Equipment management and control method for distributed energy storage system cluster based on energy storage EMS

Abstract

The invention relates to an equipment management and control method for a distributed energy storage system cluster based on an energy storage EMS, and the method comprises the steps: obtaining the power load data and prediction demands of a slave control end through a master-slave communication network, carrying out the load prediction through a load automatic prediction model, analyzing the operation characteristics of the slave control end, and optimizing an operation strategy through the combination of prediction information. And key indexes are extracted to establish a parameter form, an optimization value is dynamically configured, and dynamic feedback optimization is performed on the adjusted operation result of the energy storage system. The automatic load prediction model can extract key indexes and circularly optimize the key indexes, update weight values and verify accuracy. And the energy storage EMS system carries out operation adjustment control according to acquisition, extraction, optimization and fine adjustment results, so that the cluster achieves a preset effect. The method can be realized by an independent device or an independent application in an energy storage equipment EMS system, the energy efficiency and the adjusting effect are improved, and the cluster stability is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage system regulation, and is mainly used for a distributed energy storage system cluster to optimize the system operation strategy through an EMS system. It is applicable to distributed energy storage system clusters and also to other energy storage systems with similar structures. Background Art

[0002] In the collaborative work of existing distributed energy storage clusters, the energy storage EMS system cannot efficiently manage and distribute energy during energy conversion and energy efficiency regulation due to the inaccurate operation strategy, which affects the overall energy efficiency performance. During the operation of the distributed energy storage cluster, the energy storage EMS system cannot optimize and adjust the operation strategy in a timely manner. When factors such as power load demand and external environment change, it is unable to respond quickly, making the energy storage system unable to adapt well to new operating conditions, further reducing energy efficiency and affecting the regulation effect of the system. At present, in the operation and management of distributed energy storage clusters, manual adjustment of operation strategies is often required. However, the response speed of manual adjustment strategies is relatively slow, which makes it difficult to meet the rapidly changing operation requirements of distributed energy storage system clusters, and it is unable to respond to various emergencies in a timely and effective manner, affecting the working stability of distributed energy storage clusters. Summary of the invention

[0003] The technical problems to be solved by the present invention are: the operation strategy is not accurate enough; the operation strategy is not optimized in time; and the manual adjustment strategy has a slow response.

[0004] The present invention provides a device control method for a distributed energy storage system cluster based on energy storage EMS, which is applicable to distributed energy storage system clusters and other energy storage systems with similar structures. When a distributed energy storage system cluster works in collaboration, the EMS of one system is used as the master control end, and the remaining systems are used as slave control ends. A master-slave communication network is established using a defined data interface to perform data interaction such as operation data and policy instructions.

[0005] The master end obtains the historical power load data and power load forecasting requirements of all slave ends through the data network, and uses the pre-built load automation forecasting model to perform load forecasting. Analyze the operating characteristics of each slave EMS system, and analyze the operating strategy in combination with the load forecasting information. Extract the key indicators in the forecasting model, establish a parameter form, and dynamically configure the parameter form optimization value according to the strategy analysis results. Constantly provide dynamic feedback on the operating results of the master and slave energy storage systems adjusted by the new parameter table, and further optimize and adjust based on the feedback results.

[0006] The load automation forecasting model performs target analysis on the power load forecasting requirements reported by each energy storage EMS master and slave control terminal of the distributed energy storage cluster, obtains the forecast timeliness and forecast accuracy, and periodically pre-processes the historical power load data based on the forecast timeliness and forecast accuracy, and updates it to the load forecast data.

[0007] Obtain information on factors influencing load forecasting, such as season, weather, holidays, grid price adjustments, peak and valley changes in electricity consumption, etc. Analyze the impact of the influencing factors on power load data, extract key indicators, and obtain parameter tables. Use real-time / historical data models to cyclically optimize parameter tables.

[0008] Use the form optimization results to update the weight values ​​in the prediction model. At the same time, use the real-time optimized operation data to perform a second comparison on the simulation results after substituting the new weight values ​​into the prediction model to determine the accuracy of the aforementioned calculation results.

