Methods and systems for estimating the state of charge of batteries, electronic devices and vehicles
By combining the ampere-hour integration method and the water cycle algorithm, an equivalent circuit model of the battery is established, intermediate parameters are identified, and the state of charge (SOC) value of the lithium-ion battery is optimized. This solves the problems of insufficient accuracy and speed in existing methods and achieves efficient and accurate SOC estimation.
Patent Information
- Application Number
- CN202110276282.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-03-15
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2041-03-15
AI Technical Summary
Existing methods for estimating the state of charge (SOC) of lithium-ion batteries, such as the ampere-hour integration method and the Kalman filter method, are insufficient in terms of accuracy and convergence speed, making it difficult to achieve high-precision and fast SOC estimation at the battery pack level.
By combining the ampere-hour integration method and the water cycle algorithm, an equivalent circuit model of the battery is established, intermediate parameters are identified using the water cycle algorithm, and the actual state of charge of the battery is calculated by combining the trust factor.
It provides high-precision state of charge estimation in a short time, quickly eliminates errors, is suitable for the rapid response of battery management systems, and improves the efficiency and safety of battery packs.
Smart Images

Figure CN115079003B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of battery state of charge estimation, and more specifically, to a battery state of charge estimation method combining a water cycle algorithm and an ampere-hour integral method. The invention also relates to related systems, electronic devices, and vehicles for performing this battery state of charge estimation method. Background Technology
[0002] In recent years, as people have become increasingly concerned about environmental issues, especially global warming caused by rising CO2 concentrations, more and more regulations and policies have been formulated to reduce or limit CO2 emissions. In this context, more and more automakers are actively embracing the trend of vehicle electrification, and the share of electric vehicles in the global automotive market has been increasing year by year. In this transformation towards vehicle electrification, battery systems, as the energy source for electric vehicles, are becoming increasingly important, especially high-voltage battery systems that provide the primary power. Among the many types of batteries, lithium (lithium-ion) batteries are currently the most popular in the market. Lithium batteries are widely used due to their relatively high power density, energy density, safety and stability, and relatively long lifespan.
[0003] For safety and capacity considerations, lithium-ion battery packs consist of hundreds of lithium-ion battery cells connected in series or parallel, monitored by a battery management system (BMS). One of the most important functions of the BMS is estimating the state of charge (SOC) of each individual lithium-ion battery cell. At the battery pack level, the estimated SOC values of these individual cells ultimately form the overall SOC level of the battery pack. The battery's SOC, or the overall SOC of the battery pack, is a crucial parameter reflecting its current energy level and operating state. The accuracy of the SOC estimation affects the efficiency and safety of lithium-ion batteries during use. High-precision SOC estimation can improve the overall efficiency and lifespan of the battery pack and can be used to balance the power output of individual battery cells within the pack. Overall, stable and accurate battery SOC estimation significantly impacts the performance of electric vehicles.
[0004] In the existing technology, many commonly used methods for estimating the state of charge have been proposed, among which the ampere-hour integral method, the least squares method, and the Kalman filter method are the three most common estimation methods.
[0005] The ampere-hour integration method is the simplest and fastest estimation method. It considers the battery's real-time charge level during charging and discharging by calculating the input or output current. Therefore, the battery's state of charge (SOC) can be calculated by integrating the current over the total charging and discharging time. However, this estimation method oversimplifies the battery's physical model and fails to accurately represent the actual operating conditions. Furthermore, relying on a single variable (current) for estimation means that errors in current measurement, or even measurement errors themselves, will lead to significant discrepancies between the calculated result and the battery's actual SOC. Moreover, it fails to correct for any errors or mistakes during the calculation process, causing errors to accumulate and resulting in unsatisfactory final results. This open-loop algorithm essentially lacks a necessary correction mechanism.
[0006] Least squares is also a commonly used estimation method. However, in some cases, the identification results of the equivalent parameters of the battery cell may get stuck in local optima, and in some cases, the algorithm may not converge, thus failing to find the correct solution.
[0007] Another commonly used method for estimating the state of charge (SOC) of batteries is the Kalman filter. The Kalman filter primarily estimates the SOC of lithium batteries based on an equivalent circuit model. In recent years, to address various SOC estimation problems in the electric vehicle field, a number of improved Kalman filter methods have been proposed, including adaptive unscented Kalman filtering, extended Kalman filtering, and dual-time-scale extended Kalman filtering. However, these algorithms mainly target the accurate SOC estimation of individual battery cells. In a battery pack composed of hundreds or thousands of cells, such accurate estimation of each cell leads to an overly complex identification model and excessive computational cost.
[0008] Therefore, an algorithm with high estimation accuracy, fast convergence speed, and simple and stable performance is needed to overcome the shortcomings of the above algorithms. Summary of the Invention
[0009] To this end, the present invention proposes a method for estimating the state of charge (SOC) of a battery, which combines an ampere-hour integral algorithm and a water cycle algorithm. Specifically, the method for estimating the SOC of a battery includes:
[0010] The first state of charge of the battery is estimated based on the ampere-hour integral method;
[0011] Estimation of the second state of charge of the battery based on a water circulation algorithm; and
[0012] The actual state of charge (SOC) of the battery is estimated based on the first SOC and the second SOC.
[0013] According to an embodiment of the present invention, the expression for estimating the first state of charge of a battery based on the ampere-hour integration method is as follows:
[0014]
[0015] Where SOC0 is the initial state of charge of the battery, C N Let i be the battery capacity, i(t) be the current in the battery, t0 be the initial time, t1 be the estimated time, and SOC be the state of charge. AC This represents the battery's first state of charge (SOC).
[0016] According to an embodiment of the present invention, estimating the second state of charge of a battery based on a water cycle algorithm includes the following steps:
[0017] Establish an equivalent circuit model of the battery, which includes the battery's equivalent physical parameters;
[0018] The expression of the equivalent circuit model is discretized after undergoing a Laplace transformation to obtain the mathematical model.
