Dynamic power distribution method and device based on hybrid energy storage system

Abstract

The invention provides a dynamic power distribution method and device based on a hybrid energy storage system, and relates to the technical field of hybrid energy storage and power distribution.The method comprises the steps that input power data of a future time interval are predicted through an LSTM network model, and the input power data are separated into a high-frequency component and a low-frequency component through wavelet decomposition; the power input, the equipment charge state and the working temperature of the energy storage system are collected in real time, and a temperature compensation coefficient is generated to correct decomposed components; and in combination with the corrected components and the real-time data, calculating a power distribution proportion between the energy storage devices through a dynamic weighting function to realize dynamic optimal distribution of power. According to the method, by fusing time sequence prediction, dynamic compensation and differential distribution strategies, the power prediction precision, the equipment cooperation efficiency and the adaptability of the system under complex working conditions are remarkably improved, the equipment performance deviation caused by temperature fluctuation is effectively reduced, and the overall operation reliability and working efficiency of the hybrid energy storage system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage systems, and in particular to a method and device for dynamic power distribution based on a hybrid energy storage system. Background Art

[0002] With the large-scale access of renewable energy and the rapid development of smart grids, hybrid energy storage systems have become one of the key technologies for solving the intermittent and volatile problems of renewable energy generation because they can combine the advantages of energy-type and power-type energy storage devices. In scenarios such as wind power and photovoltaics, hybrid energy storage systems can effectively smooth out power fluctuations and improve grid stability by dynamically allocating power tasks. Currently, the deep integration of software technology and energy storage systems has become an industry trend, especially power prediction and dynamic allocation technology based on artificial intelligence, which is regarded as the core direction for improving the response speed and energy efficiency of energy storage systems. However, existing technologies still have significant bottlenecks in prediction accuracy, dynamic environmental adaptability, and multi-device collaborative optimization, resulting in the actual performance of hybrid energy storage systems not being fully utilized, which restricts their large-scale application under complex working conditions.

[0003] Existing power allocation methods for hybrid energy storage systems mostly use fixed rules or traditional signal processing techniques, including decomposing the input power into high-frequency and low-frequency components through Fourier transform or empirical mode decomposition, and allocating them to power-type and energy-type energy storage devices respectively. Such methods rely on static analysis of historical data and do not incorporate dynamic parameters such as the state of charge and temperature of the energy storage equipment in real time, resulting in a rigid allocation strategy. In addition, the power forecasting link often uses an autoregressive integral moving average model or a simple neural network, which has insufficient prediction accuracy for nonlinear and non-stationary power series and does not consider the impact of temperature on equipment performance. Allocation is made directly based on the predicted power, but the decomposed components are not temperature compensated, resulting in a large deviation between the actual allocated power and the theoretical value.

[0004] The above-mentioned existing technologies have the following technical deficiencies: Traditional prediction models have limited ability to capture complex time series characteristics, and prediction errors accumulate in the decomposition and allocation stages, reducing the overall reliability of the system; the correction of high-frequency and low-frequency components does not take into account the differentiated characteristics of energy-type and power-type devices, resulting in inaccurate allocation weight calculations and suboptimal system energy efficiency; the power allocation process lacks dynamic feedback on the real-time status of energy storage devices, and especially in temperature fluctuation scenarios, the lack of a temperature compensation mechanism leads to a mismatch between allocation strategies and actual operating conditions, exacerbating the risk of equipment aging. These technical deficiencies significantly limit the adaptability and operational efficiency of hybrid energy storage systems in dynamic environments. Summary of the Invention

[0005] The object of the present invention is to provide a method and device for dynamic power distribution based on a hybrid energy storage system to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for dynamic power distribution based on a hybrid energy storage system, the method comprising the following steps:

[0008] S1: Set the time interval length and use the LSTM network model to predict the input power of the hybrid energy storage system within a future time interval to obtain the predicted input power data;

[0009] S2: Perform wavelet decomposition on the predicted input power data to obtain the predicted high-frequency component and low-frequency component;

[0010] S3: Acquire the power input data of the energy storage system, the operating data of the energy storage device, and the operating temperature in real time within a future time interval;

[0011] S4: generating a temperature compensation coefficient of the energy storage device according to the temperature data, for correcting the predicted high-frequency component and low-frequency component;

[0012] S5: Based on the power input data of the energy storage system, the charge state of each energy storage device, and the corrected high-frequency component and low-frequency component, the input power of each energy storage device is dynamically allocated within a future time interval.

