Balanced control method for health states of multiple energy storage units connected in parallel

Abstract

The invention discloses a balance control method for health states of multiple parallel energy storage units. Firstly, a communication-free state-of-health online estimation method based on a battery semi-empirical life model is deduced, the method does not need communication, is small in calculation amount, considers the rated capacity difference, is suitable for the dynamic degradation process of a battery, and is popularized to different rated capacities by introducing the concept of weighted ampere-hour throughput. Then, in order to meet the requirements of health state equalization, accurate current distribution and average direct current bus voltage recovery, a hierarchical control structure is established. Through coordination and cooperation among the main control layer, the secondary control layer and the communication layer in the hierarchical control structure, the provided strategy can realize the control targets of health state balance, accurate current distribution and average DC bus voltage recovery, and can greatly reduce the communication burden of the system.

Description

Technical Field

[0001] The present invention belongs to the field of DC microgrids and relates to a method for balancing the health states of parallel multiple energy storage units. Background Art

[0002] In recent years, driven by the goals of "carbon peak and carbon neutrality", electric vehicles and DC microgrids composed of clean and renewable energy sources such as solar energy and wind energy have been vigorously developed. However, due to the intermittent and volatile characteristics of renewable energy, high-penetration renewable energy will affect the safe and stable operation of DC microgrids. A battery energy storage system should be set up to perform peak shaving and valley filling for the DC microgrid, thereby improving the stability of the DC microgrid. In the electric vehicle industry, when the battery capacity decays to less than 80% of its nominal capacity, it is usually retired, but it still has great remaining utilization value. When the retired battery has an intact appearance and its functions have not failed, it can be reused and applied to the battery energy storage system. With the vigorous promotion of electric vehicles in recent years, a large number of retired batteries will be generated in the future. The cascade utilization of retired batteries can effectively reduce the environmental pollution caused by retired batteries, reduce the cost of the battery energy storage system, and extend the service life of power batteries, which has important practical significance.

[0003] In a battery energy storage system, multiple recycled batteries are connected in series to form an energy storage unit. Since the capacity of a single energy storage unit is small, in order to increase the total capacity and reliability of the battery energy storage system, multiple groups of energy storage units are usually connected in parallel to the DC bus. The unbalanced health states of parallel multiple energy storage units will lead to inconsistent scrapping times of different energy storage units, resulting in the premature scrapping of a certain energy storage unit. The remaining energy storage units may not be able to withstand the load power or renewable energy generation, thus affecting the stability of the DC microgrid. Currently, the most commonly used traditional state of charge balancing method for parallel multiple energy storage units cannot achieve the health state balance between different energy storage units. Instead, it will cause the energy storage unit with the lowest health state to drop to the retired level first, and it is not applicable to the scenario of cascade utilization of retired batteries. Summary of the Invention

[0004] In order to solve the problems that the traditional state of charge balancing method in a DC microgrid cannot achieve the health state balance between multiple energy storage units and has a heavy communication burden, etc., the technical solution adopted by the present invention is: a method for balancing the health states of parallel multiple energy storage units, including the following steps:

[0005] Step 1: Build a DC microgrid structure including an energy storage unit, a power generation unit, a load unit, and a DC bus;

[0006] Step 2: Adopt an online estimation method for the health state of the battery to estimate the health state of the battery modules in each energy storage unit under different rated capacities;

[0007] Step 3: Adopt a hierarchical control method. Based on the estimated values of the state of health of the battery modules under different rated capacities in each energy storage unit, achieve the state of health balance of the battery modules in all energy storage units, and accurately distribute the current according to the ratio of the rated capacity and restore the average DC bus voltage to the rated value.

[0008] Further: The hierarchical control adopted includes a primary control layer, a secondary control layer, and a communication layer;

[0009] The primary control layer is used to control the energy storage unit by adopting a voltage-current double closed-loop PI control. Based on the adaptive droop control method, introduce a speed adjustment factor and a precision adjustment factor for controlling the state of health balance of the battery into the droop coefficient to improve the speed and precision of the SoH balance;

[0010] The secondary control layer: is used to adopt a simplified secondary control method. Based on a secondary control loop regulated by an integrator, achieve the accurate distribution of the current of the batteries in the energy storage unit according to the ratio of the rated capacity and restore the average DC bus voltage to the rated value;

[0011] The communication layer: is used to exchange information with adjacent nodes based on the method that each energy storage unit relies only on neighbor communication, and use the dynamic consensus algorithm to achieve the stable convergence of the global average state variables.

