Active battery management method for economic optimization

Abstract

Methods for active battery management to optimize battery performance are provided. The method includes providing signal injection for charging and discharging of the battery. The signal injection includes various charge and discharge profiles, rates, and endpoints. Response signals corresponding to the signal injections are received and the utility of these signals is measured. Based on the utility of the response signal, data related to charging and discharging of the battery is modified to optimize battery performance and determine when to discharge the battery into the grid in order to return power to the grid to exchange economic benefits, such as payments or discounts from the utility company.

Description

[0001] This application is a divisional application based on the Chinese patent application filed on September 10, 2019, with application number 2019800588781 and invention name “Active Battery Management Method for Economic Optimization”. Background Art

[0002] Battery management is critical to many commercial applications, from consumer electronics to automotive electrification and grid-level energy storage. Battery management is also a key part of the value proposition of any battery system, whose utility depends on reliably and safely delivering a minimum amount of energy over extended time periods, from a few years in consumer electronics to a decade in grid installations. Summary of the invention

[0003] A first method for active battery management includes injecting a randomized controlled signal into the charging or discharging of a battery, including charging the battery from a power grid; and ensuring that the signal injection occurs within normal operating ranges and constraints. The method also includes monitoring the performance of the battery in response to the controlled signal; calculating a confidence interval about the causal relationship between the battery performance and the controlled signal; and selecting an optimal signal for charging or discharging the battery based on the calculated confidence interval, including discharging the battery into the power grid in exchange for economic benefits.

[0004] A second method for active battery management includes providing a signal injection for charging or discharging a battery, including causing the battery to charge from a power grid; and receiving a response signal corresponding to the signal injection. The method also includes measuring the utility of the response signal; accessing data related to the charging or discharging of the battery, including causing the battery to discharge into the power grid in exchange for economic benefits; and modifying the data based on the utility of the response signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] The accompanying drawings are incorporated into and constitute a part of this specification and together with the description, explain the advantages and principles of the present invention.

[0006] Figure 1 A schematic diagram showing a system for implementing an active battery management method;

[0007] Figure 2 is a flow chart of a search space method for the system;

[0008] Figure 3 is a flow chart of a signal injection method for the system;

[0009] Figure 4 is a flow chart of a continuous learning method for the system;

[0010] Figure 5 is a flow chart of a memory management method for the system;

[0011] Figures 6A-6D The search space of all possible charging curves of an embodiment is shown;

[0012] Figures 7A-7C shows that the algorithm in the embodiment recognizes the different effect sizes of old batteries relative to new batteries; and

[0013] Figures 8A-8C The voltage is shown versus the capacity of the charging curve assigned by the algorithm in the embodiment. DETAILED DESCRIPTION

[0014] Embodiments of the invention include methods for improving battery management by implementing random experiments on charging and discharging variables and inferring their causal effects on utility metrics such as energy capacity, power, decay rate, charging time, internal resistance, state of health, battery imbalance, temperature, battery swelling, electrical cost, etc. A linear combination of any of the above utility metrics can also be defined to give a figure of merit that balances competing requirements. The battery management can be used, for example, in electric or hybrid vehicles, electric bicycles, consumer electronic devices, grid storage systems, and other vehicles and devices that use batteries.

[0015] The battery management method may also include economic factors such as determining when to send power back to the grid in exchange for an economic benefit (e.g., a payment, discount, or rebate from a utility company or another entity), and potentially, what type of user needs need to be addressed by the user, frequency regulation (high frequency) versus peak shaving (low frequency). Specifically, an embodiment includes a method for optimizing the discharge and charging conditions of a battery pack in an electric vehicle or uninterruptible power system so that the battery pack can be used as a source of power back to the grid without adversely affecting the life of the battery pack. The method uses deep causal learning to maximize the net benefit from the power sold back to the utility grid, or to minimize the cost of electricity consumed by the primary operator through techniques such as peak shaving. As described herein, deep causal learning is a particularly useful method for this problem because it can be continuously adjusted over the life of the battery pack.

[0016] Figure 1A schematic diagram of a system for implementing an active battery management method for a battery pack in an electric vehicle or an uninterruptible power source (or other stationary power source) is shown. The system includes a processor 10 electrically connected to a power source from a power grid 12, a load 20, and a data storage device 22. The power grid 12 provides power for charging one or more batteries 14, 16, and 18, and the batteries provide power to the load 20. In this application, the grid can be an energy source or a load when power is supplied back to the grid. The data storage device 22 (such as an electronic memory) stores curves and parameters 24, external data 26, and results 28. The results may include, for example, time series of current and voltage of each cell or battery string, energy capacity, temperature, etc.

