threshold method is used to detect the abnormal state:

[0057]

[0058] Among them, Threshold i is the adaptive threshold obtained according to the historical data analysis;

[0059] Furthermore, if the health status is lower than this threshold, a warning is triggered, indicating that the i-th energy storage unit has a fault risk.

[0060] Finally, after the above data collection and health monitoring process, the health status and abnormal information of all energy storage units are stored and output for use by subsequent distributed scheduling, load prediction and other modules. The output data structure is usually a matrix or table containing the health status of all energy storage units and related abnormal data, as the input data for subsequent scheduling decisions.

[0061] S2. Construct a multi-level distributed scheduling architecture, including a three-level scheduling architecture of a global scheduling layer, a regional scheduling layer and a local scheduling layer, where:

[0062] The global scheduling layer conducts global energy planning according to the system load demand;

[0063] The regional scheduling layer coordinates the load balancing of energy storage units within the region according to the global scheduling instructions;

[0064] The local scheduling layer dynamically adjusts the charge-discharge strategy of a single energy storage unit in combination with the health status and real-time power status of the energy storage unit.

[0065] Specifically, our goal is to design a more efficient, flexible, and intelligent distributed architecture for the energy scheduling problem of energy storage tunnels. This architecture not only considers the health status, charge-discharge requirements, and system load of each energy storage unit, but also incorporates a dynamic feedback mechanism and multi-objective optimization techniques to better achieve efficient energy scheduling and optimization.

[0066] Furthermore, to achieve efficient energy scheduling, we designed a three-level distributed architecture:

[0067] Local Scheduling Layer: Handles the charge-discharge decisions of each energy storage unit, combining its health status and current energy demand;

[0068] Regional Scheduling Layer: Coordinates the energy scheduling of energy storage units within the same region and makes appropriate adjustments according to regional demands;

[0069] Global Scheduling Layer: Conducts energy management and scheduling at the system level, comprehensively considering the demands, status, energy flow of the entire system, and feedback from each region.

[0070] Each level has specific tasks, but they coordinate with each other through information transfer and feedback mechanisms to ensure the optimal energy scheduling of the overall system.

[0071] Furthermore, the local scheduling layer:

[0072] The main task of the local scheduling layer is to determine the charge-discharge strategy of each energy storage unit and make dynamic adjustments according to the health status (HS i ), current energy state (SOC i ), and predicted demand of the energy storage unit. To maximize the utilization efficiency of the energy storage unit, we designed a new scheduling strategy considering the following factors:

[0073] The health status (HS i ) of the energy storage unit; the current charge-discharge state (SOC i ) of the energy storage unit; the charge-discharge efficiency (C e ) of the battery; the energy demand prediction (D i (t)).

[0074] Here, we propose a dynamic scheduling optimization formula with health state weighting, which further weights the impact of the health state while considering the current load of each energy storage unit, avoiding overuse of energy storage units with poor health states:

[0075]

[0076] Among them, is the charge and discharge state of energy storage unit i at time t; ΔSOC i (t) is the charge and discharge amount of energy storage unit i at time t; HS i is the health state of energy storage unit i; γ i is the regularization coefficient, which is used to adjust the influence intensity of the health state on the charge and discharge strategy.

[0077] Regional scheduling layer:

[0078] The purpose of the regional scheduling layer is to coordinate the scheduling of multiple energy storage units within the region to achieve load balancing and pre-regulate local load fluctuations. In the design, we added an energy transfer model between energy storage units to ensure that the energy scheduling between energy storage units can flexibly respond to demand fluctuations.

[0079] The core goal of regional scheduling is to optimize the scheduling strategy by minimizing factors such as overall energy loss, load imbalance, and deterioration of the health state. We introduced an objective function with regular terms for energy loss and health loss, comprehensively considering factors such as energy demand, health state of energy storage units, and load balance:

[0080]

[0081] Among them, w i is the weight of energy storage unit i, which is related to its health state (HS i ); SOC target is the target charging state of regional scheduling; is the energy change amount of energy storage unit i at time t; λ1 and λ2 are regularization coefficients, which respectively control the influence weights of energy loss and health state. The design of this objective function enables the regional scheduling layer to minimize the health deterioration of energy storage units while ensuring load balance, thereby extending the service life of the entire system.