[0019] Using a water cycle algorithm to identify intermediate parameters in a mathematical model;
[0020] The equivalent physical parameters of the battery are calculated based on the identified intermediate parameters;
[0021] The second state of charge of the battery is estimated based on equivalent physical parameters.
[0022] Advantageously, in the step of establishing the equivalent circuit model of the battery, the expression of the equivalent circuit model of the battery is as follows:
[0023]
[0024] Among them, U oc R is the open-circuit voltage of the battery. o R is the ohmic internal resistance of the battery, I is the current in the battery, and R is the current in the battery. Th C is the polarization internal resistance of the battery. Th For the polarization capacitance of the battery, U Th U is the polarization voltage of the battery. L This is the battery's terminal voltage.
[0025] Advantageously, the expression of the obtained mathematical model is as follows:
[0026] E(k)=α1E(k-1)+α2I(k)+α3I(k-1)
[0027] E(k)=U oc (k)-U L (k)
[0028]
[0029]
[0030]
[0031] τ Th =R Th C Th
[0032] Where k is the discrete time, I(k) is the discretized current, E(k) is the discretized voltage difference, representing the voltage difference between the open-circuit voltage and the terminal voltage, and U... oc (k) is the discretized open-circuit voltage, U L (k) represents the discretized terminal voltage, α1, α2, and α3 are intermediate parameters of the mathematical model, and R o R is the ohmic internal resistance of the battery. Th C is the polarization internal resistance of the battery. Th Let be the polarization capacitance of the battery, and T be the sampling period.
[0033] Advantageously, identifying the intermediate parameters of the mathematical model using the water cycle algorithm includes the following steps:
[0034] The intermediate parameters at time k are estimated using the water cycle algorithm module.
[0035] The error value e(k) was calculated using the following three expressions:
[0036]
[0037]
[0038]
[0039] in, These are estimated values of the intermediate parameters at time k, where E(k-1) is the voltage difference at time k-1, I(k) is the current at time k, and I(k-1) is the current at time k-1. It is an estimate of the voltage difference at time k. It is the estimated value of the open-circuit voltage at time k. U is the estimated value of the terminal voltage at time k. L e(k) is the terminal voltage at time k, and e(k) is the error value at time k.
[0040] Determine whether the stopping condition set based on the error value e(k) at time k is met;
[0041] If the stopping condition is not met, the iteration continues, and the estimation of intermediate parameters at the next time step is controlled by the error value e(k) at time k; if the stopping condition is met, the iterative operation of the water cycle algorithm can be stopped, and the intermediate parameters are output.
[0042] Furthermore, the error value e(k) at time k is used as the independent variable in the fitness function of the water cycle algorithm to specifically limit the fitness function, so as to achieve optimized control of parameter estimation.
[0043] Furthermore, advantageously, the fitness function in the water cycle algorithm can be set as:
[0044] C n (k) = e(k)^2, n = 1, 2, ..., N pop
[0045] Among them, C n e(k) refers to the fitness function at time k, and e(k) is the error value of the input water cycle algorithm module at time k.
[0046] Furthermore, advantageously, the stopping condition is set based on the error value e(k) at time k as e(k)^2 < 0.001.
[0047] Advantageously, the steps for estimating the second state of charge of a battery based on its equivalent physical parameters include:
[0048] The open-circuit voltage of the battery is calculated using an equivalent circuit model based on equivalent physical parameters.
[0049] Based on the calculated open-circuit voltage value, the second state of charge value of the battery is obtained according to the pre-calibrated open-circuit voltage-state of charge mapping.
[0050] According to an embodiment of the present invention, the expression for estimating the actual state of charge (SOC) of the battery based on the first SOC and the second SOC is as follows:
[0051] SOC R =δ·SOC AC +(1-δ)SOC WCA
[0052] Among them, SOC R State of Charge (SOC) represents the actual state of charge of the battery. AC The first state of charge (SOC) of the battery. WCA Let δ represent the second state of charge (SOC) of the battery, and let δ represent the trust factor. Preferably, the trust factor δ varies according to the SOC of the battery and / or the battery's operating duration.
[0053] Specifically, the trust factor δ changes as follows:
[0054] When the battery's state of charge is ≤20% or the battery's operating duration is <60s, the trust factor δ = 1;
[0055] When 20% < battery state of charge ≤ 40% and battery operating duration > 60s, the trust factor δ = 0.6;
[0056] When 45% < battery state of charge ≤ 60% and battery operating duration > 60s, the trust factor δ = 0.4;
[0057] When 60% < battery state of charge ≤ 80% and battery operating duration > 60s, the trust factor δ = 0.2;
[0058] When the battery's state of charge is greater than 80% and the battery's operating time is greater than 60 seconds, the trust factor δ = 0.
[0059] The present invention also relates to a system for estimating the state of charge (SOC) value of a battery, comprising: a first SOC value estimation module for estimating the first SOC value of the battery based on the ampere-hour integration method; a second SOC value estimation module for estimating the second SOC value of the battery based on a water cycle algorithm; and an actual SOC value estimation module for estimating the actual SOC value of the battery based on the first SOC value and the second SOC value.
[0060] In addition, the present invention relates to an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for estimating the state of charge of a battery as briefly described above.
[0061] Finally, the present invention also relates to a vehicle including a battery, the vehicle further including the electronic devices described above.
[0062] Preferably, the battery is a lithium-ion battery. More preferably, the battery is a lithium-ion battery for use in motor vehicles.
[0063] The advantage of this method, which combines the water circulation algorithm and the ampere-hour integration method for estimating the battery's state of charge (SOC), lies in introducing a SOC estimation process based on the water circulation algorithm on top of the conventional ampere-hour integration method. Since the ampere-hour integration method is essentially an open-loop algorithm and ignores the changes in the battery's equivalent physical parameters with environmental conditions and operating time, the SOC value obtained using this method has an initial error, which becomes increasingly pronounced as the battery continues to operate. The SOC estimation process based on the water circulation algorithm can identify the parameters in the battery's equivalent circuit model relatively well online, thus solving for the parameter values that best reflect reality. The SOC estimates derived from these parameters are more accurate. However, the water circulation algorithm requires a period of time and multiple iterations to achieve the desired identification accuracy. Therefore, this method, which combines the ampere-hour integration method and the water circulation algorithm, can obtain sufficient estimation accuracy in a short time to overcome the shortcomings of the water circulation algorithm, while eliminating the errors accumulated by the ampere-hour integration method over longer battery operating times. This allows for reasonably accurate SOC estimates over any length of battery operation.