[0012] Further: The expression of the online estimation method for the state of health of the battery is as follows:

[0013]

[0014] Where: SoH p and SoH p+1 are the online estimated values of SoH at time t p and t p+1 respectively. A refers to the pre-factor, E a is the activation energy, C rate is the battery charge and discharge rate, B is the compensation factor of C rate , R is the gas constant, T bat is the absolute temperature of the battery, Z is the time factor, ΔA heff is the effective ampere-hour throughput during the period from t p to t p+1 ,

[0015] The effective ampere-hour throughput ΔA under nominal operating conditions heff By introducing the weighted ampere-hour throughput, convert the actual ampere-hour throughput ΔA hact caused by any charge / discharge event into the effective ampere-hour throughput ΔA heff , and the expression is as follows:

[0016]

[0017] where Q A and Q R are the battery capacities under actual and rated operating conditions respectively, and i bat is the battery output current.

[0018] Furthermore: the expression of the droop coefficient is as follows:

[0019]

[0020] where R vi is the improved dynamic droop coefficient, Q ratedmax is the capacity of the energy storage unit (ESU) with the largest rated capacity, and all ESUs have the same value. SoH avgi is the average SoH calculated locally in the ESU i , R v0 is the initial droop coefficient, and ρ and ε are the adjustment factors that control the SoH balancing speed and accuracy respectively. ρ mainly affects the SoH balancing speed in the initial stage, and ε mainly affects the SoH balancing speed and accuracy in the final stage.

[0021] Furthermore: the average state of health S o H avgi , can be obtained through a dynamic consistency algorithm.

[0022] Furthermore: the expression of the secondary control loop regulated by an integrator is as follows:

[0023]

[0024] where: δv i is the DC bus voltage compensation term determined by the secondary control layer, κ is the integral gain, u nom is the rated DC bus voltage, ζ avgi and ψ i are the average virtual state variable and voltage compensation factor of the ESU i respectively, and are designed as:

[0025]

[0026] where ζ i is the virtual state variable of the ESU i , ζ avgi is the average virtual state variable calculated locally in the ESU i , u dci is the output voltage of the ESU i , and ξ is a factor to prevent ψ iGain of zero (0 < ξ < 1), select ξ = 1 / 2, Δu max is the maximum allowable offset amplitude of the DC bus voltage, Δu max = 0.1u nom , Δu viri is the virtual voltage drop of the ESU i and is designed as:

[0027] Δu viri = R vi i dci (6)

[0028] where R vi is the dynamic droop coefficient designed in step 3 of claim 1, i dci is the output current of the ESU i .

[0029] Furthermore: The average virtual state variable ζ avgi can be obtained through the dynamic consensus algorithm.

[0030] Furthermore: The expression of the dynamic consensus algorithm is as follows:

[0031]

[0032] where x i = [SoH avgi , ζ avgi T , γ i = [SoH i , ζ i T , x i (t) and x j (t) are the state variables of the ESU i and the ESU j respectively, γ i (t) is the local input variable of the ESU i , where a ij is the communication weight between the ESU i and the ESU j , select a ring topology as the communication topology between ESUs, that is, each ESU only communicates with adjacent ESUs to reduce communication costs and ensure the convergence of relevant variables in case of communication failures, N i is the set of nodes connected to the ESU i .

[0033] Furthermore: The master control layer uses a voltage-current double closed-loop PI control to achieve the control of the energy storage unit, and the specific process is as follows:

[0034] The rated DC bus voltage u​​nom Subtract the virtual voltage drop Δu i from the ESU viri and then add the result to the DC bus voltage compensation term δv i to obtain the output voltage reference value u i of the ESU refi ;

[0035] Subtract the actual output voltage u refi from the output voltage reference value u dci and, through PI regulation of the voltage outer loop, obtain the reference value i refi of the lithium battery inductor current;

[0036] Subtract the actual value i refi of the lithium battery inductor current from the reference value i bati of the battery inductor current and, through PI regulation of the current inner loop, obtain two complementary pulse width modulation (PWM) waves;

[0037] Control two insulated gate bipolar transistors in the bidirectional DC / DC converter respectively with the two complementary PWM waves. Since the voltage-current double closed loop has the characteristic of fast dynamic response, the actual output voltage u dci can quickly track the output voltage reference value u refi so as to achieve the purpose of controlling the power output of multiple parallel energy storage units.

[0038] Furthermore: The rules for selecting ρ and ε are as follows:

[0039] In the initial stage: When ΔSoH i is large, in order to avoid the simultaneous influence of ρ and ε on the convergence of SoH, it is necessary to set ρ|ΔSoH i | >> ε; Therefore, ρ needs to be at least one order of magnitude larger than ε, that is, ρ > 10ε. At this time, the influence of ε is ignored and the SoH equalization speed is mainly determined by ρ, and K i ≈ 1 / ρ|ΔSoH i |;

[0040] In the final stage: When ΔSoH i is close to 0, ρ|ΔSoH i | ≈ 0. In order to obtain a faster equalization speed and higher equalization accuracy when the SoH difference is small, it is necessary to have ε << 1. At this time, the SoH equalization speed and accuracy are mainly determined by ε, and K i ≈ 1 / ε.