[0082] Global scheduling layer:

[0083] The global scheduling layer is the core of the entire system's energy scheduling, responsible for coordinating the energy flow and scheduling strategies among regional scheduling layers. The global scheduling layer not only considers the current demand but also conducts predictions of future loads and optimizes the energy flow. Therefore, we introduce a distributed multi-objective optimization method that combines the demand predictions of multiple regions and the health status information of energy storage units.

[0084] The global scheduling objective function integrates factors such as regional energy balance, the health status of energy storage units, and the overall demand prediction of the system. Our goal is to minimize the overall energy loss of the system, the deterioration of the health status of energy storage units, and achieve the optimization of energy scheduling:

[0085]

[0086] where α k is the weight of the k-th region; SOC target is the global target charge state; λ1 and λ2 are regularization coefficients used to balance the impacts of energy loss and health degradation; is the state of energy storage unit i at time t. By introducing a global scheduling optimization method with multi-region collaboration, the system can adjust the global energy flow based on the feedback and health status information of each region, ensuring that the system can still operate stably under large load fluctuations and applying more protection measures to energy storage units with poor health status.

[0087] Furthermore, each scheduling layer (local, regional, and global) includes a real-time feedback mechanism, and these feedback messages are transmitted among different levels to ensure the real-time and flexibility of the scheduling strategy. The core objective of the feedback mechanism is to dynamically adjust the objective function according to the actual execution results (such as energy loss, health degradation, etc.) for optimization in the next scheduling cycle.

[0088] The design of this multi-level distributed scheduling architecture in this step solves the efficiency and coordination problems of existing scheduling methods when facing complex energy storage systems. By introducing a scheduling strategy with health status weighting, regularization terms for energy loss and health degradation, and a dynamic feedback mechanism, the system can flexibly respond to the health status fluctuations of different energy storage units and maximize the overall energy scheduling benefit. This method is particularly suitable for energy storage tunnel systems, which can consider factors such as the health status of energy storage units, the current energy demand, load fluctuations, etc. simultaneously, and avoid a significant decline in system performance due to the failure of a single energy storage unit through refined scheduling. In addition, the multi-level architecture design ensures the scalability and stability of the system, enabling more efficient and reliable management of a large number of energy storage units in practical applications.

[0089] S3. Based on historical load data, external environmental factors, and the state of the energy storage unit, predict the load demand through a weighted moving average model, and dynamically adjust the prediction result of the load demand in combination with an adjustment factor.

[0090] Specifically, in step S3, we designed an innovative adaptive load prediction and optimization algorithm, combining the multi-level distributed scheduling architecture design in the previous step (step S2) and the unique requirements of the energy storage system. The core objective of this step is to achieve high-precision prediction of the load demand and ensure that the prediction result can dynamically adapt to system state and environmental changes, thereby optimizing the scheduling strategy. The specific details are as follows:

[0091] The input data for this step mainly comes from the output of the previous stage, including:

[0092] Historical load data: D i (t), representing the load demand of energy storage unit i at time t.

[0093] Battery state: SOC i (t), representing the battery charge state of energy storage unit i at time t.

[0094] Health state: HS i (t), representing the health state of energy storage unit i, which may affect the discharge capacity.

[0095] External environmental factors: E i (t), such as external factors like climate conditions and regional load, which affect the load demand.

[0096] Furthermore, the key challenge in load prediction is to capture the dynamic changes in load demand, especially the impacts brought about by changes in the environment and battery state. To achieve this goal, we designed an adaptive load prediction model based on weighted moving average (WMA) and an adjustment factor.