[0064] Furthermore, compared to methods such as least squares, Kalman filtering, or other improved Kalman filtering methods, the water circulation algorithm converges faster and is less prone to getting trapped in local optima. Therefore, it can provide more stable parameter identification and find the optimal solution more quickly. In addition, the water circulation algorithm has a clear iterative objective and a simple algorithm structure during the optimization process, so the computational load is relatively small, making it more suitable for applications in systems that require fast response, such as battery management systems. Attached Figure Description
[0065] The objectives, technical solutions, and advantages of the present invention will become clearer by reading the following detailed embodiments of the invention with reference to the accompanying drawings, in which:
[0066] Figure 1 This invention illustrates a method for estimating the state of charge of a battery.
[0067] Figure 2 This shows the first-order RC equivalent circuit model of the battery;
[0068] Figure 3 The flowchart illustrating the principle of the water circulation algorithm is shown.
[0069] Figure 4 This demonstrates the process of identifying intermediate parameters of a battery's first-order RC equivalent circuit model using a water cycle algorithm.
[0070] Figure 5 The diagram shows the calculation of the battery's state of charge value from the identified intermediate parameters. Detailed Implementation
[0071] It should be understood that the specific embodiments described below are for illustrative purposes only and are not intended to limit the scope of the invention. It should also be noted that the same equivalent physical parameters in different figures are represented by the same reference numerals.
[0072] This invention proposes a method for estimating the state of charge (SOC) of a battery based on the water cycle algorithm (WCA) and the ampere-hour integration method. This estimation method utilizes the SOC values calculated by the water cycle algorithm (WCA) and the ampere-hour integration method to obtain the final actual SOC value, thereby accurately estimating the SOC of the battery online.
[0073] Specifically, refer to Figure 1 The flowchart shown illustrates that this method for estimating the state of charge of a battery includes the following calculation process:
[0074] The first state of charge (SOC) of the battery was estimated based on the ampere-hour integral method. AC ;
[0075] The second state of charge (SOC) of the battery was estimated based on a water cycle algorithm. WCA ;
[0076] Then, based on the first state of charge (SOC) value obtained using the two algorithms... AC Second State of Charge (SOC) WCA The final actual state of charge value is estimated.
[0077] The estimation process based on the ampere-hour integral method is performed in the first state of charge estimation module, and the estimation process based on the water cycle algorithm is performed in the second state of charge estimation module, which includes at least a water cycle algorithm module for executing the water cycle algorithm.
[0078] The specific process of the battery state-of-charge estimation method according to the present invention will be described in detail below.
[0079] First, the battery state-of-charge estimation method according to the present invention estimates the first state-of-charge (SOC) value of the battery based on the ampere-hour integration method. AC It should be noted that the ampere-hour integral method is a conventional method for estimating the state of charge of a battery, and its mathematical expression is:
[0080]
[0081] Where SOC0 is the initial state of charge of the battery, and C N t is the battery capacity, i(t) is the current, and t0 and t1 represent the initial time and the estimated time, respectively.
[0082] Therefore, according to expression (1), with the input initial state of charge SOC0, time, current and battery capacity C... N In this case, the estimated state of charge (SOC) of the battery at time t1 can be obtained. The SOC value calculated by this algorithm is called the first state of charge (SOC). AC .
[0083] The battery state-of-charge estimation process based on the ampere-hour integration method is executed in the first state-of-charge value estimation module.
[0084] Clearly, the ampere-hour integration method is a relatively simple open-loop algorithm that cannot eliminate errors in the estimation process, and these errors accumulate over time. Furthermore, during actual battery operation, the rated capacity is affected by factors such as cycle aging and temperature. Therefore, if the battery's operating time is long, the state-of-charge (SOC) value estimated by the ampere-hour integration method will be inaccurate and differ significantly from the actual value. Consequently, as the battery charging and discharging time increases, the estimated SOC value will become increasingly inaccurate. AC It cannot be used as an ideal output value for the actual state of charge of the battery.
[0085] Therefore, it is necessary to introduce an accurate calculation process that can eliminate errors, especially a calculation process that maintains high accuracy under long-term battery operation, to solve the problems caused by the ampere-hour integration method. Thus, the battery estimation method according to the present invention includes calculating the second state-of-charge (SOC) value. WCA The process.
[0086] The following will refer to Figure 2 , 3 4. This section describes in detail how to calculate the second state of charge (SOC). WCA The process. Advantageously, battery state-of-charge estimation based on the water cycle algorithm mainly includes the following steps:
[0087] First, an equivalent circuit model of the battery is established to simulate the physical changes of the battery during actual operation. This equivalent circuit model includes the battery's equivalent physical parameters.
[0088] Then, the expression of the equivalent circuit model is discretized after undergoing a Laplace transformation to obtain a discretized mathematical model.
[0089] Then, the intermediate parameters of the mathematical model are identified using the water cycle algorithm;
[0090] Then, the equivalent physical parameters of the battery are calculated based on the identified intermediate parameters;
[0091] Finally, the second state of charge of the battery is estimated based on the equivalent physical parameters.
[0092] Figure 2 The diagram illustrates the battery model required for the estimation process based on the water cycle algorithm according to the present invention. This battery model is a first-order RC equivalent circuit model. In practice, to perform online parameter identification of the battery, an equivalent circuit model of the battery must first be established. Typically, when simulating batteries, a first-order RC equivalent circuit model of the battery is used, as it is easy to build and can well reflect the charging and discharging dynamic characteristics of the battery itself, especially accurately reflecting the characteristics of lithium batteries.