[0009] The final results of all data collection, key indicator extraction, parameter form optimization, and strategy fine-tuning are operated and regulated by the energy storage EMS system, so that the distributed energy storage system cluster can achieve the preset operating effect.

[0010] The method can be implemented by independent devices acting as a master control end and a slave control end to perform data interaction and logic execution, and can also be implemented as an independent application in an energy storage device EMS system.

[0011] Technical effects of the present invention: The construction of a master-slave structure communication network provides an efficient data interaction basis for the management and control of distributed energy storage system cluster equipment, and realizes clear coordination between the master and slave ends through wired or wireless means. Data interaction and operation strategy optimization enable the master end to obtain the operation data of the slave end in real time, accurately formulate and adjust the operation strategy, realize dynamic management and control, and improve system efficiency and stability. The load forecasting model and parameter configuration provide reliable data support for the operation strategy of the energy storage system by accurately predicting the power load, dynamically configure the parameter form optimization value, and ensure accurate system management and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The present invention is further described in detail below in conjunction with the accompanying drawings: Figure 1 It is the control logic diagram of the present invention. DETAILED DESCRIPTION

[0013] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0014] The present invention provides an equipment control method for a distributed energy storage system cluster based on energy storage EMS. When the distributed energy storage system cluster works in collaboration, the EMS of one of the systems is usually used as the master control end to perform cluster collaborative operation, and a communication network with a master-slave structure is established using a defined data interface for data interaction, including operation data and policy instructions.

[0015] The master control end obtains the historical power load data and power load forecasting requirements of all slave control ends through the data network, and performs load forecasting through the pre-built load automation forecasting model; analyzes the operating characteristics of each slave control end EMS system, and analyzes the operating strategy in combination with the load forecasting information; extracts key indicators in the forecasting model to establish a parameter form, and dynamically configures the parameter form optimization value according to the strategy analysis results; and continuously provides dynamic feedback on the operating results of the master and slave control end energy storage systems adjusted by the new parameter table. This solves the technical problems of low energy efficiency and slow response to manual adjustment strategies in the collaborative work of existing distributed energy storage clusters, and the lack of accurate and timely optimization of operating strategies in the energy conversion and energy efficiency regulation process of energy storage EMS systems, so as to achieve the goals of accurately calculating power load, improving energy efficiency and regulation effects, and enhancing the stability of distributed energy storage clusters.

[0016] The load automation prediction model can perform target analysis on the power load prediction requirements reported by each energy storage EMS master and slave control terminal of the distributed energy storage cluster, obtain the prediction timeliness and prediction accuracy, and periodically preprocess the historical power load data according to the prediction timeliness and prediction accuracy, and update it to the load prediction data. Obtain information on factors affecting load forecasting, including seasons, weather, holidays, grid price adjustments, peak and valley changes in electricity consumption, etc. Extract key indicators based on whether the influencing factor information affects the power load data, and obtain a parameter form. Use the real-time / historical data model to cyclically optimize the parameter form. Use the form optimization results to update the weight values ​​in the prediction model, and use the real-time optimized operation data to perform a secondary comparison on the simulation results after substituting the new weight values ​​into the prediction model to determine the accuracy of the aforementioned calculation results.

[0017] The final results of all data collection, key indicator extraction, parameter form optimization, and strategy fine-tuning are operated and regulated by the energy storage EMS system, so that the distributed energy storage system cluster can achieve the preset operating effect.

[0018] The method described in this patent can be implemented by independent devices as the master and slave ends for data interaction and logic execution, or it can be implemented as an independent application in the EMS system of energy storage equipment.

[0019] Use the defined data interface to establish a master-slave communication network to achieve data interaction between the master and slave ends.

[0020] The data exchanged in the data interaction includes operation data and strategy instructions. Operation data mainly refers to various types of data generated by the energy storage system during operation, such as power load data and charge and discharge status data of the energy storage equipment; strategy instructions are obtained by the master control end based on load forecasting, operation strategy calculations, etc., and are used to guide the slave control end energy storage system to perform corresponding operations, such as charge and discharge power adjustment instructions, operation mode switching instructions, etc.