[0041] A method for balancing the state of health of multiple parallel energy storage units provided by the present invention first derives an online state of health estimation method without communication based on a semi-empirical battery life model, and generalizes this method to the case of different rated capacities by introducing the concept of weighted ampere-hour throughput.

[0042] Then, in order to meet the requirements of health state balance, precise current distribution, and average DC bus voltage restoration, a hierarchical control structure was established;

[0043] In the primary control layer, the droop coefficient is adaptively adjusted according to the dynamic change of the online estimated value of the health state in each energy storage unit to achieve health state balance, and two balance adjustment factors are introduced to improve the speed and accuracy of health state balance.

[0044] In the secondary control layer, a secondary control loop regulated only by one integrator is designed to eliminate the influence of line impedance on health state balance and achieve average DC bus voltage restoration.

[0045] In the communication layer, each energy storage unit (lithium battery and corresponding bidirectional DC / DC converter) exchanges information with adjacent nodes only by relying on neighbor communication, and uses the dynamic consensus algorithm to achieve stable convergence of global average state variables. This distributed control method can avoid the disadvantages of the centralized control method relying on the communication network and having a heavy communication burden, has higher control accuracy than the decentralized control method, and also has a similar "plug and play" function.

[0046] Finally, a Matlab / Simulink simulation model of the DC microgrid was built to verify the excellent performance of the proposed method in achieving health state balance, precise current distribution, and ensuring DC bus voltage stability. Brief Description of the Drawings

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0048] Figure 1 is the DC microgrid which is the research object of the present invention;

[0049] Figure 2 is the control block diagram of the energy storage unit proposed by the present invention;

[0050] Figure 3 is the dynamic droop coefficient R proposed by the present invention vi varying with ΔSoH i characteristic curve;

[0051] Figure 4(a) is the SoC change curve of the traditional SoC equalization method in Case 1, (b) is the SoH change curve of the traditional SoC equalization method in Case 1, (c) is the SoH change curve of the present invention in Case 1, (d) is the ESU output current change curve of the present invention in Case 1, and (e) is the average DC bus voltage change curve of the present invention in Case 1;

[0052] Figure 5 (a) is the SoH change curve of the present invention in Case 2, (b) is the ESU output current change curve of the present invention in Case 2, and (c) is the average DC bus voltage change curve of the present invention in Case 2; Detailed implementation manners

[0053] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other. The present invention will be described in detail below with reference to the drawings and in combination with the embodiments.

[0054] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The following description of at least one exemplary embodiment is actually only illustrative and in no way restrictive of the present invention and its application or use. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0055] A parallel multi-energy storage unit state of health equalization control method includes the following steps:

[0056] Step 1: Build a DC microgrid topology microgrid structure including an energy storage unit, a power generation unit, a load unit, and a DC bus;

[0057] Step 2: Adopt an online battery state of health estimation method to estimate the state of health of battery modules with different rated capacities in the energy storage unit;

[0058] Step 3: Adopt a hierarchical control method to achieve the state of health equalization of battery modules in all energy storage units based on the state of health estimation values of battery modules with different rated capacities in each energy storage unit, and accurately distribute the current according to the ratio of the rated capacities and restore the average DC bus voltage to the rated value.

[0059] The adoption of hierarchical control includes a main control layer, a secondary control layer, and a communication layer;

[0060] The main control layer is used to control the energy storage unit by adopting voltage-current double-closed-loop PI control. Based on the adaptive droop control method, a speed adjustment factor and a precision adjustment factor for controlling the state of health (SoH) balance of the battery are introduced into the droop coefficient to improve the balance speed and precision of SoH.

[0061] The secondary control layer: is used to adopt a simplified secondary control method and, based on a secondary control loop regulated by an integrator, achieve the accurate current distribution of the batteries in the energy storage unit according to the proportion of the rated capacity and restore the average DC bus voltage to the rated value.

[0062] The communication layer: is used to exchange information with adjacent nodes based on the method that each energy storage unit relies only on neighbor-to-neighbor communication, and use the dynamic consensus algorithm to achieve the stable convergence of the global average state variables (i.e., SoH avgi and ζ avgi these two average state variables).

[0063] As Figure 1 shown in the DC microgrid structure, it is the object of a method for balancing the state of health of multiple parallel energy storage units implemented in this application. The black solid line represents the electrical link, the blue dashed arrow represents the communication link between energy storage units, and the red dashed arrow represents the direction of energy transfer. The system architecture is divided into two parts: the physical layer and the communication layer. In the physical layer, it mainly consists of a power generation unit, an energy storage unit, a load unit, and a DC bus composed of their respective power electronic converters. Each module is connected in parallel to the common DC bus through its own converter. Each energy storage unit consists of a battery and its bidirectional DC / DC converter; in the communication layer, the information exchange between adjacent energy storage units is realized based on a neighbor-to-neighbor communication network.