[0093] In such Figure 2 In the equivalent circuit model of the battery shown, each equivalent physical parameter is labeled with a different reference numeral: U oc R is the open-circuit voltage of the battery, which is directly related to the battery's state of charge. o R is the ohmic internal resistance of the battery, which is typically formed by the electrode materials, electrolyte, separator internal resistance, and contact resistance of various components; I is the current in the battery; R Th -C Th This represents an RC circuit element used to simulate the polarization effect of a battery during charging and discharging. R Th C is the polarization internal resistance of the battery. Th For the polarization capacitance of the battery, U Th U is the polarization voltage of the battery. L This is the battery's terminal voltage.
[0094] For the first-order RC equivalent circuit model of a battery, according to Kirchhoff's laws, the following expression exists:
[0095]
[0096] Taking the Laplace transform of expression (1) yields:
[0097]
[0098] Where s is the complex frequency, and other equivalent physical parameters are as described above. It should be noted that the same notation will not be repeated in the expressions below.
[0099] It can be reasonably assumed that the various equivalent physical parameters of the battery and the environment in which the battery is located will not change during the sampling period T. Therefore, under the zero-order hold sampling, expression (3) can be discretized to obtain the following mathematical model expression:
[0100] E(k)=α1E(k-1)+α2I(k)+α3I(k-1) (4)
[0101] E(k)=U oc (k)-UL (k) (5)
[0102]
[0103]
[0104]
[0105] τ Th =R Th C Th (9)
[0106] Where k is a discrete-time variable, I(k) is the discretized current, E(k) is the discretized voltage, representing the difference between the open-circuit voltage and the terminal voltage, and U... oc (k) is the discretized open-circuit voltage, U L (k) is the discretized terminal voltage, α1, α2, and α3 are intermediate parameters, and T is the sampling period.
[0107] Based on expressions (6)-(9), it can be understood that the three intermediate parameters α1, α2, and α3 of the mathematical model can be used to derive the corresponding three equivalent physical parameters R in the battery equivalent model. Th C Th R o This allows for the further calculation of the open-circuit voltage U. oc The following text will refer to Figure 5 To describe in detail this method of calculating open-circuit voltage U based on intermediate parameters α1, α2, and α3 oc The process.
[0108] After deriving the expressions (4)-(9) of the above mathematical model, it is necessary to identify the intermediate parameters α1, α2, and α3 to obtain the various equivalent physical parameters of the battery model. Obviously, the more accurate the identification of the intermediate parameters α1, α2, and α3, the more the relevant equivalent physical parameters calculated from them will match the actual operation of the battery. Therefore, the identification of the intermediate parameters α1, α2, and α3 needs to meet certain identification accuracy requirements, and should also have the characteristics of being fast and convenient, so as to monitor the state of charge of the battery in real time and accurately. This is beneficial for the battery control device to quickly make appropriate control operations during the charging and discharging process of the battery.
[0109] Therefore, an algorithm with fast convergence speed and the ability to largely avoid local optima is needed. In this invention, a water cycle algorithm is used to identify the intermediate parameters of the mathematical model.
[0110] The water cycle algorithm is a metaheuristic algorithm inspired by the natural water cycle phenomenon. It is generally known that water in rivers, lakes, and oceans evaporates into the atmosphere, cools to form clouds, and under certain conditions, becomes rainfall, falling to the ground. Some of this water seeps into the ground to form groundwater and interacts with surface water; some is absorbed by plants and evaporates back into the atmosphere; and the remainder forms surface water that eventually flows into the ocean. Therefore, in nature, water continuously transforms in solid, liquid, and gaseous states in the Earth's surface, ocean, and atmosphere. The water cycle algorithm simulates the natural process of rainfall forming streams and the flow of streams and rivers into the ocean.
[0111] The principle of the water cycle algorithm will be described in detail below using expressions.
[0112] Now, suppose that the following four phenomena mainly exist in the water cycle: streams and rivers are formed by rainfall; streams of varying numbers flow into different rivers or oceans, resulting in different flows of different rivers or oceans; the process of streams and rivers flowing into the ocean; the evaporation of water in streams, rivers, and oceans; and the return of water to streams, rivers, and oceans as rainfall.
[0113] Since rainfall can form streams and rivers, it can be assumed that during rainfall, raindrops are unclassified individuals within the population, meaning they are not categorized as streams, rivers, or oceans. A single raindrop can be considered a single solution, then in N... var In optimization problems involving 3D variables, raindrops should be constrained as a set of vectors as follows:
[0114] Raindrop = [x1, x2, ..., x Nvar (10)
[0115] To begin the optimization algorithm, a population of N is randomly generated. POP If rainfall is applied to this population, the number of raindrops should be N. POP This results in a population of size N. POP ×N var The matrix:
[0116]
[0117]
[0118] Here, "Total population" refers to the total population size, consisting of N individuals. POP It is composed of raindrops.
[0119] Then, each raindrop in the population is classified into streams, rivers, and oceans according to the following fitness function:
[0120]
[0121] Since streams or rivers eventually flow into the sea, individuals with the best fitness function value (i.e., the smallest function value) can be selected as the sea, individuals with relatively good fitness function values or relatively small function values can be selected as rivers, and the remaining individuals can be classified as streams. The fitness function will be further described below in conjunction with the specific intermediate parameter identification process.
[0122] It is obvious that in the population, the number of individuals representing the ocean with the minimum fitness function value is 1. Let N be the total number of individuals in the population divided into ocean and river. sr Then the number of rivers should be N. sr -1, correspondingly, the number of streams N streams It should be equal to N POP -N sr Therefore, expression (11) can be transformed into the following matrix:
[0123]
[0124] Here, "Total population" refers to the total population, which consists of a corresponding number of oceans, rivers, and streams, and can therefore be represented as a variable of size N. POP A matrix of size ×Nvar.
[0125] After rainfall forms streams and rivers, the ability of rivers or oceans to converge streams is considered. Therefore, the water cycle algorithm takes into account that if the flow of a particular river or ocean is greater, the number of streams flowing into that particular river or ocean is considered to be greater. In other words, the better the fitness value of the river or ocean, the closer it is to the global optimum. Conversely, the more streams flowing into the river or ocean, the worse the fitness value, the further it is from the global optimum, and the fewer streams flow into it. The specific expression is given below:
[0126]
[0127] Among them, NS n N refers to the number of streams that flow into a specific river or ocean. streams This refers to the number of streams.