[0013] Furthermore, in step S1, the LSTM network model is used to predict the input power of the hybrid energy storage system within a future time interval to obtain predicted input power data, including the following steps:

[0014] Set the time interval length to N, collect the power time series data input into the hybrid energy storage system within the time [T0-k*N, T0], divide the input power time series data into time intervals of equal length, and number them in chronological order to form a training set. T0 represents the timestamp of the current moment, and k represents the number of divided time intervals. Build a power prediction model based on the LSTM network, use the power time series data of the previous time interval in the training set as the feature input of the power prediction model, and the power time series data of the next time interval as the label to train the power prediction model.

[0015] Input the power time series data of the kth time interval into the trained power prediction model to obtain the predicted input power data P of the hybrid energy storage system w (t), t is the time variable, and t∈[T0, T0+N].

[0016] Furthermore, when obtaining the predicted high-frequency and low-frequency components, the Daubechies wavelet function is used to transform P w(t) Perform continuous wavelet transform and set the scale parameter to obtain high-frequency components and low-frequency components. The formula is as follows:

[0017]

[0018] Among them, C a (t) and C b (t) represent the low-frequency component and high-frequency component respectively, P w (t) represents the power data at time t in the predicted input power data, t is the independent variable of the integration, ψ * represents the conjugate of the wavelet function, a L and a H Respectively represent the set low-scale parameters and high-scale parameters, and a L >a H .

[0019] Furthermore, the energy storage device includes an energy-type energy storage device and a power-type energy storage device. The operating temperature is the operating temperature at the intersection of the module current paths of the energy-type energy storage device and the operating temperature at the geometric center of the module of the power-type energy storage device. The operating data of the energy storage device includes the rated capacity and state of charge of the energy-type energy storage device and the rated capacity and state of charge of the power-type energy storage device.

[0020] Furthermore, the specific logic for generating the temperature compensation coefficient of the energy storage device based on the temperature data is as follows:

[0021]

[0022] Where: δ(T) a is the temperature compensation coefficient of the energy storage device, T is the operating temperature of the energy storage device at the current moment, is the optimal operating temperature of energy storage equipment, T a_min The lowest operating temperature allowed for energy storage devices, T a_max The maximum operating temperature allowed for energy storage devices. is the temperature sensitivity coefficient of the energy storage device;

[0023] The specific logic for generating the temperature compensation coefficient of the power-type energy storage device based on temperature data is as follows:

[0024]

[0025] Where: δ(T) b is the temperature compensation coefficient of the power type energy storage device, T is the operating temperature of the power type energy storage device at the current moment, is the optimal operating temperature of power type energy storage equipment, T b_min The lowest operating temperature allowed for power-type energy storage equipment, Tb_max The maximum operating temperature allowed for power-type energy storage equipment. is the temperature sensitivity coefficient of energy storage equipment.

[0026] Furthermore, after obtaining the temperature compensation coefficients of the energy storage device and the power storage device, the low-frequency component C a (t) and high frequency component C b (t) is corrected by the temperature compensation coefficient, based on the formula:

[0027] C a ′(t)=δ(T) a ·C a (t)

[0028] C b ′(t)=δ(T) b ·C b (t)

[0029] Where: C a ′(t) and C b ′(t) represents the corrected low-frequency component and the corrected high-frequency component, respectively.

[0030] Furthermore, the input power of each energy storage device is dynamically allocated in a future time interval, and the time interval is: [T0, T0+N]. At the time t0 where allocation is required, the predicted low-frequency component C is known after correction from the predicted data. a ′(t0), predicted high frequency component C b ′(t0), the actual power data to be allocated is collected in real time: P S (t0), so the deviation power ΔP(t0) is as follows:

[0031] ΔP(t0)=[(C a ′(t0)+c b ′(t0))-P S (t0)]

[0032] At the same time, the rated capacity E of the energy storage device is collected in real time nl and state of charge S nl (t0), rated capacity E of power type energy storage device gl and state of charge S gl (t0), design weight function:

[0033]

[0034] Where ∈ is a small constant to avoid division by zero. The deviation is weighted to distribute the corrected high-frequency component and the corrected low-frequency component to calculate the final output. The formula is as follows:

[0035] P low =C a ′(t0)+α(t0)·ΔP(t0)

[0036] P high =C b ′(t0)+(1-α(t0))·ΔP(t0)

[0037] Where: P low is the power component finally allocated to the energy storage device, P high It is the power component finally allocated to the power type energy storage device.