[0017] In use, the processor 10 uses the curves and parameters 24 and possibly external data 26 to inject signals into the grid 12 in order to assess the performance of the batteries 14, 16 and 18, such as the performance of the batteries in charging and discharging them. Performance metrics may include, for example, delivered power, energy capacity, decay rate, revenue, profitability, reliability score (i.e., a score given by the utility to market participants indicating their ability to meet demand within the capacity allocated the previous day). The processor 10 stores the responses to the signal injections as results 28, and these responses may be used to optimize the performance of the batteries. Processing for battery management may occur locally on the battery charging system using dedicated firmware, on a stand-alone PC, or based in the cloud, as well as remotely from the batteries.

[0018] The curves and parameters 24 include possible charge and discharge rates, charge and discharge curves, and curve endpoints. The curves and parameters 24 also include how often a full charge / discharge cycle is performed to achieve the current estimated state of health (defined as the maximum available capacity relative to the maximum capacity at time zero), which applications will be bid (taking into account that different applications can generate different revenues and have different effects on the life of the battery pack), and for each application, how much capacity is allocated to the "grid load" relative to the standard load (more capacity = more revenue, but also more chance of over-discharge).

[0019] The charging and discharging curves include the shape of such curves and may include the time at a specific state of charge or voltage. The charging endpoints include the percentage of charge in the battery when starting and stopping charging the battery, and the discharging endpoints include the percentage of charge in the battery when starting and stopping discharging the battery. For example, the curves and parameters may be stored in a lookup table. The external data 26 may include, for example, environmental conditions or factors such as temperature, humidity, airflow around the battery, time of day or year, time since the device or vehicle was turned on using the battery, estimated state of health (SOH) and state of charge (SOC) obtained from an existing battery model (usually provided by the battery or BMS supplier). In addition, active cooling or heating of the battery may be used as other control variables with parameters of temperature set points, rates, and temporal and spatial gradients. In the case of portable electronic devices, the external data may also include at least one of the following relative to such devices: use of applications, user settings, scheduled events or alarms, power consumption patterns, time of day, or location of the device. Similarly, in the case of electric vehicles, external data may include the time of day or next planned usage time, typical driving patterns of the vehicle, electricity cost relative to time (i.e., to avoid spike pricing, for example), predicted weather conditions, planned travel routes, or traffic conditions.

[0020] These batteries may include a single physical battery or multiple physical batteries that work together to provide power. In the case of multiple physical batteries, the batteries may have the same or different constructions or electrochemistries. The processing of the signals injected for charging and discharging the batteries attempts to optimize the charging and discharging profiles for a particular battery or battery pack. A battery pack may be considered a single battery, where the battery pack operates together, or a battery pack may be considered multiple physical batteries that are driven individually. Examples of battery types include lithium-ion batteries, reflow batteries, lead-acid batteries, etc.

[0021] Figure 2-5 Flowchart of methods for active battery management to optimize charging and discharging profiles and parameters. For example, these methods may be implemented in a software module executed by processor 10, for example.

[0022] Figure 2 Flowchart of the search space method. The search space method includes the following steps: receiving control information (including cost) 30; constructing a multidimensional space of all possible control states 32; constraining the space of potential control states 34; determining a normal / baseline sampling distribution 36; determining a maximum utility sampling distribution 38; and automating control selection within the constrained space 40.

[0023] Figure 3The signal injection method comprises the following steps: receiving a set of potential signal injections 42; calculating the spatial range and time range of the signal injections 44; coordinating the signal injections in space and time 46; implementing the signal injections 48; collecting response data 50; and associating the response data with the signal injections 52.

[0024] Signal injection is a change in the charging and discharging curves and parameters used for battery management, including determining when to deliver power back to the grid. The response to the signal injection is typically the battery performance caused by or related to the change in the curves and parameters of the signal injection. For example, the algorithm may perturb the values ​​representing the charging and discharging curves and parameters in the lookup table, and then monitor and store the corresponding battery performance response. The time range and spatial range of the signal injection involve the time and location of the response signal measured to those signal injections used to calculate causal relationships, respectively. The cost of the signal injection is typically related to how the signal injection affects the battery performance, for example, the signal injection may result in lower battery performance, and is controlled by the specified experimental range. The queue for signal injection involves the order and priority of the signal injection, and relies on blocking and randomization to ensure high internal validity at all times, even when optimizing utility. The utility of the response to the signal injection involves the effectiveness or other utility measure of the signal injection.