[0128] After classifying and categorizing streams, rivers, and oceans within the population, it's necessary to consider the movement process of streams flowing into rivers or oceans. Streams or rivers flow from one location to a new location. Therefore, in the water cycle algorithm, the expected position update of the solutions representing streams and rivers is considered to search for better solutions representing rivers or oceans, i.e., gradually approaching the river or ocean. Thus, this random position update of streams or rivers is expected to occur between the current stream or river position and a better river or ocean position, specifically limited as follows:
[0129]
[0130]
[0131]
[0132] Where i is the iteration number, and rand is a uniformly distributed random number between 0 and 1 to represent the randomness of the position update. These represent the current locations of the stream, river, and ocean, respectively. Equation (15) represents the updated positions of streams, rivers, and oceans, respectively. Equation (16) represents the position update process of streams flowing directly into the ocean to approximate the ocean position indicating the optimal solution. Equation (17) represents the position update process of streams flowing into rivers to approximate the river position indicating a relatively optimal solution.
[0133] It should be noted that if, after a location update, the fitness value of a stream is better than (i.e., less than) the fitness value of the river it is connected to, then the stream and river positions are swapped. Obviously, after a stream becomes a new river, other streams previously connected to the original river should also flow into this new river. Similarly, if the location of a river or stream flowing into the ocean is better than the location of the ocean, then the ocean and river or stream positions are swapped. In this way, the water cycle algorithm searches for the optimal solution through continuous iterative updates.
[0134] As streams or rivers flow into the sea, water evaporates, causing surface water to re-enter the atmosphere and, under certain conditions, transform into rainfall. Correspondingly, the water cycle algorithm incorporates this process of re-evaporation and rainfall. However, in practice, if the convergence speed of the water cycle algorithm is too fast, it can easily get trapped in local optima. Therefore, to avoid this, it is necessary to force the generation of new individuals under certain conditions, thereby expanding the overall search range of the optimization algorithm and increasing the number of individuals to further search for potentially better solutions.
[0135] Specifically, if it is believed that the solutions of various individuals in the search space of the optimization algorithm are too close, leading to excessively fast convergence during the algorithm's iteration, new individuals are introduced into this space. That is, in the water cycle algorithm, if the distance between a river and the ocean, or a stream and the ocean, is considered too short, evaporation of the river or stream is triggered, resulting in rainfall. The distance threshold for determining the distance between the river and the ocean, or the stream and the ocean, is set to d. max , Decision threshold d max The larger the distance between a river or stream and the ocean, the more likely it is to evaporate and produce rainfall. Since the algorithm iterates continuously, meaning the river or stream gets closer to the ocean, a threshold d needs to be determined. max The value of d is decreased accordingly as the number of iterations increases to avoid the search intensity of the algorithm remaining unduly high, which would lead to an excessively slow convergence speed. Therefore, d is gradually decreased as the iterations progress. max :
[0136]
[0137] in, The current distance threshold d max , The updated distance threshold is set to `maxiteration`, which represents the maximum number of iterations.
[0138] After the evaporation process is completed, rainfall occurs within the target space. Clearly, this new rainfall will form new raindrops, thus increasing the number of individuals within the target space. These newly generated raindrops are specifically defined as follows:
[0139]
[0140] in, LB represents the newly generated raindrop, UB represents the lower bound of the variable to be solved, and rand represents a random number uniformly distributed from 0 to 1.
[0141] Then, the algorithm is iteratively optimized in the newly generated population.
[0142] Based on the principle of the water cycle algorithm described above in conjunction with expressions (10)-(19), it can be seen that the water cycle algorithm seeks the global optimal solution step by step by updating the position of streams or rivers. At the same time, it prevents the algorithm from converging too quickly and getting trapped in a local optimal solution through the evaporation and rainfall processes. Therefore, this optimization algorithm can accurately and quickly identify the variables to be solved.
[0143] exist Figure 3 The document details the water cycle algorithm flow based on the principles of water cycle algorithms, which includes the following sub-steps:
[0144] First, set the initial parameters of the population, including the number of variables N in each individual. var The number of individuals N in the population POP The number of oceans is specifically defined as 1, and the total number of rivers and oceans is N. sr The distance threshold for determining evaporation. max The algorithm has a maximum number of iterations (max iteration) and generates an initial population based on the set initial parameters, as shown in expression (11).
[0145] Then, the fitness function value of each individual in the population is calculated according to expression (12), and the individuals in the population are divided into ocean, river and stream according to their respective fitness function values;
[0146] Then, the number of streams connecting to the river and the ocean is calculated according to expression (14);
[0147] Then, the stream flows into the river and the ocean, the river flows into the ocean, the stream position flowing into the ocean and the river position flowing into the ocean are updated according to expressions (15)-(16), and the river position flowing into the ocean is updated according to expression (17);
[0148] Then, determine whether the fitness value of the updated stream position is better than the fitness value of the updated position of the connected river or ocean. If so, swap their positions. Similarly, compare the updated river position with the updated ocean position. When the river position is better than the ocean position, swap the river and ocean positions as well.
[0149] Then, it determines whether the distance between the river and the ocean, or the distance between the stream and the ocean, is less than the distance threshold set in the i-th iteration. If so, the evaporation condition is triggered, and rainfall is performed according to expression (19);
[0150] Then, update the distance determination threshold according to expression (18);
[0151] Finally, it is determined whether the algorithm meets the preset stopping condition. If it does, the algorithm terminates; otherwise, optimization continues, and the algorithm returns to the step of classifying individuals (or raindrops) into oceans, rivers, and streams based on their fitness function values. This preset stopping condition will be further described below in conjunction with the specific parameter identification process.