[0021] The communication network adopts a master-slave structure, with one energy storage system's EMS as the master control end and the other energy storage system's EMS as the slave control end. The master control end is responsible for centralized data processing, formulating strategies and issuing instructions, while the slave control end receives instructions and executes corresponding operations, and at the same time feeds back its own operating data to the master control end.

[0022] Power load data: includes historical power load data and real-time power load data. Historical data is used to train and optimize the load forecasting model, while real-time data is used to monitor the operating status of the energy storage system in real time and adjust the operating strategy in a timely manner. Parameters such as data collection frequency, accuracy, and time span need to be determined based on the actual application scenario and the accuracy requirements of load forecasting.

[0023] Energy storage equipment status data: such as the state of charge (SOC), state of health (SOH), charge and discharge power, charge and discharge current, voltage and other parameters of the energy storage battery. These data can reflect the current operating status and performance level of the energy storage equipment, and provide a basis for the main control end to calculate the operating strategy and optimize the parameters.

[0024] Charge and discharge power instructions: Based on the load forecast and operation strategy calculation results, the master control end will send charge and discharge power instructions to the slave control end. The instructions should clearly state the specific value, duration and other parameters of the charge and discharge power to guide the slave control end energy storage device to perform charge and discharge operations according to the predetermined power.

[0025] Operation mode switching instructions: Under different operating conditions, the energy storage system may need to switch to different operation modes, such as charging mode, discharging mode, standby mode, etc. The instructions should include specific requirements for mode switching, including parameters such as switching conditions and switching time, to ensure that the slave control end can accurately perform mode switching operations.

[0026] Forecast timeliness: refers to the time range of load forecasting, that is, how long in the future the power load is predicted. For example, it can be a load forecast for the next few hours, days or weeks. The determination of forecast timeliness needs to comprehensively consider the application scenarios and actual needs of the energy storage system. For example, for energy storage systems that need to adjust their operating strategies in real time, a shorter forecast timeliness may be required to respond to load changes in a timely manner; while for long-term planning and scheduling, a longer forecast timeliness may be required.

[0027] Prediction accuracy: indicates the accuracy of load forecast results, usually measured by error range or confidence interval. Improving prediction accuracy can be achieved by optimizing the prediction model, increasing the amount of data, improving data preprocessing methods, etc. In practical applications, it is necessary to determine the appropriate prediction accuracy target based on the operating requirements and risk tolerance of the energy storage system. For example, for scenarios with high requirements for power supply reliability, higher prediction accuracy may be required.

[0028] Data cleaning parameters: including outlier detection and processing methods, missing value filling strategies, etc. For example, statistical methods can be used to detect and remove outliers, or interpolation methods can be used to fill missing values.

[0029] Data normalization parameters: Determine the scope and method of normalization, such as normalizing the data to the interval [0,1] or [-1,1]. Commonly used normalization methods include maximum-minimum normalization, Z-score normalization, etc.

[0030] Data smoothing parameters: Select appropriate smoothing algorithms and parameters, such as the window size of the moving average method, the smoothing coefficient of the exponential smoothing method, etc., to achieve the purpose of smoothing data and reducing noise.

[0031] Collect information on various influencing factors related to power load, such as seasons, weather, holidays, grid price adjustments, peak and valley changes in electricity consumption, etc. These factors have a significant impact on the fluctuation and change of power load. By obtaining this information, the change trend of power load can be predicted more accurately.

[0032] According to the acquired influencing factor information, the power load data is analyzed to extract the key indicators related to these factors. These key indicators can reflect the main characteristics and laws of load changes and provide a basis for subsequent parameter form optimization. For example, for weather factors, key indicators such as temperature, humidity, and wind speed can be extracted; for holiday factors, key indicators such as holiday type and holiday length can be extracted.