[0097] To adapt to the temporal changes in load demand, we designed a weighted moving average model, where the weights are dynamically adjusted according to historical load fluctuations. The specific formula is:

[0098]

[0099] Where, is the predicted load demand value, representing the load demand of energy storage unit i at time t + k. W(t) is the dynamic weight, which is adjusted according to the fluctuations of historical loads, assigning higher weights to periods with large load fluctuations and lower weights to periods with small fluctuations. F(t) is the adjustment factor, used to combine the impacts of the external environment and the state of the energy storage unit. The adjustment factor F(t) is used to integrate the external environment, the state of charge, and the health state of the energy storage unit to help adjust the output of the prediction model. Its design is as follows:

[0100] F(t) = β·E i (t) + γ·SOC i (t) + δ·HS i (t) (9)

[0101] where, E i (t): external environment factor (such as weather, grid load); SOC i (t): state of charge of the energy storage unit; HS i (t): health state of the energy storage unit; β, γ, δ: weight coefficients of the adjustment factor, set through training or prior knowledge. This design ensures that the load prediction can be closely related to the external environment and the health state of the energy storage system, thereby improving the prediction accuracy.

[0102] Furthermore, based on the load prediction results we further conduct scheduling optimization. To ensure that the energy storage unit can be reasonably scheduled according to the actual load demand, we design a scheduling adjustment strategy based on load differences. The specific adjustment amount ΔSOC i (t + k) can be calculated by the following formula:

[0103]

[0104] where, ΔSOC i (t + k): represents the state of charge adjustment amount of energy storage unit i at time t + k. η: adjustment coefficient, controlling the adjustment amplitude between the prediction error and the state of charge.

[0105] Through this adjustment strategy, we ensure that the state of charge of the energy storage system can be dynamically adjusted according to the load demand and the actual state of charge, thereby avoiding over-discharge or over-charging of the battery and improving the overall energy utilization efficiency of the system. The core of this step lies in designing an adaptive load prediction model by combining historical load data, external environment factors, and the state of the energy storage unit, and optimizing the load prediction results through the adjustment factor and the load difference adjustment mechanism. This method not only improves the accuracy of load prediction but also ensures that the system can achieve efficient scheduling in a dynamically changing load environment, ultimately providing strong support for the optimal operation of the energy storage system.

[0106] S4. Design an anomaly detection mechanism based on the prediction error of the load demand and the health state threshold of the energy storage unit to determine whether to trigger a fault detection. After the fault detection is triggered, the system automatically repairs itself by dynamically adjusting the power of the energy storage unit or switching to a standby unit.

[0107] Specifically, the core goal of this step is to detect and repair faults in the energy storage system in real time to ensure the efficient and stable operation of the system. Specifically, we adopt a fault detection algorithm based on load prediction error, battery state, health state, and external environmental factors, and combine dynamic threshold adjustment and repair strategies for self-repair. The following is a more specific implementation plan, analyzed in combination with actual examples.

[0108] The input data comes from the output of step 3:

[0109] Predicted load demand: (Predicted load demand of energy storage unit i at time t + k);

[0110] Actual load demand: D i (t + k) (Actual load demand of energy storage unit i at time t + k);

[0111] Battery state of charge: SOC i (t + k) (Battery state of charge of energy storage unit i at time t + k);

[0112] Battery health state: HS i (t + k) (Health state of energy storage unit i);

[0113] External environmental factor: E i (t + k) (External factors affecting the load demand, such as climate, grid load, etc.).

[0114] Furthermore, the anomaly detection mechanism:

[0115] First, calculate the load prediction error:

[0116]

[0117] For example, assume that the system predicts the load demand of energy storage unit 1 at time t + k to be But the actual demand is D1(t + k) = 400W, and the load prediction error is:

[0118] ΔD1(t + k) = |500 - 400| = 100W (12)

[0119] If this error is too large and exceeds the threshold, the system will trigger an anomaly detection. We define the dynamically adjusted error as:

[0120] ΔD adj(t + k)=ΔD i (t + k)-λ·E i (t + k)(13)