[0152] As shown above, the main characteristic of the water cycle algorithm is that after generating the initial population, the individuals in the population are classified for preliminary grading, with the optimal solution being assigned to the ocean. Then, an iterative process is performed, simulating the flow of a stream to a river and then to the ocean, optimizing towards a specific goal, thereby accelerating the search for the optimal solution. Simultaneously, the effects of evaporation and rainfall are introduced to re-expand the population size and avoid getting trapped in local optima. Therefore, the water cycle algorithm converges quickly and is not easily trapped in local optima, and its simple structure results in a relatively low computational load.
[0153] Based on the water cycle algorithm described above in terms of both principle and process, the estimation process of the second state of charge of the battery can enter the intermediate parameter identification stage after the battery model is established, that is, the intermediate parameters α1, α2, and α3 involved in expressions (6)-(9) are identified.
[0154] Figure 4 This illustrates the process of identifying intermediate parameters of a mathematical model based on a water cycle algorithm. Figure 4 In this context, the water cycle algorithm module refers to the module that runs the water cycle algorithm, which is included in the second state-of-charge estimation module. Meanwhile... Figure 4 The diagram also shows a delay module that delays the input value at the current time by one time step to avoid generating an algebraic cycle.
[0155] Combination Figure 4 It is known that in order to obtain the estimated values of the intermediate parameters, at least three physical quantities need to be input into the water cycle algorithm module. It should be explained in detail here that, in the actual calculation process, it is stipulated that the estimated values of the intermediate parameters obtained by the water cycle algorithm at time k-1 are used for the calculation at time k. Therefore, these estimated values of intermediate parameters are recorded as the estimated values of the intermediate parameters α1, α2, and α3 at time k.
[0156] Specifically, at time k-1, in order to solve for the estimated values of the intermediate parameters α1, α2, and α3 at time k, at least three physical quantities E(k-2), I(k-1), and I(k-2) at time k-1 need to be input into the water cycle algorithm module. Among them, the current I(k-1) at time k-1 and the current I(k-2) at time k-2 can be directly measured in the battery, while the specific solution process of E(k-2) will be described in the following steps of calculating the error value in the same way as the solution process of the voltage difference at any other time.
[0157] In addition to the three physical quantities mentioned above, the error value e(k) must also be input to the water cycle algorithm module during the iterative calculation. The error value e(k) is used to measure the accuracy of the intermediate parameter estimation and to control the optimization iterative calculation.
[0158] Specifically, in the process of identifying the intermediate parameters of the mathematical model using the water cycle algorithm, the error value e(k) is calculated using the following three expressions:
[0159]
[0160]
[0161]
[0162] in, These are the estimated values of the three intermediate parameters at time k; E(k-1) refers to the voltage difference at time k-1; I(k) refers to the current measured at time k, and I(k-1) refers to the current measured at time k-1. It is an estimate of the voltage difference at time k; It is the estimated value of the open-circuit voltage at time k; U is the estimated value of the terminal voltage at time k; L e(k) is the terminal voltage at time k, which can be obtained through actual measurement; e(k) is the error value, representing the error between the estimated terminal voltage at time k and the actual measured terminal voltage.
[0163] It should be understood that for iterative calculations at any given time, the current and terminal voltage of the battery at the current and previous times can be considered known quantities because they can be directly measured.
[0164] Reference Figure 4 The process of calculating the error value e(k) by combining expressions (20)-(22) is as follows:
[0165] 1) At time k-1, input the three known quantities E(k-2), I(k-1), and I(k-2) at time k-1, along with the error value e(k-1), into the water cycle algorithm module to obtain the estimated values of the intermediate parameters at time k. Record the estimated values of the intermediate parameters at time k;
[0166] 2) Estimates of the intermediate parameters at time k. and the terminal voltage U measured at time k L (k), calculate the estimated value of the open-circuit voltage for the next time step according to expressions (2)-(9), that is, the estimated value of the open-circuit voltage at time k+1. The estimated value of the open-circuit voltage at time k+1 Combined with the terminal voltage U at time k+1 L (k+1) can be used to calculate the voltage difference E(k+1) at time k+1 (in Figure 4Not shown in the figure, similarly, if the estimated value of the open circuit voltage and the terminal voltage at any time (k) are known, the voltage difference at the current time (k) can be obtained and used for subsequent iterative calculations. At the same time, the estimated value of the open circuit voltage can also be used to calculate the estimated value of the terminal voltage as defined by expression (21).
[0167] 3) At time k, based on the currently known voltage difference E(k-1) at time k-1, current I(k) at time k, current I(k) at time k-1, and the estimated value of the intermediate parameters at time k. The estimated value of the voltage difference at time k is calculated according to expression (20).
[0168] 4) Estimated value of voltage difference at time k And the estimated value of the open-circuit voltage at time k, which has already been calculated at time k-1. Based on expression (21), the estimated value of the terminal voltage at time k is calculated.
[0169] 5) Based on the estimated value of the terminal voltage at time k. The terminal voltage U at time k L (k), according to expression (22), solve for the error value e(k) at time k, and record the error value e(k) at time k.
[0170] After obtaining the error value e(k) at time k, a decision is made to stop the iteration. Here, it is determined whether the stopping condition set based on the error value e(k) at time k is met. The stopping condition can be specifically set as e(k)^2 < 0.001. Of course, the stopping condition is not limited to the above expression; any other suitable form of stopping condition is feasible. As long as it can be determined that the algorithm converges to a predetermined threshold range, the algorithm can be stopped, and the intermediate parameters are considered to have been correctly identified. These intermediate parameters are then output for the next step of the calculation.
[0171] If the stopping condition is not met, the iteration continues, and the error value e(k) calculated at time k is input to the water cycle algorithm module at a delay of one time (therefore denoted as e(k+1)) so as to control the estimation of intermediate parameters at the next time (k+1) based on the error value.
[0172] It should be noted that this parameter identification process involves at least two iterations by default, i.e., k ≥ 2. Of course, when the algorithm has not started iterating or has just begun iterating (i.e., when the battery has just started operating), the initial open-circuit voltage is a known value, which can be obtained from the initial state of charge stored in the corresponding memory of the vehicle and the pre-calibrated open-circuit voltage-state of charge mapping table described below.