[0033] The extracted parameter form is optimized cyclically using the real-time / historical data model. By continuously adjusting the parameter values ​​in the parameter form, the model can better fit the actual load data and improve the accuracy and reliability of the prediction. The optimization process can use machine learning algorithms, statistical analysis methods, etc. to adjust the parameters according to the prediction error of the model until a satisfactory optimization effect is achieved.

[0034] Seasonal parameters include season type (spring, summer, autumn, winter), season length, etc. The power load characteristics of different seasons are different. For example, in summer, the load is usually higher due to the use of cooling equipment such as air conditioners; in winter, the load is also relatively high due to the use of heating equipment.

[0035] Weather parameters include temperature, humidity, wind speed, precipitation, etc. Temperature is one of the important factors affecting power load. High or low temperature weather will cause significant changes in load. Humidity and wind speed will also affect people's electricity consumption behavior. For example, when the humidity is high, the use of dehumidification equipment may increase.

[0036] Holiday parameters include holiday type (legal holidays, weekends, etc.), holiday length, working days before and after holidays, etc. The electricity consumption pattern on holidays is quite different from that on weekdays. Usually, the load on holidays is lower, but some special holidays (such as Spring Festival, National Day, etc.) may have special electricity demand.

[0037] The grid price adjustment parameters include the time, adjustment range, and price structure of the electricity price adjustment. The adjustment of electricity prices will affect the user's electricity consumption behavior. For example, during periods of low electricity prices, users may increase their electricity consumption; during periods of high electricity prices, users may reduce their electricity consumption.

[0038] The parameters of electricity consumption peak-valley variation include the division of peak-valley time periods, peak-valley electricity price ratio, peak-valley load difference, etc. The electricity load difference between peak and valley time periods is large. By reasonably dividing the peak and valley time periods and adjusting the peak and valley electricity prices, users can be guided to use electricity during valley time periods and reduce the load pressure during peak time periods.

[0039] Before extracting key indicators, the original power load data needs to be preprocessed, such as data cleaning, data normalization, data smoothing, etc. Data cleaning can remove outliers and noise in the data; data normalization can convert the data into a unified range for subsequent processing; data smoothing can reduce data fluctuations and make the data smoother. For example, methods such as moving average method and Savitzky-Golay filter can be used for data smoothing.

[0040] According to the influencing factor information, select the key indicators related to the load change. You can use correlation analysis, principal component analysis and other methods to filter out the key indicators that have a greater impact on the load change from a large number of influencing factors. For example, by calculating the correlation coefficient between temperature and load, select temperature as one of the key indicators.

[0041] Select a suitable optimization algorithm to optimize the parameter form, such as gradient descent, genetic algorithm, particle swarm optimization algorithm, etc. Different optimization algorithms have different characteristics and applicable scenarios, and it is necessary to select a suitable algorithm according to the specific problem. For example, the gradient descent method is suitable for problems with a small parameter space and a smoother objective function; the genetic algorithm is suitable for problems with a large parameter space and a more complex objective function.

[0042] Determine the objective function for optimization, usually with the goal of minimizing the prediction error. The prediction error can be measured using indicators such as mean square error and absolute error. By minimizing the prediction error, the model can better fit the actual load data and improve the accuracy of the prediction.

[0043] Set the number of iterations of the optimization process, which is determined according to the convergence of the model and the limitation of computing resources. Too many iterations may lead to too long computing time, while too few iterations may not achieve satisfactory optimization results. You can adjust the number of iterations by observing the trend of the prediction error of the model, and stop the iteration when the prediction error tends to be stable.

[0044] Use the form optimization results to update the weight values ​​in the load forecasting model. The weight value update is based on the optimal parameters obtained during the parameter form optimization process. These parameters can better reflect the relationship between the load forecasting influencing factors and the power load, so that the forecasting model can more accurately predict the changes in the power load.

[0045] Using the real-time optimized operation data, the new weight values ​​are substituted into the prediction model for simulation to obtain the simulation results. Then, the simulation results are compared with the actual operation data for a second time, that is, comparative analysis, to determine the accuracy of the above calculation results. This process can verify the performance of the updated model, ensure that the optimization of the model is effective, and improve the accuracy and reliability of the prediction.