[0025] Figure 4 Flow chart of the continuous learning method. The continuous learning method includes the following steps: receiving a set of potential signal injections 54; receiving a current confidence state 56; calculating a learning value of the signal injection 58; receiving a cost of the signal injection 60; selecting and coordinating the signal injection 62; implementing the signal injection 64; collecting response data 66; and updating the confidence state 68.

[0026] A confidence state is a collection of different battery performance models and / or economic benefit models in response to charging and discharging. In frequency regulation applications, the discharge rate is determined by the grid load and the charge rate does not significantly affect performance because the SOC is almost always close to 50% (at this SOC level, small charges have little effect on battery life / health). The main drivers of revenue and battery life loss (the main tradeoff) are how much capacity is bid and the application given the battery health and condition. These confidence states may have accompanying uncertainty values ​​that reflect the likelihood that they are accurate given current experiments and knowledge sets that may tend to confirm or falsify these different models, and information that may further confirm or falsify the model may be included in the data, or derived from the basic characteristics of the specific model and the physical characteristics of the underlying system.

[0027] The learning value is a measure of the value that the knowledge generated as a result of the signal injection can provide to the system's subsequent decisions, such as determining that a particular charge or discharge profile is more likely to be optimal. In the sense of multi-objective optimization, this can include complex trade-offs between operating objectives (e.g., performance versus range), and where optimality can change over time. The learning value can be calculated, for example, by predicting the raw number of belief states that can be falsified based on predictions from a partially observable Markov decision process (POMDP) ​​or other statistical model, the predicted impact of signal injection on the level of uncertainty in the belief states in such a model, or an experimental power analysis based on increasing to the current sample size to calculate uncertainty reduction and confidence interval reduction.

[0028] Figure 5 Flow chart of the memory management method. The memory management method includes the following steps: receiving a set of historical clusters 70; receiving a set of historical signal injections 72; and calculating the temporal stability of the signal injections of the current cluster 74. If the signal injection from step 74 is stable 76, the memory management method performs the following steps: receiving a set of historical external factor states 78; calculating the stability of the signal injection relative to the external factor state 80; selecting two states to split the cluster horizontally 82 only if there is enough difference between the two states and there is enough data in each state (after the split) to drive the decision in each state (i.e., calculate the confidence interval); and updating the set of historical clusters 84.

[0029] A cluster is a group of experimental units that are statistically equivalent with respect to measuring causal effects. Within a cluster, the effect is measured without bias and / or confounding effects from external factors, which ensures that we are measuring causality and not just association / correlation. The distribution of the measured effects within each cluster is approximately positively distributed.

[0030] Table 1 provides an algorithm of one embodiment for automatically generating and applying causal knowledge for active battery management, the algorithm including economic factors. The algorithm may be implemented in software or firmware executed by processor 10.

[0031]

[0032] Example

[0033] 1. Methods for maximizing delivered power

[0034] Power characterizes the amount of energy per unit time. Maximizing power is a balancing act of minimizing charging time while minimizing energy capacity loss per cycle. In electric vehicle applications, increased power enables faster acceleration and higher performance. In grid applications such as peak shaving, where typical charge / discharge cycles are 24 hours, increased power translates into longer life and / or smaller installations.

[0035] In this example, we performed the following experiment: 32 batteries were connected to a Maccor cycler (see reference section for details), of which 16 batteries were partially aged (average N cycles / battery) and 16 batteries were brand new (0 cycles / battery). The search space consists of a series of spline curves, which are defined as cubic Bessel functions (see reference for definition), which specify the charging curve of each battery. For implementation purposes, these charging curves are discretized into 5 constant current charging steps. The starting point (constant current up to 3.6V at 200mA) and the end point (constant current up to 4.2V at 100mA) are both fixed; an additional constant voltage step is added at the end of the charging curve (V=4.2V, cutoff current=25mA) to ensure that each battery reaches its full capacity, as is usually done in practice. The 4 independent variables consist of the two coordinates of the two control points of the cubic Bessel function (cutoff voltage V and constant current I). Each independent variable is discretized into 8 levels, resulting in a total of 4096 possible combinations. Once fully charged, the battery is discharged with a fixed discharge profile (constant current of 250 mA to 3 V).