[0038] The present invention further provides a power dynamic distribution device based on a hybrid energy storage system, which is used to implement the above-mentioned power dynamic distribution method based on a hybrid energy storage system, specifically comprising:

[0039] The power prediction module is used to set the time interval length and use the LSTM network model to predict the input power of the hybrid energy storage system within a future time interval to obtain the predicted input power data;

[0040] The power preprocessing module is used to perform wavelet decomposition on the predicted input power data to obtain the predicted high-frequency component and low-frequency component;

[0041] The data acquisition module is used to obtain the power input data of the energy storage system, the state of charge and temperature data of each energy storage device in real time within a future time interval;

[0042] A data correction module, used to generate a temperature compensation coefficient for each energy storage device based on the temperature data, and used to correct the predicted high-frequency and low-frequency components;

[0043] The power distribution module is used to dynamically distribute the input power of each energy storage device within a future time interval based on the power input data of the energy storage system, the charge state of each energy storage device, and the corrected high-frequency component and low-frequency component.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention is based on a power dynamic allocation method for a hybrid energy storage system. It uses an LSTM network model to predict the input power of the hybrid energy storage system in a future time interval to obtain predicted input power data; based on wavelet decomposition technology, it accurately separates high-frequency and low-frequency components, constructs a temperature compensation coefficient, and performs temperature compensation correction on the predicted decomposition components. Finally, a weight function is designed to dynamically allocate power according to the charge state of the energy storage device fed back in real time. This method significantly improves power prediction accuracy, equipment coordination efficiency and system adaptability under complex working conditions by integrating time series prediction, dynamic compensation and differentiated allocation strategies, effectively reduces equipment performance deviations caused by temperature fluctuations, and improves the overall operation reliability and work efficiency of the hybrid energy storage system. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 Schematic diagram of the overall method flow of the present invention;

[0047] Figure 2 Schematic diagram of the module flow of the device of the present invention. DETAILED DESCRIPTION

[0048] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0049] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0050] Example:

[0051] See also Figure 1 , the present invention provides a technical solution:

[0052] A method for dynamic power distribution based on a hybrid energy storage system, comprising the following steps:

[0053] S1: Set the time interval length and use the LSTM network model to predict the input power of the hybrid energy storage system within a future time interval to obtain the predicted input power data.

[0054] In this embodiment, the time interval length is set to N = 15 minutes, and the power time series data input into the hybrid energy storage system in the past 2 hours is collected. The input power time series data is divided according to time intervals of equal length and numbered in chronological order to form a training set. T0 represents the timestamp of the current moment, and k represents the number of divided time intervals, that is, [T0-8*15min, T0]. A power prediction model based on the LSTM network is constructed, and the power time series data of the previous time interval in the training set is used as the feature input of the power prediction model, and the power time series data of the next time interval is used as the label to train the power prediction model.

[0055] A 15-minute time interval can account for short-term load fluctuations and changes in renewable energy. In power grids, changes in load and power generation are often dynamic. Shorter time intervals can better capture these rapid changes and provide timely information for the dynamic power allocation of energy storage systems. The selection of data from the past two hours for prediction provides sufficient historical information to improve the LSTM model's prediction accuracy, enabling it to fully learn recent load patterns and power generation characteristics, thereby optimizing dynamic power allocation strategies and balancing system flexibility and stability.

[0056] Power time series data has obvious time dependence, and past power changes can effectively affect future power demand. By using the power time series data of the previous time interval as feature input, historical information can be fully utilized to predict upcoming power changes, thereby enhancing the accuracy of model predictions. The long short-term memory network (LSTM) is specially designed to process and predict time series data. It can capture long-term dependencies through memory units. Using the data of the previous time interval as input can help LSTM learn the patterns and laws in the time series, thereby improving the prediction effect. The power time series data of the next time interval is used as a label to clearly define the learning objectives of the model and facilitate supervised learning. By learning the mapping relationship between input and label, the model can make accurate power predictions in actual environments. The power time series data of the 8th time interval is input into the trained power prediction model to obtain the predicted input power data P of the hybrid energy storage system. w (t), t is the time variable, and t∈[T0, T0+15min]. After inputting the data of the eighth time interval, the model will generate predictions based on these inputs. Since the model has been trained and the input data is based on the actual power demand in the past, it can reasonably infer the power demand changes between [T0, T0+15min].