[0121] Assume the external environmental factor E i (t + k) is the fluctuation of the load demand caused by the increase in temperature. E1(t + k)=20, the environmental factor adjustment coefficient λ = 0.5. Then the adjusted error is:

[0122] ΔD adj (t + k)=100 - 0.5·20 = 90W (14)

[0123] Next, we judge whether it is abnormal according to the dynamic threshold of the load demand error. Assume the dynamic threshold we set is:

[0124]

[0125] After substituting the values:

[0126] ΔD threshold (t + k)=50 + 0.1·500 + 0.2·20 = 50 + 50 + 4 = 104W (16)

[0127] Since ΔD adj (t + k)=90W is less than the threshold of 104W, this error is acceptable and the system will not trigger the repair mechanism. However, if the error is greater than this threshold, the system will start the fault detection mechanism.

[0128] Furthermore, when the system detects a fault, it will make automatic adjustments according to the repair strategy. The self - repair mechanism mainly includes two aspects: battery power repair and health status repair.

[0129] Battery power repair:

[0130] Assume the system detects that the battery power status of energy storage unit 1 is low (for example, SOC1(t + k)=30%), and predicts a significant deviation in the load demand. The system will adjust the battery power according to the following repair strategy. We calculate the battery power repair amount ΔSOC i (t + k):

[0131]

[0132] Set: the battery power repair coefficient η = 0.1, the repair indication function (indicating that repair is required).

[0133] Then, the battery power repair amount is:

[0134] ΔSOC1(t + k) = 0.1·(90 - 30) = 6% (18)

[0135] That is, the system will charge to increase the battery power by 6% to meet the load demand.

[0136] Battery health state repair:

[0137] If the battery health state is poor, HS1(t + k) = 0.4 (assuming the health state value is between [0, 1], 0 means the worst, 1 means the best), which is lower than the set threshold θ HS = 0.5, then the system will trigger the health state repair mechanism. The health state is repaired through the following repair function:

[0138] ΔHS1(t + k) = ζ·[HS target -HS1(t + k)] (19)

[0139] Assume:

[0140] The health state recovery coefficient ζ = 0.2,

[0141] The ideal health state value HS target = 1.

[0142] Then, the repair amount of the battery health state is:

[0143] ΔHS1(t + k) = 0.2·(1 - 0.4) = 0.12 (20)

[0144] That is, the system will restore the battery health state by 12%.

[0145] Furthermore, if the battery cannot be repaired, the system needs to switch to the backup energy storage unit to maintain the normal operation of the system. Assume the power state of the backup energy storage unit is SOC backup = 50%, and the goal of the system is to schedule it to meet the load demand:

[0146]

[0147] Assume:

[0148] The scheduling coefficient of the backup energy storage unit θ backup = 0.5,

[0149] The current load demand

[0150] The current battery state SOC1(t + k) = 30%.

[0151] Then, the scheduling amount of the backup energy storage unit is:

[0152] ΔSOC backup(t + k) = 0.5·(500 - 30) = 235W (22)

[0153] The backup energy storage unit provides 235W of energy to supplement the shortage of the main energy storage unit.

[0154] S5. Integrate the health state, power state of the integrated energy storage unit, and the load prediction result, and dynamically optimize the global charge-discharge strategy with the goal of minimizing energy loss and health degradation.

[0155] Specifically, in step 5, our goal is to achieve the global energy scheduling of the entire system and optimize the energy flow efficiency between the energy storage unit and the load. To achieve this goal, we need to comprehensively consider the power state, health state of the energy storage unit, load demand prediction, and external environmental factors, and then determine the appropriate charge-discharge strategy. In this step, we design a global optimization framework to enable the system to dynamically adjust according to the actual situation, thereby maximizing the system efficiency.

