Active Battery Management Methods for Economic Optimization

Abstract

A method for active battery management to optimize battery performance includes providing signal injections for charging and discharging a battery, the signal injections including various charge and discharge profiles, rates, and endpoints. Response signals corresponding to the signal injections are received and utilities of the signals are measured. Based on the utilities of the response signals, data regarding the charging and discharging of the battery is modified to optimize battery performance and determine when to discharge the battery into a power grid to return power to the grid in exchange for an economic benefit, such as a payment or rebate from a utility company.

Description

[Background technology]

[0001] 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 component of the value proposition of any battery system, whose utility depends on reliably and safely delivering a minimum amount of energy over long periods of time, from years for consumer electronics to decades for grid installations. Summary of the Invention

[0002] A first method for active battery management includes injecting a randomized, controlled signal during charging or discharging of a battery, the injecting step 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 battery performance in response to the controlled signal, calculating a confidence interval for a cause-and-effect relationship between battery performance and the controlled signal, and selecting an optimal signal for charging or discharging the battery based on the calculated confidence interval, the selecting step including discharging the battery to the power grid in exchange for an economic benefit.

[0003] A second method for active battery management includes providing a signal injection for charging or discharging a battery, the signal injection including charging the battery from a power grid, and receiving a response signal corresponding to the signal injection. The method also includes measuring a utility of the response signal, accessing data related to the charging or discharging of the battery, the data including discharging the battery to the power grid in exchange for an economic benefit, and modifying the data based on the utility of the response signal. [Brief explanation of the drawings]

[0004] The accompanying drawings, which are incorporated in and constitute a part of this specification, and together with the description, serve to explain the advantages and principles of the present invention. [Figure 1] FIG. 1 illustrates a system for implementing an active battery management method. [Figure 2] 1 is a flowchart of the search space method of the system. [Figure 3] 1 is a flow chart of a signal injection method for the system. [Figure 4] 1 is a flow chart of a continuous training method for the system. [Figure 5] 1 is a flowchart of a memory management method for a system. [Figure 6A] FIG. 1 illustrates a search space of all possible charging profiles for an example embodiment. [Figure 6B] FIG. 1 illustrates a search space of all possible charging profiles for an example embodiment. [Figure 6C] FIG. 1 illustrates a search space of all possible charging profiles for an example embodiment. [Figure 6D] FIG. 1 illustrates a search space of all possible charging profiles for an example embodiment. [Figure 7A] FIG. 10 illustrates that the algorithm in the example identified a difference in effect size for old versus new cells. [Figure 7B] FIG. 10 illustrates that the algorithm in the example identified a difference in effect size for old versus new cells. [Figure 7C] FIG. 10 illustrates that the algorithm in the example identified a difference in effect size for old versus new cells. [Figure 8A] FIG. 10 illustrates voltage versus capacity for a charging profile assigned by an algorithm in an embodiment. [Figure 8B] FIG. 10 illustrates voltage versus capacity for a charging profile assigned by an algorithm in an embodiment. [Figure 8C] FIG. 10 illustrates voltage versus capacity for a charging profile assigned by an algorithm in an embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0005] Embodiments of the present invention include a method for improving battery management by conducting random experiments on charge and discharge variables and inferring their causal effects on utility metrics such as energy capacity, power, fade rate, charge time, internal resistance, state of health, cell imbalance, temperature, cell expansion, and cost of electricity. Linear combinations of any of the above utility metrics can also be defined to yield figures of merit that balance competing requirements. This battery management can be used, for example, in electric or hybrid vehicles, electric bicycles, consumer electronics, grid storage systems, and other vehicles and devices that use batteries.

[0006] Battery management methods may also include economic factors, such as determining when to transmit power back to the grid in exchange for economic benefits, such as payments, discounts, or rebates from a commercial company or another entity, and the types of user demands to potentially address, such as frequency regulation (high frequency) versus peak shaving (low frequency). In particular, embodiments include methods for optimizing the discharge and charge conditions of a battery pack in an electric vehicle or uninterruptible power system so that it can be used as a source of power back to the grid without adversely affecting battery pack life. The methods use deep causal learning to maximize either net revenue from the sale of power back to the utility grid or minimize the cost of electricity to 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 applied continuously over the life of the battery pack.

