Intelligent battery charging management system and charging method for battery charging and swapping cabinet

Abstract

The invention discloses an intelligent battery charging management system for a battery charging and replacing cabinet. The intelligent battery charging management system comprises a multi-source data acquisition module 101, a spatial-temporal characteristic analysis engine 102, a demand prediction model 103, a dynamic scheduling decision maker 104, a charging parameter optimizer 105, a battery health monitoring unit 106 and a system communication module. The intelligent battery charging method for the battery charging and replacing cabinet comprises the following steps: acquiring battery state data and operation data in real time; generating a demand thermodynamic diagram and network topology features through a spatial-temporal feature analysis engine; predicting demand distribution in future 6 hours based on LSTM and random forest dual models; generating a scheduling scheme meeting the electricity price, load and SOH constraints by using mixed integer linear programming (MILP); charging parameters are dynamically adjusted according to a gradient charging strategy, and valley electricity resources are preferentially used; and monitoring the health degree of the battery and feeding back and correcting the charging plan in real time. The method effectively reduces the charging cost, prolongs the service life of equipment, improves the health degree of a battery, and guarantees the operation reliability.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery management, and particularly relates to a battery intelligent charging management system and a charging method for a charging and swapping cabinet. Background Art

[0002] The current charging and swapping cabinet used for lithium batteries of two-wheeled electric vehicles is a centralized charging facility. Generally, when the battery is put in, if there is no fault and it meets the charging requirements, it will be charged immediately, and after a long time of placement, regular supplementary charging will be carried out. The current charging system of the swapping cabinet has the following defects:

[0003] 1. The fixed charging strategy leads to too high energy costs and lacks a price period response mechanism;

[0004] 2. Continuous full-power charging accelerates equipment aging, and the energy consumption of the charger fan accounts for 15%-25%;

[0005] 3. Lack of an intelligent maintenance strategy, and frequent recharging of batteries stored for a long time leads to an increase in the capacity attenuation rate by 3-5%;

[0006] 4. Static inventory management cannot adapt to dynamic demands, resulting in resource waste or supply shortages. Summary of the Invention

[0007] The purpose of the present invention is to provide a battery intelligent charging management system and a charging method for a charging and swapping cabinet applicable to large-scale operation scenarios of lithium battery charging, which can make full use of valley electricity for charging to reduce charging costs, extend the equipment life, improve the battery health, and ensure the operation reliability in view of the defects existing in the current charging and swapping cabinet.

[0008] To achieve the above object, the present invention adopts the following technical solutions:

[0009] A battery intelligent charging management system for a charging and swapping cabinet, characterized by comprising:

[0010] (1) A multi-source data acquisition module (101), which obtains the SOC, voltage, and temperature parameters of the battery through a bus interface, and real-time obtains the operation data of the price period, user swapping demand, and equipment load;

[0011] (2) A spatio-temporal feature analysis engine (102), which performs the following operations:

[0012] a) Time dimension analysis: generating a heat map of battery demand fluctuations in each period within 24 hours;

[0013] b) Space dimension analysis: extracting a position correlation matrix based on the network topology structure of the charging cabinet; the network topology structure includes data such as cabinet capacity, covered population density, and historical demand heat;

[0014] (3) Demand forecasting model (103), adopting a dual-model architecture of an LSTM time series forecasting model and a random forest classifier, and outputting the probability distribution of the battery swapping demand within the next 6 hours;

[0015] (4) Dynamic scheduling decision maker (104), based on the mixed integer linear programming (MILP) algorithm, generating a charging scheduling instruction with electricity price, equipment load rate, and battery health (SOH) as constraint conditions;

[0016] (5) Charging parameter optimizer (105), through the objective function

[0017] min(αCenergy + βTcharge + γDhealth) dynamically calculates the charging current, where Cenergy is the energy cost, Tcharge is the charging time, and Dhealth is the battery health loss;

[0018] (6) Battery health monitoring unit (106), calculates the battery health in real time according to the formula

[0019] and feeds it back to the dynamic scheduling decision maker (104); where Cactual is the actual available capacity, Crated is the rated capacity, Rnew is the initial internal resistance of the battery, and Rinternal is the current internal resistance of the battery; the charging parameter optimizer 105 outputs a control signal to the execution system to implement functions such as charging current control and fan start / stop;

[0020] (7) System communication module, performs a handshake interaction with the battery BMS to obtain the maximum charging current / voltage parameters in real time.