[0156] Furthermore, to enable the system to operate efficiently, we design a global energy scheduling goal aimed at minimizing energy waste and ensuring the effective satisfaction of load demands. We define a global energy efficiency objective function ε total , which includes the power state of the energy storage unit and the load error adjustment, specifically:

[0157]

[0158] where η i is the efficiency coefficient of energy storage unit i, representing the energy conversion efficiency of this energy storage unit; SOC i (t + k) is the battery power of energy storage unit i at time t + k; is the predicted load demand of energy storage unit i at time t + k. In this objective function, we optimize the difference between the power state of the energy storage unit and the predicted load demand as a key factor. The goal is to improve the energy efficiency of the system by minimizing this difference, and at the same time consider the different performance characteristics of each energy storage unit through the energy efficiency coefficient η i .

[0159] Furthermore, to further optimize the system performance, we combine a dynamic scheduling strategy so that the system can dynamically adjust the charge-discharge process of the energy storage unit while meeting the load demand. At this time, the amount of power adjustment of the energy storage unit can be calculated by the following formula:

[0160]

[0161] where ΔSOC i (t + k) is the amount of power adjustment of energy storage unit i at time t + k; is the load prediction demand of energy storage unit i; SOC i (t + k) is the current battery charge of energy storage unit i; α i is the charging sensitivity coefficient of energy storage unit i.

[0162] Through this scheduling strategy, we can dynamically adjust the charging and discharging behavior of the energy storage unit according to the difference between the state of charge of the energy storage unit and the load demand. For example, if the predicted load demand is greater than the current charge, we will charge the energy storage unit; if the load demand is less than the charge, we may discharge the energy storage unit.

[0163] Furthermore, to further improve the intelligence and energy efficiency of the scheduling, we introduce a regularization term to consider the impact of load prediction errors. This regularization term controls the load prediction error to avoid overcharging and discharging behaviors of the system due to prediction deviations. Specifically, the load error regularization term is:

[0164]

[0165] where is the regularization objective of the load prediction error; β i is the error adjustment coefficient of energy storage unit i, controlling the impact of errors on scheduling; ΔD adj (t + k) is the load prediction error adjustment value, representing the difference between the predicted load and the actual load.

[0166] By introducing this regularization term, the system will preferentially avoid energy waste caused by load prediction errors during scheduling. Especially when the load prediction error is large, the system will tend to choose a more conservative scheduling strategy. This can effectively improve the stability and efficiency of the system and avoid overcharging and discharging due to inaccurate predictions.

[0167] Furthermore, in this step, our final global scheduling objective is the weighted sum of the energy efficiency optimization objective and the load error regularization term:

[0168]

[0169] By optimizing this comprehensive objective, we ensure that the energy storage system provides sufficient energy support while minimizing energy waste. The design of the comprehensive objective can ensure the optimal energy efficiency of the system and has strong robustness to cope with changes in different load demands and environmental conditions.

[0170] In summary, the present invention not only overcomes the problems existing in the existing energy storage tunnel scheduling method, such as load response lag, difficult fault repair, and low optimization efficiency, but also significantly improves the stability, reliability, and energy efficiency of the system. Compared with the traditional method, the energy storage tunnel energy scheduling system of the present invention can adapt to load fluctuations and changes in the state of energy storage units in real time, ensuring that the system can operate efficiently and stably under various complex environments, and has broad application prospects.

[0171] The above-disclosed are only some preferred embodiments of the present invention. Of course, the scope of the rights of the present invention cannot be limited thereby. Those of ordinary skill in the art can understand all or part of the processes of implementing the above embodiments, and the equivalent changes made according to the claims of the present invention still fall within the scope covered by the invention.

