Energy storage steady-state PCS power allocation algorithm based on SOC dynamic sequencing

A power allocation algorithm for energy storage PCS based on SOC sequencing is proposed, aiming at the problem that the energy management system (EMS) can allocate the power of the energy storage unit power converter (PCS) in the existing centralized electrochemical energy storage station steady-state power control, and cannot take into account the equalization of the state of charge (SOC) of the energy storage unit and the power control accuracy of the whole station. Based on the analysis of the characteristics of PCS power allocation algorithms commonly used in engineering, this paper proposes an algorithm to determine the PCS control priority based on the SOC ranking results, and then realize PCS power allocation. The simulation results show that compared with other steady-state PCS power allocation algorithms, the algorithm proposed in this paper can enable the whole station to achieve a better balance of the SOC of the energy storage unit under the premise of satisfying the power control accuracy, so that the energy storage station can provide better steady-state power support for the power grid.


Introduction
With the promotion and implementation of the energy strategic goal of "carbon neutrality and carbon peaking", the energy structure of the power grid has undergone corresponding changes.The proportion of traditional energy is getting lower and lower, and the proportion of new energy is getting higher and higher.However, such energy structure changes have brought new challenges to the safe and stable operation of the power grid.The power grid requires many flexible resources for allocation.The electrochemical energy storage system is widely used because of its excellent regulation performance, which is very suitable for the needs of power grids.Electrochemical energy storage uses electrochemical cells as energy carriers and power electronic devices for power conversion, which has the characteristics of flexible adjustment and fast response and has a wide range of application scenarios [1][2][3][4] .On the power supply side, the cooperation between traditional energy and electrochemical energy storage system can improve the overall regulation performance of the power grid.The combination of electrochemical energy storage with wind power and photovoltaic new energy can effectively improve the problem of distributed new energy generation consumption.On the grid side, the power grid uses energy storage to shave peaks and fill valleys to ease the pressure on grid operation.It can also use the frequency regulation and voltage regulation functions of energy storage to provide dynamic and stable support for the power grid.In addition, the use of energy storage accident backup and black start functions can effectively ensure the stable operation of the power grid.On the user side, in addition to being used as an accident backup, the peak-valley electricity price mechanism can be used to make profits.
The large-scale development of electrochemical energy storage puts forward higher and higher requirements for the performance of system control equipment, and the performance of the control system has an important impact on the stable operation of the energy storage system.The energy storage station control system is mainly composed of energy management system (EMS), power conversion system (PCS), and battery management system (BMS), of which the steady-state control is mainly realized by EMS, the transient control is mainly realized by PCS, and the system control principle is shown in Figure 1.Transient control involves the PCS controller's high and low penetration, primary frequency regulation, grid adaptability, inertia, and damping control, which has been extensively studied [5][6][7][8] .
Steady-state control is mainly aimed at power control in the stable state of voltage and frequency, coping with power commands in power grid dispatching and energy storage stations, and is related to the "calculation and allocation of PCS power" function of EMS.After the EMS accepts the active power instruction, it decomposes the total active instruction to each energy storage unit PCS for execution through calculation.The function of "calculating and distributing PCS power" is an intermediate link between the dispatching end and the PCS execution end, which affects the steadystate power capability of the whole station and the long-term operation life of the battery, and its importance can be understood as follows.
From the perspective of grid dispatching, it is expected that the energy storage station can run longer under the power control accuracy, which is the most beneficial to the support of the power grid.From the perspective of the energy storage station, it is expected that the energy storage unit in the energy storage station can respond to power commands to the same extent, ensure the consistency and balance of the battery, and improve the life of the battery of the energy storage unit.These two expectations are practically unified.A feature of the energy storage station is that the operation is affected by the "energy barrel effect" of the energy storage unit.When the energy storage unit is different, the unit with high energy will affect the power of the charging process, and the unit with low energy will affect the discharge process, that is, the balance of energy affects the process of power charging and discharging.The high consistency of the energy storage unit means that the "energy barrel effect" is weak, and the capacity of the energy storage station can be fully utilized, which is equivalent to the strongest support for the power grid.Therefore, the problem that needs attention to steady-state control can be considered as being able to balance the battery under the condition of ensuring the accuracy of power control.
For the steady-state power control of energy storage units, literature has been studied.A power allocation method is proposed for the hybrid battery energy storage system based on empirical mode decomposition, which only considers the problem of SOC exceeding the limit when used, does not consider the equilibrium, and is less applied in practical engineering [9][10][11] .Liu et al. [12] use distributed control algorithms to manage and control multiple energy storage and use the rated power and SOC values of multiple energy storage units to jointly determine the output power.Li and Ma [13] propose a BESS energy management method considering the health state of the battery pack, which integrates the battery health state into the battery pack charge and discharge priority sorting rule, and uses the mathematical programming algorithm to solve the distribution of power among the battery packs.Zhao et al. [14] use the exponential function relationship between charge and discharge power and SOC to design the allocation strategy, dynamically adjust the charge and discharge ratio of each battery energy storage system, and dynamically adjust the SOC by the variable interval window method.The above several PCS power allocation algorithms consider the control of SOC equilibrium, some establish the mathematical model of SOC, and some use complex mathematical methods to calculate the allocated power.In general, the equilibrium control of SOC is realized, but there is a problem that cannot be ignored, that is, the algorithm focuses on SOC equalization, and fails to consider the power accuracy control of the whole station, because the power control is controlled by inequality constraints.When the algorithm is implemented, the PCS power is first calculated by using the equation constraint, and then the PCS power is re-corrected by the over-limit judgment.If the power is corrected, it is different from the result of the original equation constraint calculation, and the power control accuracy cannot be guaranteed.In addition, the algorithms provided by the above literature are more complex and have not been used on a large scale in field engineering.
Based on the analysis of several algorithms commonly used in field engineering, this paper proposes an energy storage PCS power allocation algorithm based on SOC sequencing combined with the steady-state control principle of power distribution, and proves the effectiveness of the proposed algorithm through simulation analysis. [15]he PCS power allocation algorithm in field engineering requires a simple and simultaneous power control accuracy, and the following is a brief analysis of the commonly used algorithms in the field and the proposed new algorithms.