[0173] Furthermore, in the iterative calculation, in order to optimize the estimation of intermediate parameters at the next time step based on the error value e(k) at time k, it is necessary to control the iterative calculation of intermediate parameters α1, α2, and α3 in the direction that the error value e(k) gradually decreases. Advantageously, e(k) can be used as an independent variable to specifically define the fitness function in the water cycle algorithm, thereby achieving control over the estimation of intermediate parameters. In particular, the fitness function can be set as follows:
[0174] C n (k) = e(k)^2, n = 1, 2, ..., N pop (twenty three)
[0175] Among them, C n e(k) refers to the fitness function at time k, and e(k) is the error value of the input water cycle algorithm module at time k.
[0176] Of course, depending on the various needs in the actual operation process, C n (k) can also be other functions with e(k) as the independent variable.
[0177] After correctly identifying the three intermediate parameters α1, α2, and α3, equivalent physical parameters that conform to the actual operating conditions of the battery are calculated based on the three identified intermediate parameters α1, α2, and α3. Subsequently, the state of charge (SOC) of the battery can be estimated based on the calculated equivalent physical parameters. This SOC is called the second SOC.
[0178] Figure 5 This illustrates the process of calculating equivalent physical parameters based on identified intermediate parameters, and subsequently estimating the second state of charge (SOC) of the battery based on these equivalent physical parameters. Combined with... Figure 5 Describe the calculation process from intermediate parameters to equivalent physical parameters, and from equivalent physical parameters to the state of charge value.
[0179] The process of calculating the equivalent physical parameters based on the identified intermediate parameters is as follows: According to expressions (6)-(9), the identified intermediate parameters α1, α2, and α3 can be used to calculate the equivalent physical parameters in the first-order RC equivalent circuit model of the battery, namely the ohmic internal resistance R. o Polarization resistance R Th and polarization capacitor C Th .
[0180] The process of estimating the second state of charge of a battery based on equivalent physical parameters is as follows:
[0181] According to expression (2), given the ohmic internal resistance R... o Polarization resistance R Th and polarization capacitor C ThFurthermore, given the actual terminal voltage of the battery, the open-circuit voltage U of the battery model is calculated. oc ;
[0182] After calculating the open-circuit voltage U oc Then, the state of charge (SOC) value of the battery is obtained based on the pre-calibrated open-circuit voltage-state-of-charge (OCC) mapping. This SOC value is called the second SOC value. WCA .
[0183] It should be noted that the pre-calibrated open-circuit voltage-state of charge mapping can be obtained through a limited number of experiments, or statistically derived from the actual application of a large number of batteries, or by any other means known to those skilled in the art. Regardless of how the pre-calibrated open-circuit voltage-state of charge mapping is formed, no subjective or objective error is considered in the mapping.
[0184] Therefore, by establishing an equivalent circuit model of the battery, mathematically transforming the expression of the equivalent circuit model to obtain a mathematical model, identifying the intermediate parameters of the mathematical model using a water cycle algorithm, calculating the equivalent physical parameters based on the identified intermediate parameters, and estimating the second state of charge (SOC) value based on the equivalent physical parameters, the SOC value of the battery can be estimated. WCA .
[0185] The first state of charge (SOC) value was obtained using the battery state of charge estimation method according to the present invention. AC and the second state of charge (SOC) WCA Next, the actual state of charge value is estimated. Now return. Figure 1 To continue describing the calculation of the actual state of charge value.
[0186] Input the first state of charge (SOC) value into the actual state of charge estimation model. AC and the second state of charge (SOC) WCA Then, a summation operation is performed on the actual state-of-charge estimation model according to the following expression:
[0187] SOC R =δ·SOC AC +(1-δ)SOC WCA (twenty four)
[0188] Among them, SOC R δ represents the actual state of charge value, and δ represents the confidence factor.
[0189] The trust factor δ varies with the battery's state of charge (SOC) and / or operating duration, and this variation can be defined according to Table 1 below:
[0190] δ condition 1 The battery's state of charge is ≤20% or the battery's operating duration is <60s. 0.6 20% < state of charge value of the battery ≤ 45% and the continuous operation duration of the battery > 60s 0.4 45%<state of charge value of the battery≤60% and the continuous operation duration of the battery>60s 0.2 60%<state of charge value of the battery ≤ 80% and the continuous operation duration of the battery>60 s 0 The battery's state of charge (SOC) is >80% and the battery's operating duration is >60 seconds.
[0191] Table 1
[0192] As shown in Table 1, the trust factor δ is larger when the battery state of charge value is smaller and the battery operation time is shorter, and smaller when the battery initial state of charge value is larger and the operation time is longer.
[0193] In practice, especially in lithium-ion batteries, the state of charge (SOC) estimate obtained using the ampere-hour integral method deviates significantly from actual conditions during long-term battery operation due to the inherent characteristics of the method, while maintaining a certain level of accuracy in the short term. The water circulation algorithm, on the other hand, estimates the SOC of the battery that increasingly align with the actual operating conditions. However, the water circulation algorithm requires iterative calculations, resulting in inaccurate SOC values with significant initial errors in the early stages of SOC estimation. Therefore, it is necessary to utilize both the ampere-hour integral method and the water circulation algorithm simultaneously to achieve a balance between short-term and long-term accuracy. Thus, a confidence factor δ is used to balance the accuracy of SOC estimation under various battery operating conditions.
[0194] Expression (24) is an empirical formula, used only to illustrate that in the process of solving the estimated value of the actual state of charge according to the estimation method of the present invention, the first state of charge value and the second state of charge value are weighted and summed according to the actual operating conditions of the battery. In fact, the actual state of charge value can be obtained by any other suitable calculation method to make reasonable use of the first state of charge value and the second state of charge value. Similarly, the trust factor δ is not limited to the setting in Table 1, and can be appropriately adjusted according to the actual situation of the battery or the results obtained from experiments. For example, when the battery operating duration is <30s, the trust factor δ is equal to 1.