[0064] The power generation unit includes a photovoltaic array and the main grid;

[0065] Each energy storage unit (ESU) consists of a lithium battery and its bidirectional DC / DC converter. The corresponding DC / DC converters are connected in parallel to the common DC bus. Usually, multiple groups of ESU are connected in parallel to the DC bus to improve the total capacity and reliability of the battery energy storage system (BESS).

[0066] Photovoltaic arrays usually operate in the maximum power point tracking (MPPT) mode to make full use of solar energy. When the state of charge (SoC) of the energy storage unit exceeds the upper limit, the photovoltaic array switches to the constant voltage control mode. The battery energy storage system mainly plays the role of "peak shaving and valley filling", responsible for smoothing the power fluctuations of the photovoltaic array and the load, and maintaining the stability of the DC bus voltage. The performance of the battery energy storage system directly affects the reliability of the DC microgrid. This method can achieve the health state balance of multiple parallel energy storage units in practical applications, thus extending the service life of the battery energy storage system and improving the reliability of the DC microgrid.

[0067] As Figure 2 shown in the control block diagram of the energy storage unit, a hierarchical control structure is adopted, mainly divided into the main control layer, the secondary control layer and the communication layer.

[0068] In the main control layer, an online estimation method of the battery health state without communication and with small computational load is introduced, and this method is extended to the case of different rated capacities by introducing the concept of weighted ampere-hour throughput, so as to correct the online estimation value of the health state:

[0069]

[0070] where S O H p and S O H p+1 are the online estimation values of SoH at time t p and t p+1 , A refers to the pre-factor, E a is the activation energy (J), C rate is the battery charge and discharge rate, B is the compensation factor of C rate , R is the gas constant (J / (mol·K)), T bat is the absolute temperature of the battery (K), Z is the time factor, ΔA heff is the effective ampere-hour throughput during the period from t p to t p+1 . By introducing the concept of weighted ampere-hour throughput, the actual ampere-hour throughput (ΔA hact ) caused by any charge / discharge event can be converted into the effective ampere-hour throughput (ΔA heff ) under the nominal operating conditions, and the expression is as follows:

[0071]

[0072] where Q A and Q R are the battery capacities under actual and rated operating conditions respectively, and i bat is the battery output current.

[0073] Meanwhile, a decentralized I-V droop control is introduced to achieve the basic current distribution function of the energy storage units, and an adaptive droop controller based on the state of health is designed to achieve the state of health balance:

[0074]

[0075] where R vi is the improved dynamic droop coefficient. Q ratedmax is the capacity of the ESU with the largest rated capacity, and all ESUs have the same value. SoH avgi is the average SoH calculated locally at the ESU i , SoH avgj is the average SoH calculated locally at the ESU j , ……, all average SoHs can be obtained through the dynamic consensus algorithm, and finally all the average SoHs obtained through the dynamic consensus algorithm will converge and be equal, that is: SoH avgi = SoH avgj = ……. R v0 is the initial droop coefficient. ρ and ε are the adjustment factors that control the SoH balance speed and accuracy respectively. ρ mainly affects the SoH balance speed in the initial stage, and ε mainly affects the SoH balance speed and accuracy in the final stage.

[0076] However, there is a contradiction between accurate current distribution and voltage regulation in droop control. An excessive droop coefficient will cause the DC bus voltage to deviate seriously from the given value, affecting the power quality. To solve the above problems, it is often necessary to design multiple additional secondary control loops to eliminate the influence of line impedance on current distribution and compensate for the DC bus voltage deviation caused by the droop coefficient and line impedance. Unfortunately, designing multiple secondary control loops will greatly reduce the stability margin of the system and it is difficult to ensure the stable operation of the DC microgrid under various working conditions.

[0077] Therefore, in the secondary control layer, a virtual state variable ζ containing local current, voltage, and dynamic droop coefficient information is designed. Finally, a secondary control loop regulated by an integrator can achieve two control objectives: accurate current distribution in proportion to the rated capacity and restoring the average DC bus voltage to the rated value. The expression of the secondary control loop regulated by an integrator is as follows:

[0078]

[0079] where δv i is the DC bus voltage compensation term determined by the secondary control layer, κ is the integral gain, u nom is the rated DC bus voltage, ζ avgi and ψ iare the average virtual state variable and voltage compensation factor of the ESU i respectively, and can be designed as:

[0080]

[0081] where ζ i is the virtual state variable of the ESU i , ζ avgi is the average virtual state variable calculated locally in the ESU i . The average virtual state variable can be obtained through the dynamic consistency algorithm. u dci is the output voltage of the ESU i . ξ is a gain to prevent ψ i from being zero (0 < ξ < 1). ξ = 1 / 2 is selected. Δu max is the maximum allowable offset amplitude of the DC bus voltage. Δu max = 0.1u nom . Δu viri is the virtual voltage drop of the ESU i and is designed as:

[0082] Δu viri = R vi i dci (6)

[0083] where R vi is the designed dynamic droop coefficient, and i dci is the output current of the ESU i .