Claims

1. An energy scheduling optimization method for an energy storage tunnel, characterized in that: The method comprises the following steps: S1. Collect real-time operating data of all energy storage units in the energy storage tunnel, evaluate the health status of each energy storage unit based on a multi-dimensional health monitoring model, and identify potential faults through time series analysis and anomaly detection mechanism; the real-time operating data includes voltage, current, temperature, power and charge and discharge times; S2. Construct a multi-level distributed scheduling architecture, including a three-level scheduling architecture of global scheduling layer, regional scheduling layer and local scheduling layer, where: The global scheduling layer performs global energy planning according to system load requirements; The regional scheduling layer coordinates the load balancing of energy storage units in the region according to the global scheduling instructions; The local scheduling layer dynamically adjusts the charging and discharging strategy of a single energy storage unit based on the health status and real-time power status of the energy storage unit; S3, based on historical load data, external environmental factors and energy storage unit status, predict load demand through weighted moving average model, and dynamically adjust the load demand prediction result in combination with adjustment factors; S4. Design an abnormality detection mechanism based on the load demand prediction error and the energy storage unit health status threshold to determine whether to trigger fault detection. After the fault detection is triggered, the system is automatically repaired by dynamically adjusting the energy storage unit power or switching the backup unit. S5. Comprehensively consider the health status, charge status and load forecast results of the energy storage unit, dynamically optimize the global charging and discharging strategy with the goal of minimizing energy loss and health degradation.

2. The energy scheduling optimization method for energy storage tunnel according to claim 1 is characterized in that: The multi-dimensional health monitoring model is used to monitor the health status of the energy storage unit; the health status of the energy storage unit is obtained by performing regression analysis on real-time operation data based on historical data and an empirical model of a system knowledge base.

3. The energy scheduling optimization method for energy storage tunnel according to claim 1 is characterized in that: In step S1, for the operating data of each energy storage unit, a wavelet transform method is used to extract key features in the time series data; a weighted fusion method is used based on the key features of the operating data of each energy storage unit to integrate multi-dimensional health assessment data, and anomaly detection is performed through an adaptive threshold method.

4. The energy scheduling optimization method for energy storage tunnel according to claim 1, characterized in that: The charging and discharging strategy of the local scheduling layer is dynamically adjusted through the health status weighted formula: in, is the charge and discharge state of energy storage unit i at time t; ΔSOC i (t) is the charge and discharge amount of energy storage unit i at time t; HS i is the health status of energy storage unit i; γ i is the regularization coefficient, which is used to adjust the impact of the health status on the charging and discharging strategy.

5. The energy scheduling optimization method for energy storage tunnel according to claim 1, characterized in that: The purpose of the regional dispatching layer is to coordinate the dispatching of multiple energy storage units in the region to achieve load balance and pre-regulate local load fluctuations: Among them, w i is the weight of energy storage unit i, and is related to the health status HS i Related; SOC target is the target charging state of the regional dispatch; is the energy change of energy storage unit i at time t; λ1 and λ2 are regularization coefficients, which control the influence weights of energy loss and health status respectively.

6. The energy scheduling optimization method for energy storage tunnel according to claim 4 or 5, characterized in that: The global scheduling layer is used to minimize the overall energy loss of the system, the degradation of the health status of the energy storage unit, and optimize the energy scheduling: Among them, α k is the weight of the kth region; SOC target is the global target charging state; λ1 and λ2 are regularization coefficients used to balance the effects of energy loss and health loss; is the state of energy storage unit i at time t.

7. The energy scheduling optimization method for energy storage tunnel according to claim 6 is characterized in that: The anomaly detection mechanism is implemented through dynamic threshold judgment: The prediction error of the load demand is obtained. If the prediction error exceeds a threshold based on an external environmental factor, an anomaly detection is triggered; the external environmental factor is determined based on the fluctuation of the temperature on the load demand.

8. The energy scheduling optimization method for energy storage tunnel according to claim 6 is characterized in that: The optimization objective function of the global charge and discharge strategy is: Among them, η i is the efficiency coefficient of the energy storage unit, β i is the error adjustment coefficient, ΔD adj (t+k) is the load prediction error adjustment value.

9. The energy scheduling optimization method for energy storage tunnel according to claim 1, characterized in that: The automatic repair includes at least one of the following methods: (a) Adjustment by charging Repair power; (b) According to the health status recovery amount ΔHS1(t+k)=ζ·[HS target -HS1(t+k)]Repair health status; (c) Switch to the backup energy storage unit, and its dispatch amount is backup (+)= backup ·[^(+)-(+)] calculation.

Images