Algorithm 1: Distribute power based on the proportion of the maximum available power value
The algorithm is proportionally distributed according to the maximum available power of each currently functioning PCS, as shown in Equation (1).In the algorithm, the ratio of the maximum available power per PCS to the sum of the maximum available power of all PCSs is used as a factor, and the active setpoint is multiplied by the active share borne by each PCS.
where P i is the power value of the energy storage converter; P i,max is the maximum rechargeable (discharged) power value of the energy storage converter i, and P Set sets the total power for energy storage.
The advantage of the algorithm is that EMS uses the maximum available power sent by each PCS for control Its disadvantage is the battery consistency of the energy storage system battery and the balance factor of SOC.In the algorithm operation, it is easy to lead to the long-term charging and discharge of the energy storage system after the energy storage unit battery cell difference is large, and the "energy barrel effect" is further prominent, which is not conducive to the steady-state power control of the whole station.

Algorithm 2: Allocate power based on the ratio of SOC to power weight
The algorithm comprehensively considers the maximum chargeable and discharging power of the PCS and the battery SOC of the energy storage system to allocate the power target value of each PCS, and the algorithm is shown in Equation (2).
where f pi is the ratio of the maximum chargeable and dischargeable power value of group i to the maximum chargeable and discharging power of all groups, f si is the SOC percentage of the stack of group i, and wp and ws are the weighted values of the two factors.
The advantage of this algorithm is that the power capability of PCS and battery SOC value are comprehensively considered in the form of weighting, so that the power issued during the charging and discharging process is constrained by two aspects at the same time, and will not only consider the power factor as in Algorithm 1.By combining the SOC factors of the energy storage unit battery, the consistency level of the energy storage unit battery can be improved.The disadvantage of this algorithm is that the weighted value used has no relevant reference basis, the effect varies greatly under different weighted values, and the weighted value calculation is difficult.