[0084] In the communication layer, in order to calculate SoH avgi and ζ avgi using fewer communication resources, a neighbor-to-neighbor sparse communication network is constructed. Each energy storage unit exchanges information only with adjacent energy storage units through Low Bandwidth Communication (LBC), and the stable convergence of the global average state variable is achieved through iterative estimation by the dynamic consistency algorithm:

[0085]

[0086] where x i = [SoH avgi , ζ avgi T , γ i = [SoH i , ζ i T , x i (t) and x j (t) are the ESU i and ESU respectively​​j The state variable, γ i (t) is the ESU i The local input variable, where a ij is the ESU i and the ESU j The communication weight between the ESU and the ESU. The ring topology is selected as the communication topology between the ESU, that is, each ESU only communicates with the adjacent ESU to reduce the communication cost and ensure the convergence of relevant variables in case of communication failures. N i is the set of nodes connected to the ESU i

[0087] Such as Figure 3 The R shown vi The characteristic curve of R varying with ΔSoH i is further discussed for the rationality of the selection rules of the adjustment factors ρ and ε.

[0088] Assume that ΔSoH i = SoH avgi - SoH i , R v0 = 3, Q ratedi = Q ratedmax . In the application of the battery energy storage system, since the capacity of the recycled battery is usually between 60% and 80%, ΔSoH i is usually not more than 0.2. According to Equation (3), the characteristic curve of R varying with ΔSoH vi can be plotted for different values of ρ and ε. i

[0089] At the initial stage, when the difference in SoH is large and ε is set to a fixed value, the smaller ρ is, the larger R vi of the ESU with a lower SoH is, then i bat is smaller and the decrease in SoH is slower;

[0090] The initial stage is the stage when the difference in SoH is large, for example, the difference in SoH is more than 1%.

[0091] On the contrary, the smaller R vi of the ESU with a higher SoH is, then i bat is larger and the decrease in SoH is faster. Based on this rule, SoH will gradually equalize during the charge and discharge process;

[0092] At the final stage, when the difference in SoH is small and ρ is set to a fixed value, the smaller ε is, the more obvious the R vi difference is and the faster the SoH equalization speed is.

[0093] The final stage is the stage when the difference in SoH is small, for example, the difference in SoH is less than 1%. ​​

[0094] The above analysis results show that ρ mainly affects the SoH equalization speed in the initial stage, and ε mainly affects the SoH equalization speed and accuracy in the final stage. By introducing these two adjustment factors, the SoH equalization speed and accuracy can be greatly improved, especially when the SoH difference is small, a faster equalization speed and higher equalization accuracy can be obtained.

[0095] In summary, the following rules for selecting ρ and ε are given.

[0096] 1) When ΔSoH in the initial stage i is large, in order to avoid the simultaneous influence of ρ and ε on SoH convergence, it is necessary to set ρ|ΔSoH i | >> ε. Therefore, ρ needs to be at least one order of magnitude larger than ε, that is, ρ > 10ε. At this time, the influence of ε can be ignored, and the SoH equalization speed is mainly determined by ρ, and K i ≈ 1 / ρ|ΔSoH i |.

[0097] 2) When ΔSoH in the final stage i is close to 0, ρ|ΔSoH i | ≈ 0. In order to obtain a faster equalization speed and higher equalization accuracy when the SoH difference is small, it is necessary to have ε << 1. At this time, the SoH equalization speed and accuracy are mainly determined by ε, and K i ≈ 1 / ε.

[0098] In the main control layer, voltage-current double closed-loop PI control is adopted to control the energy storage unit. The droop coefficient is adaptively adjusted according to the dynamic change of the online estimated value of SoH in each ESU to achieve SoH equalization, and two equalization adjustment factors are introduced to improve the SoH equalization speed and accuracy; in the secondary control layer, a secondary control loop based on only one integrator adjustment is designed to eliminate the influence of line impedance on SoH equalization and achieve the recovery of the average DC bus voltage; in the communication layer, each ESU exchanges information with adjacent nodes only through adjacent communication, and the dynamic consensus algorithm is used to achieve the stable convergence of the global average state variables. The specific process of voltage-current double closed-loop PI control is as follows:

