A traceless Kalman SOC estimation method for hybrid power system battery

Abstract

The present invention discloses a hybrid power system battery untraceable Kalman SOC estimation method, the steps are as follows: establish an equivalent circuit model based on fractional order theory; conduct capacity test of lithium iron phosphate battery of hybrid UAV under various working conditions; conduct open circuit voltage hysteresis characteristic test of lithium iron phosphate battery and establish a simplified open circuit voltage hysteresis model; conduct parameter identification test and use optimization algorithm for parameter identification; fractional order untraceable Kalman filter SOC estimation. The present invention can better simulate the battery terminal voltage and improve the accuracy of SOC estimation under the condition of using fewer parameters and simpler hysteresis model.

Description

Technical Field

[0001] The invention belongs to the field of unmanned aerial vehicle energy management, and in particular relates to a method for estimating the state of charge (SOC) of a battery in a hybrid power system. Background Art

[0002] The power battery SOC is a common indicator for measuring the remaining battery power. Its estimation method is an important research topic in the field of hybrid UAV energy management. Its accuracy directly affects the effectiveness of the energy management system and the power control system. It is also an important basis for ensuring and extending the health of the UAV power battery. However, there are a lot of system noise interference in the operation of hybrid UAVs, and the current measurement is often not accurate enough, which makes the commonly used ampere-hour integration method have error accumulation and is limited by algorithm defects. The error cannot be eliminated to make the algorithm converge. Therefore, it is often used in engineering in combination with the open circuit voltage method. However, the accurate open circuit voltage of the power battery needs to be obtained by static. Usually, for the convenience of use, the accuracy of the open circuit voltage can only be sacrificed. It is approximately assumed that the battery terminal voltage under low current state is equal to the open circuit voltage. The above problems can be solved to a certain extent by simulating the battery terminal voltage characteristics through the battery equivalent circuit model, measuring the actual terminal voltage and current of the battery, and combining Kalman filtering and its various forms to estimate SOC. However, the accuracy of the algorithm depends to a large extent on the accuracy of the equivalent circuit model. However, for UAV carriers, there are many factors that affect the parameters of the equivalent circuit model, and incompleteness will greatly reduce the accuracy of the model terminal voltage. Considering that for lithium iron phosphate batteries, their open circuit voltage, as one of the most important parts of the equivalent circuit model, has a hysteresis characteristic, which is manifested as the open circuit voltage in the charging state is greater than the discharge state, and when the charge and discharge states are switched, the open circuit voltage changes to the corresponding state after a period of time.

[0003] At present, the commonly used integer-order equivalent circuit models cannot simulate the battery terminal voltage characteristics well when the order is low, but the system state dimension and the filter calculation complexity will increase with the increase of the order. In recent years, the equivalent circuit model based on fractional-order theory has attracted widespread attention due to its advantage of being able to better simulate the battery charging and discharging state and the resistance and capacitance characteristics of the static state with a smaller model order.

[0004] Therefore, based on the shortcomings and deficiencies of the current SOC estimation methods, it is particularly important to establish a traceless Kalman SOC estimation method for hybrid power system batteries. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a hybrid system battery traceless Kalman SOC estimation method to solve the deficiencies in the existing SOC estimation methods in hybrid UAVs; the present invention can better simulate the battery terminal voltage and improve the accuracy of SOC estimation by using fewer parameters and a simpler hysteresis model.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A hybrid power system battery traceless Kalman SOC estimation method of the present invention comprises the following steps:

[0008] 1) Establish an equivalent circuit model based on fractional-order theory;

[0009] 2) Conduct capacity tests of lithium iron phosphate batteries for hybrid UAVs under various operating conditions;

[0010] 3) Conduct open circuit voltage hysteresis characteristic test of lithium iron phosphate battery and establish a simplified open circuit voltage hysteresis model;

[0011] 4) Conduct parameter identification tests and use optimization algorithms for parameter identification;

[0012] 5) Fractional-order unscented Kalman filter SOC estimation.