[0036] We focus primarily on the charge curve for purposes of illustration and simplicity, and the disclosed method can be extended to the discharge curve and any other variable associated with the battery cycle. The quality factor (FOM) is defined as the delivered power calculated as the discharge energy divided by the cycle time (i.e., charge time + discharge time). Before any significant aging of the battery, both the discharge energy and the discharge time are almost constant over the entire battery (or cycle), and the FOM is mainly driven by the charge time. Additional dependent variables are recorded for each cycle, including the charge energy and the charge time. External variables are also recorded for each cycle to explore the possible impact of the dependent variables and find clustering opportunities. External variables include battery identification (ID), old batteries relative to new batteries, discharge energy, and discharge time.

[0037] Data is recorded over several weeks. During the initial exploration phase, the algorithm randomly assigns charging curves throughout the search space and builds confidence intervals (CIs) around the expected causal effect of each IV level on the FOM. Once some of the CIs are significantly different and non-overlapping (this happens after exploring about 30% of the total search space), the algorithm begins to exploit this knowledge by assigning the highest utility more frequently, resulting in a gradual increase in the FOM. The algorithm also begins to identify clusters of various causal effects that are statistically different. While the optimal levels may not change between clusters, their relative effects do change due to different aging and different initial states of health.

[0038] Figures 6A-6D The search space of all possible charging curves for the described experiment is shown. Fig. 6A In FIG. 1 , the optimal charge curve is shown as a dashed line, and its endpoints and control handle points are shown as dotted lines. The selected FOM (total delivered power) is plotted against the discharge energy. The power values ​​associated with the optimal charge curve are shown as dark gray circles. The FOM and utilization frequency are also plotted as a function of time. Fig. 6A The search space of all possible charging profiles is shown. Figure 6B The FOM is shown versus the discharge energy. Figure 6C The FOM is shown versus time. Fig.6D The utilization frequency is shown.

[0039] Figures 7A-7C It is shown that over time, the algorithm identifies different effect sizes of old batteries relative to new batteries. Once clustering is initiated, the algorithm utilizes the charging curve that is optimal for each different cluster, which may or may not be the same between clusters. Fig. 7A and Figure 7B An embodiment of a method for maximizing delivered power is shown, where E-Cap refers to energy capacity. Figure 7C An embodiment of a method for maximizing delivered power under ambient conditions is shown.

[0040] 2. Methods for measuring battery internal resistance

[0041] Data from previous experiments are also analyzed to characterize the internal resistance of the battery during each cycle and over time. Mapping the internal resistance as a function of current and state of charge (SOC) can usually be achieved by performing complete cycles at various rates. This analysis is time consuming and furthermore represents only the internal resistance at the beginning of the battery life. This analysis may also be inaccurate because the results of one cycle are independent of the previous cycle. This effect is mitigated by randomization with the algorithm disclosed herein.

[0042] Here, the internal resistance was estimated using data from each actual cycle in the following way: For each constant current step, the internal resistance value was approximated by R = ΔV / ΔI, where ΔV is the average potential change compared to a reference cell cycled at 25 mA, and ΔI is the current difference compared to the reference cell. R was estimated for each cell, each cycle, and for each of the five constant current steps. An almost continuous curve of internal resistance versus state of charge was obtained, which shows the expected behavior of a Li-ion battery. These results are shown in Figures 8A-8C middle.

[0043] Fig. 8A The voltage is shown versus the capacity of the charge curve assigned by the algorithm (solid line) and versus the reference C / 20 charge curve (dashed line), which shows the overpotential ΔV. Figure 8B SOC and overpotential versus current are shown. Figure 8C The internal resistance is shown relative to the SOC.

[0044] The knowledge of the in-situ internal resistance over time can be used to improve the performance and safety of the battery management system by eliminating charging curves in the search space that can cause significant over- and under-voltage, heating, and degradation of the battery. The internal resistance map can be constructed from subgroups within the data, and examples of possible subgroups include, but are not limited to: battery age, number of cycles, temperature exposure, cumulative discharge energy, average discharge current, average charge current, maximum charge current, average voltage, manufacturing batch. The change in internal resistance can also be used to detect abnormal conditions, such as a short circuit in the battery pack.