[0099] (1) Subtract the virtual voltage drop Δu nom of the ESU i from the rated DC bus voltage u viri , and then add the result to the DC bus voltage compensation term δv i to obtain the output voltage reference value u i of the ESU refi ;

[0100] (2) Subtract the actual output voltage u refi from the output voltage reference value u dciTake the difference, and obtain the reference value \(i\) of the inductor current of the lithium battery through the PI regulation of the voltage outer loop refi ;

[0101] (3) Take the difference between the reference value \(i\) of the inductor current of the lithium battery refi and the actual inductor current value \(i\) of the lithium battery bati Take the difference, and obtain two complementary Pulse Width Modulation (PWM) waves through the PI regulation of the current inner loop;

[0102] (4) Control two Insulate Gate Bipolar Transistors (IGBTs) in the bidirectional DC / DC converter respectively with the two complementary PWM waves. Due to the fast dynamic response characteristics of the voltage-current double closed loop, the actual output voltage \(u\) dci can quickly track the reference value \(u\) of the output voltage refi , so as to achieve the purpose of controlling the power output of multiple parallel energy storage units.

[0103] To further verify the effectiveness of the proposed method, a DC microgrid simulation model based on Matlab / Simulink was built. This patent studied two different experimental cases to verify the effectiveness of the proposed method under different working conditions. Since the SoH equalization is a long-time scale process, the simulation mainly focuses on the transient and steady-state performance of the proposed control method on a short-time scale. Therefore, the initial SoH difference and the ESU capacity were reduced to meet the simulation purpose. The simulation parameters are shown in Table 1.

[0104]

[0105]

[0106] In Case 1, the proposed method was compared with the traditional SoC equalization method under load power fluctuations to prove the limitations of the traditional SoC equalization method and the effectiveness of the proposed SoH equalization method under normal working conditions. The initial SoC of each ESU was set to 80%, 70%, and 60% respectively, and the initial SoH was set to 80%, 79%, and 78% respectively. The rated capacity ratio of each ESU was \(Q\) rated1 : \(Q\) rated2 : \(Q\) rated3 = 1:1:1, all 2 Ah. The experimental process was divided into the following four stages:

[0107] Stage 1 (0 - 50 s): The power generation power of the photovoltaic array and the load power were 20 kW and 23 kW respectively;

[0108] Stage 2 (50 - 100 s): At \(t = 50\) s, the load power decreased from 23 kW to 17 kW;

[0109] Stage 3 (100 - 150 s): At t = 100 s, the load power increases from 17 kW to 23 kW;

[0110] Stage 4 (150 - 200 s): At t = 150 s, the load power decreases from 23 kW to 17 kW.

[0111] Figure 4 (a) is the SoC change curve of the traditional SoC equalization method in Case 1, (b) is the SoH change curve of the traditional SoC equalization method in Case 1, (c) is the SoH change curve of the present invention in Case 1, (d) is the ESU output current change curve of the present invention in Case 1, and (e) is the average DC bus voltage change curve of the present invention in Case 1;

[0112] From Figure 4 (a)(b), it can be seen that when the traditional SoC equalization method is adopted, the ESU with a higher SoC releases more current during the discharge process and absorbs less current during the charging process. On the contrary, the ESU with a lower SoC releases less current during the discharge process and absorbs more current during the charging process. Therefore, the SoC of each ESU gradually converges, and SoC equalization is achieved at about t = 36 s. Whether it is the discharge process or the charging process, the greater the current, the greater the change rate of SoH, that is, the greater the aging rate of the ESU. With the SoC equalization, the aging rates of each ESU will tend to be the same. The traditional SoC equalization method will not be able to achieve SoH equalization between different ESUs, but will instead cause the ESU with the lowest SoH to drop to the retirement level first. From Figure 4 (c)(d)(e), it can be seen that when the proposed SoH equalization method is adopted, the ESU with a higher SoH releases / absorbs more current during the discharge / charging process and has a greater aging rate. On the contrary, the ESU with a lower SoH releases / absorbs less current and has a smaller aging rate. Therefore, the SoH of each ESU gradually converges. In addition, when the load power fluctuates, the average DC bus voltage can quickly recover and stabilize at the rated value after a small fluctuation. Finally, SoH equalization and precise current distribution are achieved at about t = 148 s.

[0113] The simulation results of Case 1 show that the traditional SoC equalization method has the limitation of being unable to achieve SoH equalization between different ESUs. The proposed method can achieve SoH equalization and precise current distribution during the charge and discharge processes, quickly eliminate the influence of line impedance on the current distribution effect, and ensure that the average DC bus voltage quickly recovers and stabilizes at the rated value when the load power fluctuates.