Algorithm 3: Energy storage PCS power allocation based on SOC sequencing
The basic principle is to determine the PCS control priority with the SOC sorting result, and then realize the PCS power allocation, and the algorithm principle is as follows.N is the number of energy storage units, power_avail is the PCS available power of each unit, and battery_SOC is the state-ofcharge SOC of the battery.
In the implementation process, the PCS call priority of the energy storage system is first set based on the battery SOC ranking result of the energy storage system, that is, the SOC of the battery of the energy storage unit is sorted from largest to smallest, and the higher the order, the higher the priority of the corresponding PCS call of the energy storage unit.Then, according to the PCS call priority, the power Pset issued by EMS is first borne by the PCS with the highest call priority.If the Pset power is greater than power_avail the current maximum power available to the PCS, then the remaining power (Pset-power_avail) is then borne by the PCS with the second highest priority, and so on.The power Pset is gradually decomposed by each PCS until all the power is allocated.
The use of SOC sequencing in the implementation process is to make SOC more balanced.In the discharge process, the energy storage unit with higher SOC should be discharged first.In the charging process, the energy storage unit with low SOC should be charged preferentially to reduce the impact of the "energy barrel effect".Since the PCS power is not allocated proportionally, but according to the SOC priority order, the allocation process considers the power-bearing capacity of each energy storage unit, and there will be no need to correct the final output power in Algorithms 1 and 2, to ensure that all power can be fully allocated.Therefore, the algorithm can ensure the power control accuracy of the whole station under the premise of SOC equalization control.
One difference between this algorithm and conventional algorithms is that the PCS power sent to each energy storage unit needs to be dynamically adjusted several times during operation.

The power allocation algorithm based on SOC sequencing is implemented 2.4.1. Basic steps of the algorithm
(1) It is initialized after the information of each energy storage unit of the energy storage station is entered, such as battery state of charge SOC, battery state of charge limit SOC_max, SOC_min, SOC original sorting order orig_sequence, SOC sorting order sort_sequence, PCS available power power_avail, PCS final setting power power_set, and the whole station setting active power P set .
(2) It sorts based on orig_sequence from largest to smallest to obtain sort_sequence.
(3) It compares the power_avail of P set with the current energy storage unit i.If P set is greater than power_avail, the power_set of the current energy storage unit is set to power_avail, and P set is corrected to (Pset-power_set), and then the next energy storage unit is calculated.If P set t is less than power_avail, the power_set of the current energy storage unit is set to P set , and P set is corrected to zero.
(4) It stops the calculation of this round after P set t is zero.If there is no stop instruction, it repeats Steps (1) ~ (4) according to a certain time step.

Algorithm flowchart
The algorithm execution flow chart is shown in Figure 2.

Simulation verification
According to the above three algorithms, an energy storage station operating environment is constructed for simulation verification and comparison.The energy storage station consists of 5 energy storage units, each of which has a PCS-rated power of 30 kW, the value range of state-ofcharge SOC is 5%~95%, and the active power Pset is set to 50 kW.The operating conditions are that each energy storage unit of the energy storage station runs in a discharged state, the available power of each energy storage unit does not change during operation, and when the SOC of the energy storage unit touches the SOC value boundary, the energy storage unit stops running.The initial parameters of each energy storage unit of the energy storage station are shown in Table 1.
The set initial parameter table shows that the state of charge and available power of the five energy storage units of the energy storage station are not equal, indicating that the consistency of the batteries of the five energy storage units is not good, the PCS power capacity is also different, and the initial situation is unfavorable to the balanced control of the energy storage station.

Simulation analysis of Algorithm 1
The operation simulation of the energy storage station is carried out according to the algorithm of "distributed power based on the proportion of the maximum available power value", and the simulation curve is shown in Figure 3.The figure shows that the SOC of each energy storage unit in the initial state is quite different, and the consistency of SOC is not improved during operation, and the SOC difference of each energy storage unit reaches the maximum when it is iterated 45 times, and the overall SOC equilibrium control ability is poor.The algorithm is good in the accuracy of active power control, and the output power can track the set power Pset for 52 iterations, and then the output power drops due to the stop operation of some energy storage units.Although the algorithm can control the discharge at different powers, eventually making the SOC of all energy storage units tend to the lower limit, such equilibrium is meaningless because the SOC equilibrium cannot be guaranteed during the operation of a given power.
Figure 3. SOC equilibrium trend chart