[0046] Determine the update method of the weight value, such as using a simple replacement update, that is, directly replacing the weight value in the original model with the optimized new weight value; or using a progressive update, gradually adjusting the weight value according to a certain learning rate, so that the model can smoothly transition to the new parameter state.

[0047] In progressive updates, the learning rate is an important parameter that controls the speed and magnitude of weight value updates. A learning rate that is too large may cause the model parameters to update too quickly, resulting in oscillations; a learning rate that is too small will cause the model to update too slowly, affecting optimization efficiency. It is usually necessary to determine the appropriate learning rate value through experiments and debugging.

[0048] Determine the time range of the simulation, that is, how long in the future the power load situation will be simulated. The simulation time range should match the forecasting requirements in actual applications. For example, if short-term load forecasting is required, the simulation time range can be a few hours or days in the future; if long-term load forecasting is required, the simulation time range can be a few weeks or months in the future.

[0049] Simulation step size: The simulation step size refers to the time interval between each simulation during the simulation process. The choice of simulation step size will affect the simulation accuracy and computational efficiency. A smaller simulation step size can improve the simulation accuracy, but will increase the amount of computation; a larger simulation step size can improve the computational efficiency, but may reduce the simulation accuracy. The appropriate simulation step size needs to be determined based on actual needs and computing resource limitations.

[0050] Select appropriate comparison indicators to measure the difference between simulation results and actual operation data. Common comparison indicators include mean square error (MSE), mean absolute error (MAE), root mean square error (RMSE), etc. These indicators can reflect the size of the prediction error from different angles. By comparing the values ​​of these indicators, the performance of the updated model can be evaluated.

[0051] Verification threshold: Set the threshold of the verification index. When the value of the verification index is lower than the threshold, the model update is considered to be effective and the prediction result has a high accuracy. When the value of the verification index is higher than the threshold, further analysis of the cause is required, and the model parameters or optimization methods may need to be readjusted. The setting of the threshold needs to be determined according to the actual application scenario and the requirements for prediction accuracy, and can usually be obtained through experiments and experience.

[0052] The energy storage EMS system performs operation adjustment and control on the distributed energy storage system cluster based on the final results of data collection, key indicator extraction, parameter form optimization, and strategy fine-tuning. This process is to apply the results of the previous series of analysis, optimization, and calculation to the actual operation of the energy storage system, and to enable the distributed energy storage system cluster to work according to the preset operation effect by adjusting the parameters such as the charging and discharging power and operation mode of the energy storage equipment.

[0053] Through the operation regulation and control of the energy storage EMS system, the distributed energy storage system cluster can achieve the preset operation effect. The preset operation effect may include but is not limited to load balancing, cost optimization, improved system stability, faster response speed, etc. For example, when the grid load is peak, the energy storage system can discharge in time to relieve the pressure on the grid; when the load is low, the energy storage system can charge efficiently to reduce operating costs.

[0054] The charge and discharge power control includes the set value, regulation rate, power range, etc. of the charge and discharge power. The set value is calculated based on the load forecast and operation strategy, and is used to guide the charge and discharge power of the energy storage device at a specific moment; the regulation rate refers to the speed at which the charge and discharge power changes in a unit of time. A too fast regulation rate may cause damage to the energy storage device, and a too slow regulation rate may not be able to respond to load changes in time; the power range is the charge and discharge power interval in which the energy storage device can operate safely. Exceeding this range may cause equipment failure.

[0055] Operation mode control involves the switching of operation modes of energy storage equipment, such as charging mode, discharging mode, standby mode, power control mode, SOC control mode, etc. The switching conditions, duration, parameter changes during the switching process, etc. of each mode need to be clarified. For example, in the power control mode, the target power value and the allowable range of power fluctuation need to be set; in the SOC control mode, the upper and lower limits of the SOC and the adjustment accuracy need to be set.