[0007] 1 illustrates a system for implementing an active battery management method for a battery pack in an electric vehicle or uninterruptible power supply (or other stationary power source). The system includes a processor 10 electrically coupled to a power source from a power grid 12, a load 20, and data storage 22. The power grid 12 provides power to charge one or more batteries 14, 16, and 18, which in turn power the load 20. In this application, the power grid can be the energy source or the load in providing power back to the grid. The data storage 22, such as an electronic memory, stores profiles and parameters 24, external data 26, and results 28. The results can include, for example, time series of current and voltage, energy capacity, temperature, etc., for each cell or cell string.

[0008] During use, processor 10 uses profiles and parameters 24, and possibly external data 26, to inject signals into power grid 12 to evaluate the performance of batteries 14, 16, and 18, e.g., how they charge and discharge. Performance metrics may include, for example, delivered power, energy capacity, fade rate, revenue, profitability, and reliability score (i.e., a score given by utility to market participants, representing the ability to meet demand within the allocated capacity from the previous day). Processor 10 stores responses to the signal injections as results 28 and can use those responses to optimize battery performance. Processing for battery management may be performed locally on a standalone PC, with dedicated firmware on the battery charging system, or cloud-based and remote from the batteries.

[0009] The profiles and parameters 24 include possible charge and discharge rates, charge and discharge profiles, and profile endpoints. The profiles and parameters 24 also include how often to perform a full charge / discharge cycle to obtain a new estimate of the state of health (defined as maximum utilization relative to maximum capacity at time zero), which applications to request use for (taking into account that different applications may generate different revenues and have different impacts on the life of the battery pack), and for each application, how much capacity to allocate to "grid load" versus standard load (more capacity = more revenue, but more opportunities for discharge).

[0010] Charge and discharge profiles include the shape of such profiles and, in some cases, the time at a particular state of charge or voltage. Charge endpoints include the percentage of charge in the battery when charging begins and stops, and discharge endpoints include the percentage of charge in the battery when discharging begins and stops. Profiles and parameters may be stored, for example, in a lookup table. External data 26 may include, for example, environmental conditions or factors such as the temperature, humidity, and airflow surrounding the battery; time of day or season; time since the device or vehicle was turned on using the battery; estimated state of health (SOH); and state of charge (SOC) from an existing battery model (typically provided by the battery or BMS supplier). Active cooling or heating of the battery may also be used as another control variable with parameters for temperature setpoint, rate, and slope over time and space. For portable electronic devices, external data may also include the following about such devices: application usage, user settings, scheduled events or alarms, power consumption patterns, time of day, or device location. As with electric vehicles, external data may include the time of day or next scheduled use, the vehicle's typical driving patterns, electricity costs over time (e.g., to avoid peak prices), predicted weather conditions, planned route of travel, or traffic conditions.

[0011] A battery can include a single physical battery or multiple physical batteries that collectively provide power. In the case of multiple physical batteries, the batteries may have the same or different constructions or electrochemistries. The process of injecting signals for battery charging and discharging seeks to optimize the charge and discharge profile of a particular battery or pack of batteries. A pack of batteries can be considered a single battery where the pack operates collectively together, or the pack of batteries can be considered multiple physical batteries powered one by one. Examples of battery types include lithium ion, reflow, lead, and others.

[0012] 2-5 are flowcharts of methods for active battery management to optimize charge and discharge profiles and parameters. These methods may be implemented, for example, in software modules for execution by processor 10.

[0013] Figure 2 is a flowchart of the system's search space method, which includes steps 30 of receiving control information (including costs), 32 of constructing a multidimensional space of all possible control states, 34 of constraining the space of potential control states, 36 of determining the normal / baseline sampling distribution, 38 of determining the sampling distribution with the highest utility, and 40 of performing automated control selection within the constrained space.