[0057] S2: Perform wavelet decomposition on the predicted input power data to obtain predicted high-frequency components and low-frequency components.

[0058] In this embodiment, when obtaining the predicted high-frequency component and low-frequency component, Daubechies wavelet function is used to calculate P w (t) Perform continuous wavelet transform and set the scale parameter to obtain high-frequency components and low-frequency components. The formula is as follows:

[0059]

[0060] Among them, C a (t) and C b (t) represent the low-frequency component and high-frequency component respectively, P w (t) represents the power data at time t in the predicted input power data, t is the independent variable of the integration, ψ * It represents the conjugate of the wavelet function. The scale parameter determines the degree of expansion and contraction of the wavelet function and directly affects the frequency band range after decomposition. L , corresponding to the low-frequency component, capturing the long-term trend; the smaller the scale a H , corresponding to high-frequency components, captures short-term fluctuations, and satisfies a L >a H . Combined with the specific application scenario, the parameters can be set to low-scale parameters a L =8, high-scale parameter a H =4. The high-frequency component represents the short-term fluctuations and transient changes of the corresponding signal. These components change rapidly and have high amplitudes, requiring rapid response capabilities. Power-type energy storage devices have fast charge and discharge rates and high power density, making them suitable for handling high-frequency transient power demands. They also require high response speeds. Power-type devices can quickly handle energy, suppress transient fluctuations, and ensure grid stability. The low-frequency component represents the long-term trend and stable changes of the corresponding signal. These components change slowly and last for a long time, requiring a large energy capacity. Energy-type energy storage devices have high energy density and large capacity, making them suitable for long-term charging and discharging. For sustained energy demands, energy-type devices can provide stable and long-lasting energy support, optimizing system economics.

[0061] Wavelet transform is a multi-scale analysis tool that can analyze signals at different time scales. Through wavelet decomposition, complex signals can be decomposed into different frequency components. Low-frequency components correspond to the long-term trend and short-term fluctuations of the signal, respectively. By setting a larger scale parameter for the low-frequency component, the long-term trend and periodic changes in the signal can be captured. By setting a smaller scale parameter for the high-frequency component, the short-term fluctuations and transient changes in the signal, such as load spikes and abnormal fluctuations, can be captured.

[0062] The advantage of using Daubechies wavelet function for wavelet transform is that it can provide good localization characteristics in time domain and frequency domain. By decomposing the signal into low-frequency and high-frequency components, the patterns and changes in power demand can be more clearly identified in subsequent analysis. The predicted input power data is subjected to wavelet decomposition to obtain high-frequency and low-frequency components. This is not only based on the mathematical characteristics of wavelet transform, but also to effectively capture the long-term trends and short-term fluctuations in the signal, providing more valuable information and data support for the management and dispatch of the power system.

[0063] S3: Obtain the power input data of the energy storage system, the operating data of the energy storage device, and the operating temperature in real time within a future time interval.

[0064] In this embodiment, the energy storage device includes an energy-type energy storage device (supercapacitor) and a power-type energy storage device (lithium-ion battery). The operating temperature refers to the operating temperature at the intersection of the lithium-ion battery module's current paths and the operating temperature at the geometric center of the supercapacitor module. Lithium-ion batteries typically have large capacities and long charge and discharge times. Local overheating may lead to the risk of thermal runaway. By monitoring the temperature at the intersection, power distribution can be adjusted promptly to reduce this risk. Furthermore, the current path intersection is an area of ​​high current density, prone to localized heating. Monitoring the temperature there can more sensitively reflect the internal thermal state of the device, especially the transient temperature rise during high-current charge and discharge. The efficiency of a supercapacitor is significantly affected by its internal resistance, which is closely related to its overall temperature. Monitoring the geometric center temperature can optimize its transient power response efficiency and achieve faster charge and discharge rates, but with lower single-shot energy throughput and relatively uniform heat distribution. The geometric center temperature is a better representation of the device's overall thermal state. The operating data of the energy storage device includes the rated capacity and state of charge of the lithium-ion battery and the rated capacity and state of charge of the supercapacitor. Rated capacity is the basis for calculating the state of charge and determines the upper limit of the device's energy storage. The state of charge reflects the device's current energy storage level.