[0045] 3. Methods for balancing cells in a battery pack

[0046] In addition to maximizing power, causal knowledge of the effects of charging curve variables on battery power can also be used to optimize other utility metrics, such as balancing the state of health (SOH) or the internal resistance across the battery. While battery aging tends to be fairly uniform (within manufacturing tolerances) at the beginning of the battery life, it becomes increasingly non-uniform and unpredictable with each additional charge / discharge cycle. This is an important consideration when repackaging used batteries into battery packs for new applications as a way to extend their life in less demanding applications. An example is the reuse of EV batteries for grid-level energy storage. The development of intelligent battery management systems capable of balancing battery aging is critical to ensuring safe, reliable, and durable operation and making the application economically viable. Accurately measuring the internal resistance of each cell and across clusters of homogeneous battery cells is an effective mechanism for quantitatively achieving battery balancing in practice.

[0047] Similarly, hybrid vehicles may use systems that combine different types of batteries with different performance and aging characteristics. The electric grid may use systems that combine high power and high energy storage. This translates into greater data variance, making it more difficult to apply standard data analysis techniques for battery management. The algorithms disclosed herein can address this type of problem by automatically identifying a minimal set of homogeneous clusters over time that can be used for reliable causal inference. The algorithms disclosed herein can also be used to conduct experiments on batteries between different vehicles.

[0048] 4. Methods for maximizing delivered power under different environmental conditions

[0049] In many applications, batteries are exposed to different environmental conditions due to weather, infrequent operation, etc. For illustration purposes, we focus on ambient temperature as an external factor, which is common to many applications and has a huge known effect on battery performance. We performed a second set of experiments in which 8 batteries were placed in an oven at an elevated ambient temperature of 45°C. The results show how the optimal charging curve varies with environmental conditions.

[0050] 5. Vehicle to Grid

[0051] Vehicle to Grid (V2G) refers to a system in which a plug-in electric vehicle, such as a battery-powered electric vehicle, an electric hybrid vehicle, or a hydrogen fuel cell electric vehicle, is connected to the grid to sell demand response services by returning power to the grid or by throttling its charging rate. Therefore, V2G can be used with plug-in electric vehicles that have grid capacity. Since most vehicles are parked at any given time, the batteries in the electric vehicles can be used to allow power to flow from the vehicle to the distribution grid and back. V2G storage capabilities can also enable electric vehicles to store and discharge power generated from renewable energy sources such as solar and wind, whose output fluctuates depending on the weather and time of day.

[0052] Batteries have a finite number of charge cycles and shelf life. Therefore, using a vehicle as a grid storage device can impact battery life. For example, cycling a battery two or more times per day has shown a dramatic reduction in capacity and greatly shortened life. However, battery capacity is a complex function of factors such as battery chemistry, charge and discharge rates, temperature, state of charge, and age.

[0053] Deep causal learning can be used to optimize the profit and / or total cost of ownership of electric vehicles, given the uncertainty about whether it is economically advantageous or beneficial to sell a portion of the available capacity in an electric vehicle battery pack (energy arbitrator) at any given time given a particular battery history. For example, a vehicle or fleet owner can set a maximum allowed depth of discharge (or equivalently an estimate of the minimum range of the vehicle if required before a full recharge) and a daily schedule that the vehicle will need to be available for driving next (presumably with a full charge). Within the battery management system, the parameters that control the charge curve from a given depth of discharge, the state of the battery pack's external heater, and the discharge rate (up to the maximum allowed by the circuit the vehicle is plugged into) are also independent variables.

[0054] Ignoring the electricity costs associated with the external heater of the battery pack, the potential profit (dependent variable) of selling some of the next available hour of available energy stored in the battery pack can be calculated as described in the following formula (1) in a publication by SB Peterson, JF Whitacre and J. Apt, J. Power Sources 195, (2010) 2377-2384):

[0055]

[0056] in

[0057] LMP 销售 is the electricity sales price at the current time ($ / kWh);

[0058] LMP 销售 is the electricity purchase price ($ / kWh) at the planned recharging time;

[0059] DCH 效率 is the discharge efficiency;

[0060] CH 效率 is the charging efficiency; and

[0061] T&D is the transmission and distribution cost ($ / kWh).

[0062] The degradation cost associated with discharging the energy arbitrator depends on the battery replacement cost, the current estimated state of health (SOH) of the battery, the state of health (SOH) at which the battery is no longer considered usable for its primary application, and the min， Typically set at 80% for vehicles or other stationary battery backup systems) and a V2G Deg factor, which is the marginal impact of the next discharge / charge cycle on the battery's health.