[0014] 3 is a flowchart of a signal injection method that includes steps of receiving 42 a set of potential signal injections, calculating 44 the spatial and temporal reach of the signal injections, adjusting 46 the signal injections in space and time, performing 48 the signal injections, collecting 50 response data, and associating 52 the response data with the signal injections.

[0015] Signal injection is a change in charge / discharge profile and parameters for battery management, including determining when to deliver power back to the power grid. The response to a signal injection is typically battery performance resulting from or related to the changes in profile and parameters resulting from the signal injection. For example, an algorithm can perturb values ​​in a lookup table representing the charge / discharge profile and parameters, then monitor and store the corresponding battery performance response. The temporal and spatial reach of the signal injection are related to when and where to measure the response signal to the signal injection, respectively, which is used to calculate causality. The cost of the signal injection is typically related to how much the signal injection affects battery performance, e.g., whether the signal injection may cause a degradation in battery performance, and is controlled by the specific experimental scope. The queue for signal injection includes the order and priority of the signal injection, and relies on blocking and randomization to always ensure high internal validity, even when optimizing utility. The utility of the response to the signal injection includes another measure of the effectiveness or utility of the signal injection.

[0016] 4 is a flowchart of the continuous learning method, which includes steps of receiving 54 a set of potential signal injections, receiving 56 a current belief state, calculating 58 a learned value for the signal injections, receiving 60 a cost for the signal injections, selecting and adjusting 62 a signal injection, performing 64 a signal injection, collecting 66 response data, and updating 68 a belief state.

[0017] Belief states are a set of different models of battery performance and / or economic benefit models as a function of charge and discharge. 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 that SOC level, small charges have little impact on battery life / health). The primary source of revenue and battery life loss (the main trade-off) is the capacity required for usage demands, and application, given the battery's health and condition. These belief states can be thought of as uncertainty values ​​that reflect the likelihood of their accuracy given the current set of tests and knowledge that may tend to confirm or falsify these various models. Information that can further confirm or falsify the models may be contained in this data or may be derived from the fundamental properties of the particular model and the physical properties of the underlying system.

[0018] A learned value is a measurement where knowledge generated as a result of signal injection can inform subsequent decision-making by the system, such as determining that a particular charge or discharge profile is more likely to be optimal. In terms of multi-objective optimization, this can involve complex trade-offs between operational goals (e.g., performance vs. range), and optimality can change over time. A learned value may be calculated, for example, by predicting raw numerical data of belief states that can be falsified according to the predictions of a partially observable Markov decision process (POMDP) ​​or other statistical model, and the predicted impact of signal injection on the uncertainty level in the belief states in such a model, or an experimental power analysis calculates the reduction in uncertainty and narrowing of confidence intervals based on an increase to the current sample size.

[0019] 5 is a flowchart of a memory management method. The memory management method includes receiving 70 a set of historical clusters, receiving 72 a set of historical signal injections, and calculating 74 the temporal stability of the signal injections for the current cluster. If the signal injections from 74 are stable (76), the memory management method performs the following steps: receiving 78 a set of historical exogenous factor states, calculating 80 the stability of the signal injections relative to the exogenous factor states, selecting two states, splitting the cluster (82) only if there is sufficient variance across the two states and enough data within each state (after the split) to be able to drive a decision at each state (i.e., calculate a confidence interval) and updating 84 the set of historical clusters.

[0020] A cluster is a group of experimental units that are statistically equivalent with respect to the causal effect measured. Within a cluster, effects are measured without bias and / or confounding effects from extraneous factors, thereby ensuring that causation is measured and not just correlation / association. The distribution of measured effects within each cluster is approximately normally distributed.

[0021] Table 1 describes an embodiment algorithm for automatically generating and applying causal knowledge for active battery management, including economic factors. The algorithm may be implemented in software or firmware for execution by processor 10. [Table 1] [Example]

[0022] 1. How to maximize delivered power Power characterizes the amount of energy per unit time. Maximizing power is a balancing act that minimizes charge time while minimizing the loss of energy capacity with each cycle. In electric vehicle applications, increased power allows for increased acceleration and improved performance. In grid applications such as peak shaving, where a typical charge / discharge cycle is 24 hours, increased power translates into longer lifespans and / or smaller equipment.