[0021] Further: The dynamic scheduling decision maker (104) includes:

[0022] A sliding window calculation module, dynamically calculates the minimum inventory threshold of each charging cabinet according to the following formula:

[0023]

[0024] where μd is the historical demand mean, σd is the demand fluctuation standard deviation, Vpeak is the peak demand sequence, and α is a safety factor, α ∈ [1.2, 1.5].

[0025] Its further feature is that: The charging parameter optimizer (105) adopts a dynamic current regulation algorithm to dynamically calculate the charging current:

[0026]

[0027] where twait is the waiting time from the current moment to the next valley electricity price, β is the delay attenuation factor, η(T) is the temperature compensation factor, Tvalley is the total duration of the valley electricity price period, Cremain is the remaining charging capacity to be charged, and Τ is the delayed charging time threshold.

[0028] Preferably: The typical value of η(T) is 1 at 25 degrees and 0.925 at 40 degrees.

[0029] A battery intelligent charging method for a charging and swapping cabinet, which uses the above-mentioned battery intelligent charging management system for a charging and swapping cabinet, is characterized by including the following steps:

[0030] (1) Real-time collect battery status data and operation data;

[0031] (2) Generate a demand heat map and network topology features through a spatio-temporal feature analysis engine;

[0032] (3) Predict the demand distribution in the next 6 hours based on a dual model of LSTM and random forest;

[0033] (4) Generate a scheduling plan that meets the electricity price, load, and SOH constraints by using mixed integer linear programming (MILP);

[0034] (5) Dynamically adjust the charging parameters according to the gradient charging strategy and preferentially use valley electricity resources;

[0035] (6) Monitor the battery health and real-time feedback to correct the charging plan.

[0036] Its further feature is that the charging time decision process is as follows: A new battery is connected to the charging and swapping cabinet; Determine whether the number of batteries in the charging and swapping cabinet is less than Rmin. If it is less than Rmin, immediately enter the fast charging mode; If the number of batteries is not less than Rmin, enter the charging optimization queue, and then calculate the valley electricity period; Determine whether it can be delayed until the valley electricity period. If so, set the charging time for delayed low-speed charging; If not, enter the gradient boosting charging mode.

[0037] Furthermore: The gradient boosting charging mode is as follows: In stage 1, charge at a constant current of 10A, in stage 2, charge at a constant current of 8A, and in stage 3, charge at a constant current of 5A; If SOC < 50%, perform stage 1 charging; If 50% < SOC < 90%, then perform stage 2 charging; If SOC ≥ 90%, perform stage 3 charging; In case of an emergency demand at any stage, immediately switch to the fast charging mode to charge at the maximum current.

[0038] Preferably: In the low-speed charging mode during the valley electricity period, the emergency demand charging mode, and the gradient charging boost mode, when the battery SOC ≥ 95%, enter the constant voltage trickle charging state, otherwise maintain the original charging state. The low-speed current during the valley electricity period remains at 5 - 8A.

[0039] Preferably, the charging method further includes a battery maintenance charging step; when the battery placement time reaches trecharge, the battery is subjected to maintenance charging;

[0040]

[0041] wherein: kbat is the battery attenuation coefficient, Tenv is the environmental temperature influence factor, SOCthreshold is the recharge threshold, and SOCcurrent is the currently detected remaining battery charge.