[0114] Case 2 mainly studies the control effect of the proposed method under the conditions of communication status and equipment status changes of ESU with inconsistent rated capacities. The initial SoH of each ESU is set to 80%, 79%, and 78% respectively, and the rated capacities are 1 Ah, 2 Ah, and 3 Ah respectively, with a ratio of Q rated1 :Q rated2 :Q rated3 =1:2:3. The experimental process can be divided into the following seven stages:

[0115] Stage 1 (0 - 50 s): The power generation power of the photovoltaic array and the load power are 20 kW and 23 kW respectively;

[0116] Stage 2 (50 - 80 s): At t = 50 s, the load power drops from 23 kW to 17 kW;

[0117] Stage 3 (80 - 100 s): At t = 80 s, the communication between ESU1 and ESU2 fails;

[0118] Stage 4 (100 - 120 s): At t = 100 s, the load power rises from 17 kW to 23 kW;

[0119] Stage 5 (120 - 130 s): At t = 120 s, the communication between ESU1 and ESU2 is restored;

[0120] Stage 6 (130 - 140 s): At t = 130 s, ESU3 quits running due to a fault;

[0121] Stage 6 (140 - 150 s): At t = 140 s, ESU3 runs again;

[0122] Stage 7 (150 - 200 s): At t = 150 s, the load power drops from 23 kW to 17 kW.

[0123] Figure 5 (a) is the SoH change curve of the present invention in Case 2, (d) is the ESU output current change curve of the present invention in Case 2, and (e) is the average DC bus voltage change curve of the present invention in Case 2;

[0124] After the communication between ESU1 and ESU2 fails, since there is still a minimum directed spanning tree in the communication network, the dynamic consistency algorithm is still effective and does not affect the control effect of the proposed method. Therefore, the SoH of each ESU gradually converges, and the first SoH balance is achieved at about t = 110 s, and the output current is distributed in the ratio of 1:2:3. After the communication between ESU1 and ESU2 is restored, the system continues to maintain a stable operation state. At t = 130 s, ESU3 quits running due to a fault, and its output current i dc3= 0 A, the aging rate of ESU3 becomes 0. Since the communication network composed of the remaining ESUs still has a minimum directed spanning tree, the remaining two ESUs can still maintain the SoH balance state. The output current is redistributed in a 1:2 ratio, the aging rate increases, and the average DC bus voltage only drops slightly. At t = 140 s, ESU3 runs again. After a small fluctuation, the average DC bus voltage quickly returns to the rated value. The SoH of each ESU converges again, and the second SoH balance is achieved at about t = 180 s. The output current is redistributed in a 1:2:3 ratio.

[0125] The simulation results of Case 2 show that the proposed method can achieve the control objectives of SoH balance, accurate current distribution, and average DC bus voltage recovery under the conditions of inconsistent ESU rated capacities, communication state changes, and equipment state changes. This verifies the excellent performance of the proposed method in being plug-and-play and resisting the risks of communication and equipment failures.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for controlling the health status of multiple energy storage units in parallel, characterized in that: The following steps are involved: Step 1: Build a DC microgrid structure including an energy storage unit, a power generation unit, a load unit and a DC bus; Step 2: Using an online battery health status estimation method, estimate the battery health status of the battery module in each energy storage unit at different rated capacities; Step 3: A hierarchical control method is used to balance the health status of the battery modules in all energy storage units based on the estimated battery health status of the battery modules in each energy storage unit at different rated capacities. The current is accurately distributed in proportion to the rated capacity and the average DC bus voltage is restored to the rated value.

2. A method for controlling the health status of multiple energy storage units in parallel according to claim 1, characterized in that: The hierarchical control comprises a primary control layer, a secondary control layer and a communication layer; The main control layer is used to control the energy storage unit by adopting voltage and current double closed-loop PI control, and introduces the speed adjustment factor and precision adjustment factor for controlling the balance of the battery health state into the droop coefficient based on the adaptive droop control method to improve the speed and precision of SoH balance. The secondary control layer is used to adopt a simplified secondary control method, based on a secondary control loop regulated by an integrator, to achieve accurate current distribution of the batteries in the energy storage unit in proportion to the rated capacity and restore the average DC bus voltage to the rated value; The communication layer is used to exchange information with adjacent nodes based on each energy storage unit relying only on inter-neighbor communication, and to achieve stable convergence of global average state variables using a dynamic consistency algorithm.

3. According to claim 1, a method for controlling the health status of multiple energy storage units in parallel is characterized in that: The battery health status online estimation method is expressed as follows: Of which: SoH p and SoH p+1 is the time t p and t p+1 SoH online estimation value when A refers to the pre-factor, E a is the activation energy, C rate is the battery charge and discharge rate, B is C rate The compensation factor is R, the gas constant, and T bat is the absolute temperature of the battery, Z is the time factor, ΔA heff Yes p to p+1 The effective ampere-hour throughput during the period, Effective ampere-hour throughput ΔA under nominal operating conditions heff By introducing a weighted ampere-hour throughput, the actual ampere-hour throughput ΔA caused by any charge / discharge event is hact Converted to effective ampere-hour throughput ΔA heff , the expression is as follows: Where Q A and Q R are the battery capacities under actual and rated operating conditions, respectively, i bat Output current for the battery.