[0023] In this example, the following experiment was performed: 32 cells were connected to a Maccor cycler (see the reference section for details), 16 cells were partially aged (average N cycles / cell) and 16 cells were new (0 cycles / cell). The search space consisted of a family of spline curves defined as cubic Bézier functions (see references for definitions) that specified the charging profile for each cell. For implementation purposes, these charging curves were divided into five constant-current charging steps. Both the start point (constant current up to 3.6 V at 200 mA) and the end point (constant current up to 4.2 V at 100 mA) were fixed, and an additional constant-voltage step (V = 4.2 V, cutoff current = 25 mA) was added at the end of the charging profile to ensure that each cell reached its full capacity, as is commonly done in practice. The four independent variables consisted of the two coordinates of the two control points of the cubic Bézier function (cutoff voltage: V, and constant current: I). Each independent variable was divided into eight levels, giving a total of 4096 possible combinations. Once fully charged, the cells were discharged under a fixed discharge profile (constant current at 250 mA to 3 V).

[0024] While the inventors focused primarily on charge profiles for illustrative purposes and simplicity, the disclosed method can be extended to discharge profiles as well as any other variables related to cell cycling. The figure of merit (FOM) was defined as the delivered power calculated as the discharge energy divided by the cycle time (i.e., charge time + discharge time). Prior to significant cell aging, both the discharge energy and discharge time were approximately constant across the cell (or cycle), and the FOM varied primarily with charge time. Additional dependent variables were recorded at each cycle, including charge energy and charge time. External variables were also recorded at each cycle to explore the possible effects of the dependent variables and identify clustering opportunities. The external variables included cell identification (ID), old vs. new cell, discharge energy, and discharge time.

[0025] Data were recorded over several weeks. During the initial exploration phase, the algorithm randomly assigned charge profiles across the search space and constructed confidence intervals (CIs) around the expected causal effect of each IV level on the FOM. After exploring approximately 30% of the total search space, some of the resulting CIs were significantly distinct and non-overlapping. The algorithm then began to utilize that knowledge by assigning levels with the highest utility more frequently, resulting in a gradual increase in the FOM. The algorithm also began to identify clusters across statistically distinct causal effects. While optimal levels may not change across clusters, their relative effects changed due to differences in deterioration over time and differences in initial health states.

[0026] 6A-6D show the search space of all possible charging profiles for the described embodiment. In FIG. 6A, the optimal charging profile is shown as a dashed line, and its endpoints and control handle points are shown as dash-dot lines. The selected FOM (total delivered power) is plotted against discharge energy. The power values ​​associated with the optimal charging profile are shown as dark gray circles. The FOM and utilization frequency are also plotted as a function of time. FIG. 6A shows the search space of all possible charging profiles. FIG. 6B shows the FOM against discharge energy. FIG. 6C shows the FOM against time. FIG. 6D shows the utilization frequency.

[0027] 7A-7C show the algorithm identifying differences in effect size of old cells versus new cells over time. Once clustering begins, the algorithm utilizes an optimal charge profile for each distinct cluster, which may or may not be the same across clusters. Figures 7A and 7B show an example of a method for maximizing delivered power, where E-Cap refers to energy capacity. Figure 7C shows an example of a method for maximizing delivered power under environmental conditions.

[0028] 2.Method for measuring the cell internal resistance Data from previous experiments was also analyzed to characterize the cell's internal resistance during each cycle and over time. Mapping the internal resistance as a function of current and state of charge (SOC) can typically 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 cell's life. This analysis can also be inaccurate because the results of one cycle are not independent of previous cycles. The randomization algorithm disclosed herein mitigates this effect.