[0042] The present invention has the following technical effects:

[0043] 1. Reduce charging costs: The utilization rate of valley electricity is increased to more than 85%;

[0044] 2. Prolong the equipment life: The operating temperature of the charger is reduced by 10 - 15 °C;

[0045] 3. Improve the battery health: The cycle life is increased by 200 - 500 times;

[0046] 4. Ensure operation reliability: The out - of - stock rate is lower than 0.5%. Description of the Drawings

[0047] Figure 1 It is a schematic diagram of the system architecture of the present invention.

[0048] Figure 2 It is a schematic diagram of the method flow of the present invention.

[0049] Figure 3 It is a schematic diagram of the dual - mode charging decision - making model flow.

[0050] Figure 4 It is a schematic diagram of the current - optimized charging process. Detailed Embodiments

[0051] As Figure 1 shown, a battery intelligent charging management system for a charging and swapping cabinet

[0052] (1) The multi - source data acquisition module 101 obtains the SOC, voltage, and temperature parameters of the battery through a bus interface, and simultaneously obtains the operation data of the electricity price period, user swapping demand, and equipment load in real time. The battery data is accessed through the CAN / RS485 bus, and the operation data is accessed through the HTTP API. A circular buffer is used to store the real - time data stream for data caching.

[0053] (2) The spatio - temporal feature analysis engine 102 performs the following operations:

[0054] a) Time - dimension analysis: Generate a heat map of the battery demand fluctuation in each period within 24 hours;

[0055] b) Spatial dimension analysis: Extract the location correlation matrix based on the charging cabinet network topology; the network topology includes data such as cabinet capacity, covered population density, and historical demand popularity.

[0056] (3) Demand prediction model 103, adopting a dual-model architecture of an LSTM time series prediction model and a random forest classifier, and outputting the probability distribution of the battery swapping demand within the next 6 hours.

[0057] (4) Dynamic scheduling decision maker 104, based on the mixed integer linear programming MILP algorithm, generates charging scheduling instructions with electricity price, equipment load rate, and battery health SOH as constraint conditions.

[0058] The dynamic scheduling decision maker 104 includes:

[0059] A sliding window calculation module that dynamically calculates the minimum inventory threshold for each charging cabinet according to the following formula:

[0060]

[0061] where μd is the historical demand mean, σd is the demand fluctuation standard deviation, Vpeak is the peak demand sequence, α is the safety factor, and α ∈ [1.2, 1.5].

[0062] (5) Charging parameter optimizer 105, through the objective function

[0063] min(αCenergy + βTcharge + γDhealth) dynamically calculates the charging current, where Cenergy is the energy cost, Tcharge is the charging time, and Dhealth is the battery health loss.

[0064] The charging parameter optimizer 105 adopts a dynamic current regulation algorithm to dynamically calculate the charging current:

[0065]

[0066] where twait is the waiting time from the current moment to the next valley electricity price, β is the delay decay factor, η(T) is the temperature compensation factor, Tvalley is the total duration of the valley electricity price period, Cremain is the remaining charging capacity to be charged, and Τ is the delayed charging time threshold. Preferably: the typical value of η(T) is 1 at 25 degrees and 0.925 at 40 degrees.

[0067] The charging parameter optimizer 105 outputs a control signal to the execution system to achieve functions such as charging current control and fan start / stop.

[0068] (6) Battery health monitoring unit 106, according to the formula

[0069] Calculate the battery health in real time and feedback it to the dynamic scheduling decision maker 104. Among them, Cactual is the actual available capacity, Crated is the rated capacity, Rnew is the initial internal resistance of the battery, and Rinternal is the current internal resistance of the battery.

[0070] (7) The system communication module conducts handshake interaction with the battery BMS and obtains the maximum charging current / voltage parameters in real time.