4. According to claim 2, a method for controlling the health status of multiple energy storage units in parallel, characterized in that: The expression of the droop coefficient is as follows: Where R vi is the improved dynamic droop coefficient, Q ratedmax is the capacity of the energy storage unit (ESU) with the largest rated capacity. All ESUs have the same value. SoH avgi For ESU i The average SoH calculated locally, R v0 is the initial droop coefficient, ρ and ε are the adjustment factors for controlling the SoH balancing speed and accuracy, respectively. ρ mainly affects the SoH balancing speed in the initial stage, and ε mainly affects the SoH balancing speed and accuracy in the final stage.

5. A method for controlling the health status of multiple energy storage units in parallel according to claim 1, characterized in that: The average state of health SoH avgi , which can be obtained through a dynamic consistency algorithm.

6. A method for controlling the health status of multiple energy storage units in parallel according to claim 1, characterized in that: The expression of the secondary control loop based on an integrator regulation is as follows: Where: δv i is the DC bus voltage compensation term determined by the secondary control layer, κ is the integral gain, u nom is the rated DC bus voltage, avgi and ψ i ESU i The average virtual state variable and voltage compensation factor are designed as: where ζ i for ESU i The virtual state variable, ζ avgi For ESU i The locally calculated average virtual state variable, u dci for ESU i output voltage, ξ is a i For a gain of zero (0<ξ<1), select ξ=1 / 2, Δu max is the maximum allowable deviation of the DC bus voltage, Δu max =0.1u nom , Δu viri for ESU i The virtual voltage drop is designed to be: Δu viri =R vi and dci (6) Where R vi is the dynamic droop coefficient designed in step 3 of claim 1, i dci for ESU i of output current.

7. A method for controlling the health status of multiple energy storage units in parallel according to claim 2, characterized in that: The average virtual state variable ζ avgi , which can be obtained through a dynamic consistency algorithm.

8. A method for controlling the health status of multiple energy storage units in parallel according to claim 5 or 7, characterized in that: The expression of the dynamic consistency algorithm is as follows: where x i =[SoH avgi ,ζ avgi ] T , γ i =[SoH i ,ζ i ] T , x i (t) and x j (t) are ESU i and ESU j The state variable, γ i (t) is ESU i Local input variables, where a ij It's ESU i and ESU j The communication weight between ESUs is chosen as the ring topology, that is, each ESU communicates only with adjacent ESUs to reduce the communication cost and ensure the convergence of related variables in the event of a communication failure. i Is with ESU i A collection of connected nodes.

9. A method for controlling the health status of multiple energy storage units in parallel according to claim 2, characterized in that: The main control layer uses voltage and current double closed-loop PI control to control the energy storage unit. The specific process is as follows: The rated DC bus voltage u nom with ESU i Virtual voltage drop Δu viri The result is then subtracted from the DC bus voltage compensation term δv i Add them together to get ESU i Output voltage reference value u refi ; The output voltage reference value u refi The actual output voltage u dci The voltage outer loop PI regulation is used to obtain the lithium battery inductor current reference value i refi ; The battery inductor current reference value i refi Compared with the actual lithium battery inductor current value i bati The difference is made, and two complementary pulse width modulation (PWM) waves are obtained through the current inner loop PI regulation; The two complementary PWM waves are used to control the two insulated gate bipolar transistors in the bidirectional DC / DC converter. Since the voltage and current double closed loop has fast dynamic response characteristics, the actual output voltage u dci Capable of quickly tracking the output voltage reference value u refi , thereby achieving the purpose of controlling the power output of multiple energy storage units in parallel.

10. A method for controlling the health status of multiple energy storage units in parallel according to claim 4, characterized in that: The rules for selecting ρ and ε are as follows: When the initial stage: ΔSoH i When it is large, in order to avoid both ρ and ε affecting the SoH convergence at the same time, it is necessary to set ρ|ΔSoH i |>>ε; therefore, ρ needs to be at least one order of magnitude larger than ε, that is, ρ>10ε. At this time, ignoring the influence of ε, the SoH equilibrium speed is mainly determined by ρ, K i ≈1 / ρ|ΔSoH i |; When the final stage: ΔSoH i When it is close to 0, ρ|ΔSoH i |≈0, in order to obtain faster equalization speed and higher equalization accuracy when the SoH difference is small, ε<<1 is required. At this time, the SoH equalization speed and accuracy are mainly determined by ε, K i ≈1 / ε.

Images