[0029] Here, data from each actual cycle was used to estimate the internal resistance in the following manner: for each constant current step, the internal resistance value was approximated by R = ΔV / ΔI, where ΔV was the change in average potential compared to a reference cell cycled at 25 mA, and ΔI was the current difference compared to the reference cell. R was estimated for each cell, and for each cycle and each of the five constant current steps. A nearly continuous curve of internal resistance versus state of charge was obtained, demonstrating the expected behavior of a lithium-ion battery. These results are shown in Figures 8A-8C.

[0030] Figure 8A shows voltage vs. capacity for the charge profile assigned by the algorithm (solid line) for the reference C / 20 charge profile (dashed line), with overvoltage ΔV illustrated. Figure 8B shows SOC and overvoltage vs. current. Figure 8C shows internal resistance vs. SOC.

[0031] Knowledge of in-situ internal resistance over time can be used to improve the performance and safety of battery management systems by eliminating charging profiles within the search space that may result in significant over- and under-voltages, cell heating, and degradation. Internal resistance maps can be constructed from subgroups within the data, with examples of possible subgroups including, but not limited to, cell age, cycle count, temperature exposure, cumulative discharge energy, average discharge current, average charge current, maximum charge current, average voltage, and manufacturing batch. Changes in internal resistance can also be used to detect abnormal conditions, such as the development of short circuits within a battery pack.

[0032] 3. How to balance cells in a pack In addition to maximizing power, causal knowledge of the effects of charge profile variables on cell power can also be used to optimize other utility metrics, such as balancing state-of-health (SOH) or overall cell internal resistance. Battery aging tends to be fairly uniform (within manufacturing tolerances) at the beginning of a battery's life, but becomes increasingly heterogeneous and unpredictable with each additional charge / discharge cycle. This is an important consideration when repackaging used cells for new applications as a way to extend their life in less demanding applications. One example is reusing EV batteries for grid-level energy storage. Developing smart battery management systems that can balance cell aging is critical to ensuring safe, reliable, and durable operation and making applications economically viable. Accurate determination of each cell's internal resistance, as well as clustering across homogeneous cell groups, is an effective mechanism for quantitatively implementing cell balancing in practice.

[0033] Similarly, hybrid vehicles can use systems that combine different types of cells with different performance and aging characteristics. Power grids can use systems that combine high power and high energy storage. This leads to greater dispersion in the data, 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 minimum set of homogeneous clusters that can be used for reliable causal inference over time. The algorithms disclosed herein can also be used to conduct experiments on batteries across different vehicles.

[0034] 4. How to maximize delivered power under different environmental conditions In many applications, batteries are exposed to different environmental conditions due to weather, irregular operation, etc. For illustrative purposes, we focused on ambient temperature as an external factor. This is common in many applications and has a known large effect on cell performance. We conducted a second set of experiments with eight cells placed in an oven at a high ambient temperature of 45°C. The results showed how the optimal charge profile changed with environmental conditions.

[0035] 5. Vehicle to Grid Vehicle-to-grid (V2G) refers to a system in which plug-in electric vehicles, such as battery-electric electric vehicles, electric hybrid vehicles, or hydrogen fuel cell electric vehicles, communicate with the power grid to return power to the grid or sell demand response services by controlling their charging rate. Therefore, V2G can be used with plug-in electric vehicles that have grid capacity. Because most vehicles are parked at any given time, batteries within the electric vehicle are used to enable the flow of electricity between the vehicle and the electric distribution network. V2G storage capacity can also enable electric vehicles to store and discharge electricity generated from renewable energy sources, such as solar and wind, whose output varies depending on the weather and time of day.

[0036] Batteries have a finite number of charge cycles and a finite shelf life. Therefore, using a vehicle as a grid storage device can affect battery life. For example, cycling a battery more than twice a day significantly reduces capacity and shortens lifespan. However, battery capacity is a complex function of factors such as battery chemistry, charge / discharge rate, temperature, state of charge, and age.