[0195] In summary, the method for estimating the state of charge (SOC) of a battery, combining the water cycle algorithm and the ampere-hour integration method proposed in this invention, can accurately estimate the SOC of the battery. While the ampere-hour integration method provides a stable estimate of SOC during battery operation due to its simple computation, it is an open-loop algorithm that ignores the variation of battery parameters with physicochemical conditions. Therefore, this estimate is inaccurate, and the error accumulates with increasing battery operating time. The SOC estimation process based on the water cycle algorithm identifies the parameters involved in the battery model online, thus accurately reflecting the actual SOC of the battery after multiple iterations. Therefore, it is necessary to combine these two estimation processes to further accurately estimate the actual SOC of the battery, achieving a balance between short-term and long-term accuracy in estimating the battery's SOC.
[0196] Furthermore, the present invention also relates to a system for estimating the state of charge (SOC) of a battery, the system comprising: a first SOC estimation module, which estimates the first SOC of the battery based on the ampere-hour integration method. AC The second state-of-charge (SOC) estimation module estimates the battery's SOC based on a water circulation algorithm. WCA ; and an actual state of charge (SOC) estimation module, which estimates the SOC of the battery based on the first and second SOC values. R .
[0197] Furthermore, the present invention also relates to an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for estimating the state of charge of a battery as described above.
[0198] Finally, the present invention also relates to a vehicle including a battery, preferably a lithium-ion battery, and the vehicle further including electronic equipment according to the above description.
[0199] The above embodiments are merely illustrative of the present invention, enabling those skilled in the art to understand it, and are not intended to limit the scope of protection of the present invention. Equivalent variations or modifications made based on the principles and design concepts disclosed in this invention all fall within the scope of protection of this invention.
Claims
1. A method for estimating the state of charge (SOC) of a battery, characterized in that, include: The first state of charge of the battery is estimated based on the ampere-hour integral method; The second state of charge value of the battery is estimated based on the water circulation algorithm; as well as The actual state of charge (SOC) of the battery is estimated based on the first SOC and the second SOC. Estimating the second state of charge of the battery based on the water cycle algorithm includes the following steps: An equivalent circuit model of the battery is established, and the equivalent circuit model contains the equivalent physical parameters of the battery; The expression of the equivalent circuit model is discretized after undergoing a Laplace transformation to obtain the mathematical model. The water cycle algorithm is used to identify intermediate parameters in the mathematical model; The equivalent physical parameters are calculated based on the identified intermediate parameters; The second state of charge value of the battery is estimated based on the equivalent physical parameters; The mathematical model is expressed as follows: Where k is the discrete time, I(k) is the discretized current, E(k) is the discretized voltage difference, representing the voltage difference between the open-circuit voltage and the terminal voltage, and U... oc (k) is the discretized open-circuit voltage, U L (k) represents the discretized terminal voltage, α1, α2, and α3 are intermediate parameters of the mathematical model, and R o R is the ohmic internal resistance of the battery. Th C is the polarization internal resistance of the battery. Th Let be the polarization capacitance of the battery, and T be the sampling period.
2. The method for estimating the state of charge (SOC) of a battery according to claim 1, characterized in that, The expression for estimating the first state of charge of the battery based on the ampere-hour integral method is as follows: Among them, SOC AC Here, SOC0 is the first state of charge (SOC) of the battery, and C is the initial state of charge (SOC) of the battery. N Let i be the capacity of the battery, i(t) be the current of the battery, t0 be the initial time, and t1 be the estimated time.
3. The method for estimating the state of charge (SOC) of a battery according to claim 1, characterized in that, The equivalent circuit model expression of the battery is as follows: Among them, U oc R is the open-circuit voltage of the battery. o R is the ohmic internal resistance of the battery, I is the current of the battery, and R is the internal resistance of the battery. Th C is the polarization internal resistance of the battery. Th U is the polarization capacitance of the battery. Th U is the polarization voltage of the battery. L The terminal voltage of the battery is denoted as .
4. The method for estimating the state of charge (SOC) of a battery according to claim 1, characterized in that, Identifying intermediate parameters in the mathematical model using the water cycle algorithm includes the following steps: The intermediate parameters at time k are estimated using the water cycle algorithm module. The error value e(k) is calculated using the following expression: Where is the estimated value of the intermediate parameters at time k, E(k-1) refers to the voltage difference at time k-1, I(k) refers to the current at time k, and I(k-1) refers to the current at time k-1. It is an estimate of the voltage difference at time k. It is the estimated value of the open-circuit voltage at time k. U is the estimated value of the terminal voltage at time k. L e(k) is the terminal voltage at time k, and e(k) is the error value at time k. Determine whether the stopping condition set based on the error value e(k) at time k is met; If the stopping condition is not met, the iteration continues, and the estimation of the intermediate parameter at the next time step is controlled according to the error value e(k) at time k; if the stopping condition is met, the intermediate parameter is output.
5. The method for estimating the state of charge (SOC) of a battery according to claim 1, characterized in that, Estimating the second state of charge value of the battery based on the equivalent physical parameters includes: The open-circuit voltage of the battery is calculated using the equivalent circuit model based on the equivalent physical parameters. Based on the calculated open-circuit voltage value, the second state of charge value of the battery is obtained according to the pre-calibrated open-circuit voltage-state of charge mapping.
6. The method for estimating the state of charge (SOC) of a battery according to claim 1, characterized in that, The expression for estimating the actual state of charge (SOC) of the battery based on the first SOC and the second SOC is as follows: Among them, SOC R The SOC represents the actual state of charge (SOC) of the battery. AC The first state of charge (SOC) of the battery. WCA The second state of charge value of the battery is δ, which represents the trust factor, and the trust factor δ varies according to the state of charge value of the battery and / or the operating duration of the battery.
7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method for estimating the state of charge of a battery as described in any one of claims 1-6.
8. A vehicle comprising a battery, characterized in that, The vehicle includes the electronic equipment according to claim 7.
Citation Information
Patent Citations
Method and device for estimating state of charge of battery and computer readable storage medium
CN109613432A
Battery state of charge estimation method and battery state of health estimation method
CN110286324A