[0056] Load balance includes load deviation rate, load balance degree, etc. Load deviation rate refers to the ratio of the deviation between actual load and predicted load to predicted load. By controlling the load deviation rate within a certain range, relative load balance can be achieved; load balance degree is an indicator to measure the uniformity of load distribution between energy storage devices in the energy storage system cluster. A higher load balance degree helps to improve the overall operating efficiency and reliability of the system.

[0057] Cost optimization includes operating costs, maintenance costs, equipment depreciation costs, etc. By optimizing the operation strategy of the energy storage system, such as charging when electricity prices are low and discharging during peak hours, the operating costs can be reduced; at the same time, reasonable operation regulation and control can reduce the failure rate of equipment, reduce maintenance costs and equipment depreciation costs, thereby achieving comprehensive cost optimization.

[0058] System stability, such as voltage stability, frequency stability, power factor, etc. Voltage stability and frequency stability are important indicators for measuring the stable operation of the power system. The energy storage system can adjust the voltage and frequency of the power grid through charging and discharging operations to maintain it within a stable range; the power factor reflects the energy exchange efficiency between the energy storage system and the power grid. A higher power factor helps improve the stability and operating efficiency of the system.

[0059] Response speed refers to the time it takes for an energy storage system to respond to load changes or control commands. When the grid load changes rapidly or an emergency occurs, the energy storage system needs to be able to respond quickly, adjust the charge and discharge power, or switch the operating mode in a timely manner. The speed of response directly affects the energy storage system's ability to support the grid and handle emergencies. It is usually necessary to improve the response speed by optimizing the control algorithm and improving equipment performance.

[0060] Finally, it should be pointed out that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the above embodiments, a person skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A device control method for a distributed energy storage system cluster based on energy storage EMS, characterized in that: When a distributed energy storage system cluster works together, one of the EMS systems is used as the master control end, and the other systems are used as slave control ends. The master control end first obtains the historical power load data and power load forecast requirements of all slave control ends through the data network; Import the index value into the load automation prediction model for load prediction; according to the operating characteristic values ​​of each slave EMS system, collect the load prediction information to calculate the operation strategy; extract the key index result value in the prediction model, and dynamically configure the parameter form optimization value according to the strategy analysis results; obtain the operation result feedback of the master and slave energy storage systems after the EMS is adjusted according to the new parameter table; The operation feedback is used as the basis for the next model operation to iterate and optimize the model.

2. The device control method of a distributed energy storage system cluster based on energy storage EMS according to claim 1, characterized in that: Use the defined data interface to establish a master-slave communication network to exchange data such as operating data and policy instructions.

3. The device control method of a distributed energy storage system cluster based on energy storage EMS according to claim 2 is characterized in that: The load automation forecasting model performs target analysis on the power load forecasting requirements reported by each energy storage EMS master and slave control terminal of the distributed energy storage cluster, obtains the forecast timeliness and forecast accuracy, and periodically pre-processes the historical power load data based on the forecast timeliness and forecast accuracy, and updates it to the load forecast data.

4. The device control method of a distributed energy storage system cluster based on energy storage EMS according to claim 3 is characterized in that: Obtain information on factors influencing load forecasting, such as season, weather, holidays, grid price adjustments, changes in electricity consumption peaks and valleys, etc., extract key indicators based on the impact of influencing factor information on power load data, obtain parameter forms, and use real-time / historical data models to cyclically optimize the parameter forms.

5. The device control method of a distributed energy storage system cluster based on energy storage EMS according to claim 4 is characterized in that: Use the form optimization results to update the weight values ​​in the prediction model. At the same time, use the real-time optimized operation data to perform a second comparison on the simulation results after substituting the new weight values ​​into the prediction model to determine the accuracy of the aforementioned calculation results.

6. The device control method of a distributed energy storage system cluster based on energy storage EMS according to claim 5 is characterized in that: The final results of data collection, key indicator extraction, parameter form optimization, and strategy fine-tuning are all operated and regulated by the energy storage EMS system, so that the distributed energy storage system cluster can achieve the preset operating effect.

Images