[0037] Given the uncertainty regarding whether it would be economically profitable or beneficial to sell a portion of the available capacity in an electric vehicle battery pack (energy arbitrage) at any given time given a particular cell history, deep causal learning can be used to optimize the profitability and / or total cost of ownership of an electric vehicle. For example, a vehicle or fleet owner would set a maximum allowable depth of discharge (or equivalently estimated minimum range of the vehicle, if required before a full recharge) and a daily schedule for when the vehicle is next required to be driven (perhaps on a full charge). Within a battery management system, parameters controlling the charging profile from a given depth of discharge, the state of an external heater for the battery pack, and the discharge rate (up to the maximum allowed by the circuitry to which the vehicle is plugged) are also independent variables.

[0038] Ignoring the electricity costs associated with an external heater for the battery pack, the potential profit (dependent variable) from selling a portion of the available energy stored in the battery pack for the next available time can be calculated as set forth in the following equation (1) in the publication by S.B. Peterson, J.F. Hitacre and J. Apt., J. Power Sources 195, (2010) 2377-2384:

number

[0039] The degradation costs associated with energy arbitrage discharge are calculated based on the battery replacement cost, the current estimated state of health (SOH) of the battery, and the state of health (SOH) when the battery is no longer useful for its primary use. 最小 , typically set at 80% for automotive or other stationary battery backup systems), by the V2G Deg factor, which is the marginal effect of the next discharge / charge cycle on the battery's health.

number

[0040] kWh 取引済 (or available for trading) is the battery's State of Charge (SOC) minus the maximum depth of discharge allowed.

[0041] In the absence of uncertainty about the battery's SOH or SOC, the problem reduces to a scheduling problem with uncertainty in the price of electricity per hour, which can be handled using traditional Monte Carlo techniques. However, both SOC and SOH (including the effect of internal resistance) are complex functions of cell history. Thus, by continuously experimenting with the independent variables described below, deep causal learning can optimize a policy for when and how to discharge / recharge the battery pack to maximize profit or total cost of ownership (initial cost minus total profit over the battery's lifetime) while still leaving sufficient charge available for the vehicle's primary use when needed.

[0042] The independent variables include: C 放電 = discharge rate (kW / h). DoD = Maximum Depth of Discharge (relative to initial capacity). Charging must be completed at the scheduled time. Charging profile (current, voltage, and temperature profile through each segment). ΔN = number of cycles between full discharges to assess state of charge.

[0043] A similar analysis applies when using the energy capacity of any other stationary storage or battery backup system, such as an uninterruptible power system, but not necessarily for automotive power. In a frequency regulation market, the battery pack owner or provider provides a certain amount of power to the grid at a price that the provider believes results in a net benefit greater than zero from equation (1). If the provider wins the usage request for that period, the provider must comply with or be penalized for future usage requests. The penalty factor is an exogenous variable in the economic analysis of this type of system.

[0044] See Examples 1 to 4 Cell: Rechargeable lithium-ion polymer battery rolled pouch cell, 5.0 x 30 x 35 mm, nominal capacity 500 mAh; graphite anode and LiCoO2 cathode (E-Group, Mt. Laurel, New Jersey).

[0045] Maccor: 96 channels, Series 4000 Maccor (Tulsa, Oklahoma).

[0046] Cubic Bezier functions: Four points P0, P1, P2, and P3 in a plane or higher-dimensional space define a cubic Bézier curve. The curve starts at P0, goes to P1, and reaches P3 from the direction of P2. Normally, the curve does not pass through P1 or P2; these points exist only to provide directional information. The distance between P1 and P2 determines "how far" and "how fast" the curve travels toward P1 before turning toward P2.