[0063]

[0064] kWh交易的 (or can be used for trading) is the battery's state of charge (SOC) minus the maximum allowed depth of discharge.

[0065] In the absence of uncertainty about the SOH or SOC of the battery, the problem reduces to a planning problem with uncertainty in hourly electricity pricing that can be handled using conventional Monte Carlo techniques. However, both SOC and SOH (which includes the effect of internal resistance) are complex functions of the battery's history. Therefore, by continuously experimenting with the independent variables provided below, deep causal learning can optimize the policy on when and how to discharge / recharge the battery pack to maximize the profit or total cost of ownership (initial cost - total profit over the battery's life) while still retaining enough charge for the primary vehicle when needed.

[0066] The independent variables include:

[0067] C dis = discharge rate (kW / h);

[0068] DoD = maximum depth of discharge (relative to initial capacity);

[0069] The electricity consumption must be completed within the planned time;

[0070] Charging curve (current, voltage and temperature profile through each segment); and

[0071] ΔN = number of cycles between full discharge and estimated state of charge.

[0072] A similar analysis applies to the case where the energy capacity of any other stationary storage device or battery backup system (such as an uninterruptible power system) is used, although there is no constraint that the system must be used for vehicle power. In a frequency regulation market, the battery bank owner or provider provides a certain amount of power to the grid at a certain price at which the battery bank owner or provider believes that the net profit from formula (1) will be greater than zero. If the provider wins the bid during this time period, the provider must follow up or be penalized in future bidding rounds. In the economic analysis of such systems, the penalty factor becomes an external variable.

[0073] References to Examples 1-4 :

[0074] Cell: Rechargeable Li-ion polymer battery wrap-around pouch cell, 5.0 x 30 x 35 mm, nominal capacity 500 mAh; graphite anode and LiCoO2 cathode (E-Group, Mt. Laurel, New Jersey).

[0075] Maccor: 96-channel, 4000 Series Maccor (Tulsa, Oklahoma).

[0076] Cubic Bessel function:

[0077] Four points P0, P1, P2, and P3 in a plane or higher dimensional space define a cubic Bezier curve. The curve starts at P0 towards P1 and reaches P3 from the direction of P2. Typically, the curve will not pass through P1 or P2; the points are only there to provide directional information. The distance between P1 and P2 determines "how far" and "how fast" the curve moves towards P1 before turning towards P2.

[0078] For P i , P j and P k Define the quadratic Bezier curve B Pi,Pj,Pk (t), a cubic Bezier curve can be defined as an affine combination of two quadratic Bezier curves:

[0079] B(t)=(1-t)B P0,P1,P2 (t)+tB P1,P2,P3 (t),0≤t≤1

[0080] The explicit form of this curve is:

[0081] B(t)=(1-t) 3 P0+3(1-t) 2 tP1+3(1-t)t 2 P2+t 3 P3,0≤t≤1

[0082] For some choices of P1 and P2, the curve may intersect itself or contain a cusp.

Claims

1. A method for active battery management, comprising the following steps: injecting a randomized controlled signal into the charging or discharging of a battery, including causing the battery to be charged from a power grid; ensuring that said signal injection occurs within normal operating ranges and constraints; In response to the signal injection, monitoring the performance and economic efficiency of the battery; calculating causal knowledge about the relationship between the signal injection, the performance of the monitored battery, and the economic benefit; selecting an optimal signal for charging or discharging the battery based on the causal knowledge; as well as The battery is discharged into the power grid in exchange for economic benefits. 2 . The method of claim 1 , wherein the controlled signal comprises a charging curve or a discharging curve. The method of claim 1 , wherein the controlled signal comprises a charging rate or a discharging rate. The method of claim 1 , wherein the controlled signal comprises a charge curve endpoint or a discharge curve endpoint. 5 . The method of claim 1 , wherein the normal operating range comprises a multi-dimensional space of possible control states generated based on control information and operating constraints.

6. The method of claim 1, wherein the selecting step further comprises selecting the optimal signal based on external data, the external data comprising environmental conditions.

7. The method of claim 6, wherein the environmental condition comprises at least one of temperature, humidity, or airflow around the battery.

8. The method of claim 1, wherein the selecting step further comprises selecting a depth of discharge back into the grid.

9. The method of claim 1, wherein the battery comprises a battery pack.

10. The method of claim 1, wherein the economic benefit is one of net profit or total cost of ownership.

Images