[0065] S4: Generate temperature compensation coefficients for energy storage devices and power storage devices based on temperature data. The specific logic is as follows:

[0066] Allowable temperature range (T∈[T min ,T max ]): The compensation coefficient is 1, meaning no correction is applied. This implicitly assumes that the device is stable within the allowable temperature range and does not require additional compensation.

[0067] When the temperature range is exceeded The compensation coefficient is 1+k T (TT opt ) 2 , the nonlinear term reflects the effect of temperature deviation from the optimal value. k TTemperature sensitivity coefficient: Indicates the sensitivity of the device to temperature. Smaller, because its capacity decay is relatively slow; using (TT opt ) 2 This demonstrates nonlinear compensation: the further the temperature deviates from the optimal value, the greater the compensation. For example, for every 1°C increase in temperature, the capacity may decrease by 0.2% to 0.5%. The square term can amplify the impact of extreme temperatures.

[0068] The specific logic for constructing the temperature compensation coefficient of energy storage equipment based on temperature data is as follows:

[0069]

[0070] Where: δ(T) a is the temperature compensation coefficient of the energy storage device, T is the operating temperature of the energy storage device at the current moment, The optimal operating temperature for energy storage devices is 25°C, T a_min The lowest operating temperature allowed for energy storage devices is -20°C, T a_max The maximum operating temperature allowed for energy storage devices is 50°C. The temperature sensitivity coefficient of energy storage devices is 0.001 to 0.003. Energy storage devices are relatively sensitive to temperature changes, but the response is relatively slow. When the temperature rises or falls by 1°C, the capacity decays by about 0.2% to 0.5%.

[0071] The specific logic for constructing the temperature compensation coefficient of energy storage equipment based on temperature data is as follows:

[0072]

[0073] Where: δ(T) b is the temperature compensation coefficient of the power type energy storage device, T is the operating temperature of the power type energy storage device at the current moment, The optimal operating temperature for power-type energy storage equipment is 30°C, T b_min The lowest operating temperature allowed for power storage equipment is -10°C, T b_max The maximum operating temperature allowed for power storage devices is 50°C. The temperature sensitivity coefficient of energy storage devices is 0.003 to 0.008, while the instantaneous power throughput of power devices is large, and temperature changes have a more significant impact on efficiency. When the temperature rises or falls by 1°C, the internal resistance may increase by 1% to 3%.

[0074] After obtaining the temperature compensation coefficients of energy storage devices and power storage devices, the low-frequency component Ca (t) and high frequency component C b (t) is corrected by the temperature compensation coefficient, based on the formula:

[0075] C a ′(t)=δ(T) a ·C a (t)

[0076] C b ′(t)=δ(T) b ·C b (t)

[0077] Where: C a ′(t) and C b ′(t) represents the corrected low-frequency component and the corrected high-frequency component, respectively.

[0078] S5: Dynamically allocate the input power of each energy storage device in a future time interval. The time interval is [T0, T0+N]. At the time t0 where allocation is required, the corrected low-frequency component C is obtained from the corrected prediction data. a ′(t0), the corrected high frequency component C b ′(t0), the corrected total power (C a ′(t0)+C b ′(t0)), the actual power data to be allocated P collected in real time S (t0), that is, the deviation power ΔP(t0) is as follows:

[0079] ΔP(t0)=[(C a ′(t0)+C b ′(t0))-P S (t0)]

[0080] When ΔP(t0) is positive, the excess needs to be stored in the energy storage device. When ΔP(t0) is negative, the storage capacity needs to be reduced or energy needs to be released from the energy storage device to make up for the shortfall. At the same time, the rated capacity E of the energy storage device is collected in real time. nl and state of charge S nl (t0), rated capacity E of power type energy storage device gl and state of charge S gl (t0), according to the current SOC of the energy storage device, α(t0) satisfies the interval [0, 1] to avoid overcharging or over-discharging. At the same time, the rated power limit of the energy storage device should be taken into consideration. The weight function is designed as follows:

[0081]