[0047] P i , P j , and P k B for the quadratic Bézier curve defined by Pi,Pj,Pk Writing (t), a cubic Bézier curve can be defined as an affine combination of two quadratic Bézier curves. B(t)=(1-t)B P0,P1,P2 (t)+tB P1,P2,P3 (t), 0≦t≦1

[0048] The explicit form of the curve is: B(t)=(1-t) 3 P0+3(1-t) 2 tP1+3(1-t)t 2 P2+t 3 P3,0≦t≦1 For some choices of P1 and P2, the curve may intersect itself or include a cusp. In addition to the above-described embodiments, the following aspects will be noted. (Appendix 1) 1. A method for active battery management, comprising: injecting a randomized and controlled signal during charging or discharging of a battery, the injecting step comprising charging the battery from a power grid; ensuring that said signal injection occurs within normal operating ranges and constraints; monitoring the performance of the battery in response to the controlled signal; calculating a confidence interval for a causal relationship between the battery performance and the controlled signal; selecting an optimal signal for the charging or discharging of the battery based on the calculated confidence interval, the selecting step comprising discharging the battery to the power grid in exchange for an economic benefit. (Appendix 2) 2. The method of claim 1, wherein the controlled signal comprises a charge profile or a discharge profile. (Appendix 3) 2. The method of claim 1, wherein the controlled signal comprises a charge rate or a discharge rate. (Appendix 4) 2. The method of claim 1, wherein the controlled signal comprises a charge profile endpoint or a discharge profile endpoint. (Appendix 5) 2. 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. (Appendix 6) 2. The method of claim 1, wherein the selecting step further comprises selecting the optimal signal based on external data including environmental conditions. (Appendix 7) 7. The method of claim 6, wherein the environmental conditions include at least one of temperature, humidity, or airflow around the battery. (Appendix 8) 2. The method of claim 1, wherein the selecting step further comprises selecting a depth of discharge back to the power grid. (Appendix 9) 2. The method of claim 1, wherein the battery comprises a pack of batteries. (Appendix 10) 2. The method of claim 1, wherein the economic benefit is one of net benefit or total cost of ownership. (Appendix 11) 1. A method for active battery management, comprising: providing a signal injection for charging or discharging a battery, including charging the battery from a power grid; receiving a response signal corresponding to the signal injection; measuring the utility of the response signal; accessing data regarding the charging or discharging of the battery, including discharging the battery into the power grid in exchange for an economic benefit; and modifying the data based on the utility of the response signal. (Appendix 12) 12. The method of claim 11, wherein the signal injection comprises a charge profile or a discharge profile. (Appendix 13) 12. The method of claim 11, wherein the signal injection comprises a charge rate or a discharge rate. (Appendix 14) 12. The method of claim 11, wherein the signal injection comprises a charge profile endpoint or a discharge profile endpoint. (Appendix 15) 12. The method of claim 11, wherein the accessing step includes accessing a lookup table. (Appendix 16) 12. The method of claim 11, wherein the signal injection has a spatial reach. (Appendix 17) 12. The method of claim 11, wherein the signal injection has a temporal reach. (Appendix 18) 12. The method of claim 11, wherein the modifying step further comprises modifying the data based on external data including environmental conditions. (Appendix 19) 19. The method of claim 18, wherein the environmental conditions include at least one of temperature, humidity, or airflow surrounding the battery. (Appendix 20) 12. The method of claim 11, wherein the providing step further comprises selecting a depth of discharge back to the power grid. (Appendix 21) 12. The method of claim 11, wherein the battery comprises a pack of batteries. (Appendix 22) 12. The method of claim 11, wherein the economic benefit is one of net benefit or total cost of ownership.

Claims

1. 1. A method for active battery management by a battery management system, comprising: injecting signals into the battery management system that cause changes in battery charge and discharge profiles and parameters; monitoring the battery for changes in performance in response to the injected signal; calculating a confidence interval for a causal relationship between the battery's performance change and the injected signal; selecting optimal charge / discharge profiles and parameters for charging or discharging the battery based on the calculated confidence interval.

2. The method of claim 1 , wherein the injected signal comprises a charge profile or a discharge profile.

3. A method for active battery management, comprising: providing a signal injection for charging or discharging a battery, including charging the battery from a power grid; receiving a response signal corresponding to the signal injection; measuring the utility of the response signal; accessing data regarding the charging or discharging of the battery, including discharging the battery into the power grid in exchange for an economic benefit; and modifying the data based on the utility of the response signal.

Images