Simulation analysis of Algorithm 2
According to the algorithm of "allocated power based on SOC and power weight ratio", the operation simulation of the energy storage station was carried out, and the weighted values wp and ws were set to (0.0, 1.0), (0.2, 0.8), (0.8, 0.2), and (1.0, 0.0), respectively.The simulation curve is shown in Figure 4~7.(1) w p =0.0, w s =1.0 Figure 4 shows that the SOC value of each energy storage unit tends to be the same during operation, indicating that the purpose of SOC equalization can be effectively satisfied, but the setting power accuracy cannot be guaranteed.The figure shows that as the operation continues, the actual output power value deviates more and more from the set power P set .The reason for the deviation is that the power weight w p in the equation is small, the SOC weight w s is large, and the calculated P i will deviate from the maximum available power of the PCS, and it will need to be corrected during the actual delivery process, so that the actual delivery value will deviate from P set .
This weight combination shows that the weight ratio based on SOC weights can better meet the requirements of SOC equilibrium, but cannot meet the requirements of output power accuracy.
(2) w p =0.2, w s =0.8 Figure 5 shows that compared to the weight combination simulation in (1), the SOC value equalization trend is weaker, but the power control accuracy is improved.It shows that increasing the power weight ratio can improve the power control accuracy, but will reduce the ability of SOC equalization.
(3) w p =0.8, w s =0.2 Figure 6 shows that compared with the weight combination simulation in ( 1) and ( 2), the SOC value equalization is basically out of control, the power control accuracy is further strengthened, and the actual output power deviates from the set power after about 48 iterations.
(4) w p =1.0, w s =0.0 Figure 7 shows that compared with the weight combination simulation in (1), (2), and (3), the SOC value equalization is out of control, the power control accuracy reaches the strongest, and the actual output power deviates from the set power after about 52 iterations.This weight combination shows that the weight ratio based on the maximum available power weight of PCS can better meet the requirements of output power accuracy, but cannot meet the requirements of SOC equalization.It is worth noting that the simulation results of the algorithm under this weight are consistent with the simulation results of Algorithm 1, indicating that Algorithm 2 can be converted into Algorithm 1 under certain circumstances.
From the above simulation of the "distributed power based on SOC and power weight ratio" algorithm, the algorithm based on weight ratio cannot consider the requirements of SOC equilibrium and power control accuracy, there is no corresponding basis for how to set the weight ratio, and the application of this algorithm is not conducive to the steady-state control of energy storage stations.

Simulation analysis of Algorithm 3
The operation simulation of the energy storage station is carried out according to the algorithm of "distributed power based on SOC sequencing", and the simulation curve is shown in Figure 8, and the local curve is shown in Figure 9. Figure 8 shows that the SOC of each energy storage unit in the initial state is quite different, and it is significantly improved after about 8 iterations, and the SOC deviation difference does not exceed 5%, indicating that the SOC equilibrium control goal has been effectively achieved.The energy storage unit batt4 in the whole process due to the low available power power_avail, the whole process output is not enough, the SOC change is always very slow, but it does not affect the implementation of the control strategy.Before the iteration 62 times, the output power can track the set power Pset, but after that, due to the stop of some energy storage units, the power drops, compared with the previous two control algorithms, the algorithm fully meets the requirements of SOC equalization control and high-precision control of the whole station power.IOP Publishing doi:10.1088/1742-6596/2584/1/01204710 energy storage unit is called first, and then controlled again after a step cycle.The algorithm process effectively controls the equilibrium trend of the SOC of energy storage units and ensures the SOC consistency of multiple energy storage units.
From the above simulation of the "Distributed Power Based on SOC Sequencing" algorithm, the algorithm effectively achieves the goals of SOC equilibrium control and active accurate control.It is worth noting that the algorithm requires the PCS to repeatedly adjust the output power to meet the needs of SOC equalization, which increases the complexity of control compared to the previous two algorithms, but such a control burden is acceptable because the PCS itself is in a hot standby state and can accept changing power instructions at any time.

Conclusion
Based on the steady-state power control characteristics of energy storage stations, this paper puts forward the goals of SOC equilibrium control and power control accuracy that should be considered by EMS.Aiming at the problems of commonly used PCS power allocation algorithms in the field, an energy storage PCS power allocation algorithm based on SOC sequencing is proposed.The results of the study are as follows: 1) The "Calculate PCS Power Allocation" function of EMS needs to consider the two principles of ensuring that the energy storage station provides maximum active power support to the grid and ensuring the consistency and optimization of the battery of the energy storage station.These two principles can eventually be transformed into SOC equilibrium control and active precision control.
2) The algorithm of "distributed power based on the proportion of the maximum available power value" and the algorithm of "distributed power based on the ratio of SOC and power weight" commonly used in field engineering cannot consider the needs of SOC equilibrium and the demand for active control accuracy, which is not conducive to the long-term operation of energy storage stations.
3) The "Energy Storage PCS Power Based on SOC Sequencing" algorithm enables the energy storage unit to be balanced to the greatest extent under the premise of satisfying the power accuracy through the SOC dynamic ranking algorithm, which has the characteristics of simplicity and efficiency compared with other algorithms and is suitable for on-site engineering implementation.