[0071] A battery intelligent charging method for a charging and swapping cabinet, as Figure 2 shown, adopts the above-mentioned battery intelligent charging management system for a charging and swapping cabinet, including the following steps:

[0072] (1) Collect battery status data and operation data in real time;

[0073] (2) Generate a demand heat map and network topology features through a spatio-temporal feature analysis engine;

[0074] (3) Predict the demand distribution in the next 6 hours based on the dual models of LSTM and random forest;

[0075] (4) Generate a scheduling plan that meets the electricity price, load, and SOH constraints by using mixed integer linear programming MILP;

[0076] (5) Dynamically adjust the charging parameters according to the gradient charging strategy and preferentially use valley electricity resources;

[0077] (6) Monitor the battery health and feedback and correct the charging plan in real time.

[0078] As Figure 3 shown, the charging time decision adopts a dual-mode charging decision model, and the process is as follows: A new battery is connected to the charging and swapping cabinet; judge whether the number of batteries in the charging and swapping cabinet is less than Rmin. If it is less than Rmin, immediately enter the fast charging mode; if the number of batteries is not less than Rmin, enter the charging optimization queue, and then calculate the valley electricity period; judge whether it can be delayed to the valley electricity period. If so, set the charging time for delayed low-speed charging; if not, enter the gradient boosting charging mode.

[0079] The gradient boosting charging mode is as follows: In stage 1, charge at a constant current of 10A, in stage 2, charge at a constant current of 8A, and in stage 3, charge at a constant current of 5A; if the SOC < 50%, perform stage 1 charging; if 50% < SOC < 90%, then perform stage 2 charging; if the SOC ≥ 90%, perform stage 3 charging; in case of an emergency demand at any stage, immediately switch to the fast charging mode to charge at the maximum current.

[0080] As Figure 4As shown, in the low-speed charging mode, emergency demand charging mode, and gradient charging boost mode during off-peak electricity periods, when the battery SOC ≥ 95%, it enters the constant voltage trickle charging state; otherwise, it maintains the original charging state. The low-speed current during off-peak electricity periods remains at 5 - 8A.

[0081] Preferably, the charging method further includes a battery maintenance charging step; when the battery placement time reaches trecharge, the battery is subjected to maintenance charging;

[0082]

[0083] Where: kbat is the battery attenuation coefficient, Tenv is the environmental temperature influence factor, SOCthreshold is the recharge threshold, and SOCcurrent is the currently detected remaining battery charge.

[0084] Adopting the system and method of the present invention has obvious effects, which can reduce the charging cost: the utilization rate of off-peak electricity is increased to more than 85%; extend the equipment life: the working temperature of the charger is reduced by 10 - 15°C; improve the battery health: the cycle life is increased by 200 - 500 times; ensure the operation reliability: the out-of-stock rate is lower than 0.5%.

Claims

1. A battery intelligent charging management system for a charging and swapping cabinet, characterized in that include: (1) A multi-source data acquisition module (101) acquires the SOC, voltage, and temperature parameters of the battery through a bus interface, and acquires in real time the operational data of the electricity price period, the user's battery replacement demand, and the equipment load; (2) The spatiotemporal feature analysis engine (102) performs the following operations: a) Time dimension analysis: Generate a heat map of battery demand fluctuations for each period within 24 hours; b) Spatial dimension analysis: extracting the location correlation matrix based on the charging cabinet network topology; (3) Demand forecasting model (103), which uses a dual-model architecture of LSTM time series forecasting model and random forest classifier to output the probability distribution of battery replacement demand within the next 6 hours; (4) a dynamic scheduling decision maker (104), based on a mixed integer linear programming (MILP) algorithm, generates charging scheduling instructions with electricity price, equipment load factor and battery health (SOH) as constraints; (5) Charging parameter optimizer (105), through the objective function min(αCenergy+βTcharge+γDhealth) dynamically calculates the charging current, where Cenergy is the energy cost, Tcharge is the charging time, and Dhealth is the battery health loss; (6) Battery health monitoring unit (106), according to the formula Calculate the battery health in real time and feed it back to the dynamic scheduling decision maker (104), where Cactual is the actual available capacity, Crated is the rated capacity, Rnew is the initial internal resistance of the battery, and Rinternal is the current internal resistance of the battery; (7) System communication module, which performs handshake interaction with the battery BMS to obtain the maximum charging current / voltage parameters in real time.