[0082] Among them E gl·(1-S gl (t0)) represents the power type chargeable reserve, E nl ·(1-S nl (t0)) represents the energy type rechargeable remaining capacity, E gl ·S gl (t0) represents the power type discharge margin, E nl ·S nl (t0) represents the energy type discharge margin, ∈ is a small constant to avoid division by zero (such as 10 -6 ) Avoid dividing by zero. The design idea of ​​the weight function is: power-type devices with low state of charge have high charging weights and quickly absorb excess power; energy-type devices with high state of charge have high discharge weights and provide continuous energy support. That is, when ΔP(t0)>0, the actual power exceeds the predicted allocation. If the power-type rechargeable margin is larger, α(t0) is larger, and more power is allocated to the power-type device; when ΔP(t0)<0, the actual power is less than the predicted allocation. If the power-type discharge margin is smaller, α(t0) is smaller, and more power is borne by the energy-type device. The deviation is weighted and the corrected high-frequency component and the corrected low-frequency component are used to calculate the final output. The formula is as follows:

[0083] P low =X a ′(t0)+α(t0)·ΔP(t0)

[0084] P high =C b ′(t0)+(1-α(t0))·ΔP(t0)

[0085] Where: P low is the power component finally allocated to the energy storage device, P high is the power component finally allocated to the power type energy storage device. The corrected low frequency component C a ′(t0) and high frequency component C b ′(t0) is directly assigned to the corresponding device to maintain its basic task, and the weight function is used to allocate the deviation power to complete the final allocation.

[0086] See also Figure 2 The present invention further provides a power dynamic distribution device based on a hybrid energy storage system, wherein the device is used to execute the above-mentioned power dynamic distribution method based on a hybrid energy storage system, comprising:

[0087] The power prediction module is used to set the time interval length and use the LSTM network model to predict the input power of the hybrid energy storage system within a future time interval to obtain the predicted input power data;

[0088] The power preprocessing module is used to perform wavelet decomposition on the predicted input power data to obtain the predicted high-frequency component and low-frequency component;

[0089] The data acquisition module is used to obtain the power input data of the energy storage system, the state of charge and temperature data of each energy storage device in real time within a future time interval;

[0090] A data correction module, used to generate a temperature compensation coefficient for each energy storage device based on the temperature data, and used to correct the predicted high-frequency and low-frequency components;

[0091] The power distribution module is used to dynamically distribute the input power of each energy storage device within a future time interval based on the power input data of the energy storage system, the charge state of each energy storage device, and the corrected high-frequency component and low-frequency component.

[0092] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0093] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution.

[0094] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0095] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A power dynamic allocation method based on a hybrid energy storage system, characterized in that: The dynamic power allocation method comprises the following steps: S1: Set the time interval length and use the LSTM network model to predict the input power of the hybrid energy storage system within a future time interval to obtain the predicted input power data; S2: Perform wavelet decomposition on the predicted input power data to obtain the predicted high-frequency component and low-frequency component; S3: Acquire the power input data of the energy storage system, the operating data of the energy storage device, and the operating temperature in real time within a future time interval; S4: generating a temperature compensation coefficient of the energy storage device according to the temperature data, for correcting the predicted high-frequency component and low-frequency component; S5: Based on the power input data of the energy storage system, the charge state of each energy storage device, and the corrected high-frequency component and low-frequency component, the input power of each energy storage device is dynamically allocated within a future time interval.

2. A method for dynamic power distribution based on a hybrid energy storage system according to claim 1, characterized in that: In step S1, the LSTM network model is used to predict the input power of the hybrid energy storage system within a future time interval to obtain predicted input power data, including the following steps: Set the time interval length to N, collect the power time series data input into the hybrid energy storage system within the time [t0-k*N, T0], divide the input power time series data into time intervals of equal length, and number them in chronological order to form a training set. T0 represents the timestamp of the current moment, and k represents the number of divided time intervals. Build a power prediction model based on the LSTM network, use the power time series data of the previous time interval in the training set as the feature input of the power prediction model, and the power time series data of the next time interval as the label to train the power prediction model. Input the power time series data of the kth time interval into the trained power prediction model to obtain the predicted input power data P of the hybrid energy storage system w (t), t is the time variable, and t∈[T0, T0+N].

3. The method for dynamic power distribution based on a hybrid energy storage system according to claim 2, characterized in that: When obtaining the predicted high-frequency and low-frequency components, the Daubechies wavelet function is used to transform R w (t) Perform continuous wavelet transform and set the scale parameter to obtain high-frequency components and low-frequency components. The formula is as follows: Among them, C a (t) and C b (t) represent the low-frequency component and high-frequency component respectively, P w (t) represents the power data at time t in the predicted input power data, t is the independent variable of the integration, ψ * represents the conjugate of the wavelet function, a L and a H Respectively represent the set low-scale parameters and high-scale parameters, and a L >a H .