Figure 1 .
Figure 1.Diagram of control principle system Transient control is mainly aimed at the emergency response of the energy storage station in the case of sudden changes in voltage and frequency.Transient mutations happen fast and change frequently, which requires transient control to respond in time.The control task is generally undertaken by the PCS controller with high real-time performance, instead of EMS for unified control.Transient control involves the PCS controller's high and low penetration, primary frequency regulation, grid adaptability, inertia, and damping control, which has been extensively studied[5][6][7][8] .Steady-state control is mainly aimed at power control in the stable state of voltage and frequency, coping with power commands in power grid dispatching and energy storage stations, and is related to the "calculation and allocation of PCS power" function of EMS.After the EMS accepts the active power instruction, it decomposes the total active instruction to each energy storage unit PCS for execution through calculation.The function of "calculating and distributing PCS power" is an intermediate link between the dispatching end and the PCS execution end, which affects the steadystate power capability of the whole station and the long-term operation life of the battery, and its importance can be understood as follows.From the perspective of grid dispatching, it is expected that the energy storage station can run longer under the power control accuracy, which is the most beneficial to the support of the power grid.From the perspective of the energy storage station, it is expected that the energy storage unit in the energy storage station can respond to power commands to the same extent, ensure the consistency and balance of the battery, and improve the life of the battery of the energy storage unit.These two expectations are practically unified.A feature of the energy storage station is that the operation is affected by the "energy barrel effect" of the energy storage unit.When the energy storage unit is different, the unit with high energy will affect the power of the charging process, and the unit with low energy will affect the discharge process, that is, the balance of energy affects the process of power

Figure 7 .
Figure 7. w p =1.0, w s =0.0, SOC equilibrium trend chart The simulation results under different weighted values are analyzed below.(1)w p =0.0, w s =1.0 Figure4shows that the SOC value of each energy storage unit tends to be the same during operation, indicating that the purpose of SOC equalization can be effectively satisfied, but the setting power accuracy cannot be guaranteed.The figure shows that as the operation continues, the actual output power value deviates more and more from the set power P set .The reason for the deviation is that the power weight w p in the equation is small, the SOC weight w s is large, and the calculated P i will deviate from the maximum available power of the PCS, and it will need to be corrected during the actual delivery process, so that the actual delivery value will deviate from P set .This weight combination shows that the weight ratio based on SOC weights can better meet the requirements of SOC equilibrium, but cannot meet the requirements of output power accuracy.(2)w p =0.2, w s =0.8Figure5shows that compared to the weight combination simulation in (1), the SOC value equalization trend is weaker, but the power control accuracy is improved.It shows that increasing the power weight ratio can improve the power control accuracy, but will reduce the ability of SOC equalization.(3)w p =0.8, w s =0.2 Figure6shows that compared with the weight combination simulation in (1) and (2), the SOC value equalization is basically out of control, the power control accuracy is further strengthened, and the actual output power deviates from the set power after about 48 iterations.(4)w p =1.0, w s =0.0 Figure7shows that compared with the weight combination simulation in (1), (2), and (3), the SOC value equalization is out of control, the power control accuracy reaches the strongest, and the actual output power deviates from the set power after about 52 iterations.This weight combination shows that the weight ratio based on the maximum available power weight of PCS can better meet the requirements of output power accuracy, but cannot meet the requirements of SOC equalization.It is worth noting that the simulation results of the algorithm under this weight are consistent with the simulation results of Algorithm 1, indicating that Algorithm 2 can be converted into Algorithm 1 under certain circumstances.From the above simulation of the "distributed power based on SOC and power weight ratio" algorithm, the algorithm based on weight ratio cannot consider the requirements of SOC equilibrium and power control accuracy, there is no corresponding basis for how to set the weight ratio, and the application of this algorithm is not conducive to the steady-state control of energy storage stations.

Figure 9 .
Figure 9. Local SOC equilibrium trend chart Figure 9 is a part of the SOC equilibrium trend graph corresponding to Algorithm 3. It can be seen that batt 1, batt 2, and batt 3 have reciprocal cross-changes in SOC values after 7 iterations, indicating that the PCS output is controlled by algorithm optimization during operation so that the SOC larger

Table 1 .
Initial parameters of energy storage unit