2. The battery intelligent charging management system for a charging and swapping cabinet according to claim 1, characterized in that: The dynamic scheduling decision maker (104) comprises: The sliding window calculation module dynamically calculates the minimum inventory threshold of each charging cabinet according to the following formula: Where μd is the historical demand mean, σd is the standard deviation of demand fluctuation, Vpeak is the peak demand sequence, and α is the safety factor, α∈[1.2, 1.5].

3. The battery intelligent charging management system for a charging and swapping cabinet according to claim 1, characterized in that: The charging parameter optimizer (105) uses a dynamic current control algorithm to dynamically calculate the charging current: where t wait is the waiting time from the current moment to the next valley electricity price, β is the delay attenuation factor, η(T) is the temperature compensation factor, Tvalley is the total duration of the valley electricity price period, Cremain is the remaining charging capacity, and Τ is the delayed charging time threshold.

4. The battery intelligent charging management system for a charging and swapping cabinet according to claim 3, characterized in that: The typical value of η(T) is 1 at 25 degrees and 0.925 at 40 degrees.

5. A battery intelligent charging method for a charging and swapping cabinet, using the battery intelligent charging management system for a charging and swapping cabinet according to any one of claims 1 to 4, characterized in that The following steps are involved: (1) Real-time collection of battery status data and operation data; (2) Generate demand heat map and network topology features through spatiotemporal feature analysis engine; (3) Predict the demand distribution in the next 6 hours based on the LSTM and random forest dual models; (4) Generate a dispatching plan that satisfies the constraints of electricity price, load and SOH using mixed integer linear programming (MILP); (5) Dynamically adjust charging parameters according to the gradient charging strategy, giving priority to valley power resources; (6) Monitor battery health and provide real-time feedback to correct charging plans.

6. The battery intelligent charging method of the charging and swapping cabinet according to claim 5, characterized in that The charging time decision process is as follows: connect the new battery to the charging and swapping cabinet; determine whether the number of batteries in the charging and swapping cabinet is less than Rmin. If it is less than Rmin, enter the fast charging mode immediately; if the number of batteries is not less than Rmin, enter the charging optimization queue and calculate the valley power period; Determine whether the charging can be delayed to the valley period. If so, set the charging time for delayed low-speed charging; if not, enter the gradient boost charging mode.

7. The intelligent battery charging method of the charging and swapping cabinet according to claim 6, characterized in that The gradient boost charging mode is: charging at a constant current of 10A in stage 1, charging at a constant current of 8A in stage 2, and charging at a constant current of 5A in stage 3; if SOC is less than 50%, charging is performed in stage 1; if 50% is less than SOC is less than 90%, charging is performed in stage 2; If SOC ≥ 90%, proceed to stage 3 charging; if there is an emergency need at any stage, immediately switch to fast charging mode and charge with maximum current.

8. The intelligent battery charging method of the charging and swapping cabinet according to claim 7, characterized in that: In the low-speed charging mode during valley power period, the emergency demand charging mode and the gradient charging boost mode, when the battery SOC ≥ 95%, it enters the constant voltage trickle charging state, otherwise it maintains the original charging state; During valley power period, the low speed current is maintained at 5-8A.

9. The intelligent battery charging method of the charging and swapping cabinet according to claim 5, characterized in that: The charging method also includes a battery maintenance charging step; when the battery is placed for a period of time of t recharge Then perform maintenance charging on the battery; Among them: kbat is the battery attenuation coefficient, Tenv is the ambient temperature influence factor, SOCthreshold is the recharge threshold, and SOCcurrent is the current remaining power detected.

Images