4. The method for dynamic power distribution based on a hybrid energy storage system according to claim 1, characterized in that: The energy storage device includes an energy-type energy storage device and a power-type energy storage device. The operating temperature is the operating temperature at the intersection of the module current paths of the energy-type energy storage device and the operating temperature at the geometric center of the module of the power-type energy storage device. The operating data of the energy storage device includes the rated capacity and state of charge of the energy-type energy storage device and the rated capacity and state of charge of the power-type energy storage device.

5. The method for dynamic power distribution based on a hybrid energy storage system according to claim 4, characterized in that: Generate the temperature compensation coefficient of the energy storage device based on the temperature data. The specific logic is: Where: δ(T) a is the temperature compensation coefficient of the energy storage device, T is the operating temperature of the energy storage device at the current moment, is the optimal operating temperature of energy storage equipment, T a_min The lowest operating temperature allowed for energy storage devices, T a_max The maximum operating temperature allowed for energy storage devices. is the temperature sensitivity coefficient of the energy storage device; Generate the temperature compensation coefficient of the power type energy storage device based on the temperature data. The specific logic is: Where: δ(T) b is the temperature compensation coefficient of the power type energy storage device, T is the operating temperature of the power type energy storage device at the current moment, is the optimal operating temperature of power type energy storage equipment, T b_min The lowest operating temperature allowed for power-type energy storage equipment, T b_max The maximum operating temperature allowed for power-type energy storage equipment. is the temperature sensitivity coefficient of energy storage equipment.

6. A method for dynamic power distribution based on a hybrid energy storage system according to claim 5, characterized in that: After obtaining the temperature compensation coefficients of energy storage devices and power storage devices, the low-frequency component C a (t) and high frequency component C b (t) is corrected by the temperature compensation coefficient, based on the formula: C a ′(t)=δ(T) a ·C a (t) C b ′(t)=δ(T) b ·C b (t) Where: C a ′(t) and C b ′(t) represents the corrected low-frequency component and the corrected high-frequency component, respectively.

7. The method for dynamic power distribution based on a hybrid energy storage system according to claim 6, characterized in that: The input power of each energy storage device is dynamically allocated in a future time interval. The time interval is: [T0, T0+N]. For the time t0 at which allocation is required, the predicted low-frequency component C is obtained after correction of the predicted data. a ′(t0), predicted high frequency component C b ′(t0), the actual power data to be allocated is collected in real time: P S (t0), so the deviation power ΔP(t0) is as follows: ΔP(t0)=[(C a ′(t0)+C b ′(t0))-P S (t0)] At the same time, the rated capacity E of the energy storage device is collected in real time nl and state of charge S nl (t0), rated capacity E of power type energy storage device gl and state of charge S gl (t0), design weight function: Where ∈ is a small constant to avoid division by zero. The deviation is weighted to distribute the corrected high-frequency component and the corrected low-frequency component to calculate the final output. The formula is as follows: P low =C a ′(t0)+α(t0)·ΔP(t0) P high =C b ′(t0)+(1-α(t0))·ΔP(t0) Where: P low is the power component finally allocated to the energy storage device, P high It is the power component finally allocated to the power type energy storage device.

8. A power dynamic distribution device based on a hybrid energy storage system, characterized in that: The distribution device is used to execute the power dynamic distribution method based on a hybrid energy storage system according to any one of claims 1 to 7, comprising: The power prediction module is used to set the time interval length and use the LSTM network model to predict the input power of the hybrid energy storage system within a future time interval to obtain the predicted input power data; The power preprocessing module is used to perform wavelet decomposition on the predicted input power data to obtain the predicted high-frequency component and low-frequency component; The data acquisition module is used to obtain the power input data of the energy storage system, the state of charge and temperature data of each energy storage device in real time within a future time interval; A data correction module, used to generate a temperature compensation coefficient for each energy storage device based on the temperature data, and used to correct the predicted high-frequency and low-frequency components; The power distribution module is used to dynamically distribute the input power of each energy storage device within a future time interval based on the power input data of the energy storage system, the charge state of each energy storage device, and the corrected high-frequency component and low-frequency component.

Images