[0013] Furthermore, the step 1) specifically includes: based on a first-order integer RC model, establishing a first-order fractional equivalent circuit model including internal resistance, polarization resistance, CPE fractional-order elements and corresponding orders, and deriving a corresponding mathematical model based on the selected model.

[0014] Furthermore, the step 2) specifically includes: setting the ambient temperature to 25°C and the air pressure to 100Kpa, charging the battery to the charging cut-off voltage at a C / 3 current, and then charging at a constant voltage until the current drops to 0.05C, and standing for 1 hour for the battery to return to a balanced state; using a high-altitude environment simulation box, at the selected temperature and air pressure nodes, constant current discharge is performed at the selected discharge rate until the discharge cut-off voltage, and the discharged capacity is recorded; this cycle is repeated until all temperature nodes, air pressure nodes, and discharge rate nodes are completed, and the battery discharge capacity is considered equal to the battery charging capacity.

[0015] Further, the step 3) specifically includes: selecting the average value of the hybrid UAV flight environment temperature -20°C and the average value of the air pressure 50Kpa as the test environment conditions, performing the open circuit voltage test of the lithium iron phosphate battery in the charging state and the open circuit voltage test in the discharging state, and obtaining the battery open circuit voltage hysteresis characteristic curve; the open circuit voltage data processing method is as follows:

[0016] U OCV =(U OCV_c-U OCV_d )*0.2+U OCV_d

[0017] Where U OCV_c U is the open circuit voltage in the charging state; OCV_d is the open circuit voltage in the discharge state;

[0018] According to the processed data, an 8th-order polynomial was used to fit it, and a simplified open-circuit voltage model considering the hysteresis characteristics of lithium iron phosphate was obtained.

[0019] Furthermore, the step 4) specifically includes: selecting SOC=90%, 50% and 10% as test nodes respectively, and based on the test temperature and air pressure nodes in step 2), using the above different node combinations as test conditions, and performing double pulse parameter identification tests under charge and discharge states respectively, and using the obtained test end voltage data using a genetic algorithm to perform parameter identification, and obtaining fractional-order equivalent circuit model parameters under different temperatures, air pressures, SOCs and charge and discharge states.

[0020] Furthermore, the step 5) specifically includes: establishing an unscented Kalman filter algorithm based on a fractional-order equivalent circuit model, and estimating the SOC of the hybrid UAV based on the results of capacity tests, parameter identification tests, and a simplified open-circuit voltage hysteresis model.

[0021] Beneficial effects of the present invention:

[0022] The present invention establishes a fractional-order unscented Kalman filtering algorithm that takes into account temperature, air pressure and battery hysteresis characteristics, realizes dynamic estimation of SOC, improves the terminal voltage accuracy of the equivalent circuit model, and also greatly improves the accuracy of SOC estimation when the hybrid UAV is flying at high altitude. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Flow chart of the method of the present invention.

[0024] Figure 2 It is the circuit diagram of the fractional-order equivalent circuit model.

[0025] Figure 3 It is a schematic diagram of the connection of the test bench in the embodiment of the invention.

[0026] Figure 4 This is the genetic algorithm flow chart. DETAILED DESCRIPTION

[0027] In order to facilitate the understanding of those skilled in the art, the present invention is further described below in conjunction with embodiments and drawings. The contents mentioned in the implementation modes are not intended to limit the present invention.

[0028] Reference Figure 1As shown, a hybrid system battery traceless Kalman SOC estimation method of the present invention comprises the following steps:

[0029] Step 1: Establish an equivalent circuit model based on fractional order theory;

[0030] Choose GL type as the fractional order definition:

[0031]

[0032]

[0033] Where h is the sampling interval, α is the fractional order, i is the historical data sequence, and t is time.

[0034] Establish as Figure 2 The first-order fractional equivalent circuit model shown in the figure, U OCV is the open circuit voltage of the battery, R0 is the internal resistance of the battery, U0 is the internal resistance terminal voltage, R1 is the polarization internal resistance, the capacitance of the CPE element is C1, the order is α1, U1 is the polarization effect terminal voltage, which affects the terminal voltage change trend during battery charging and discharging and static state, U t is the battery output voltage. The discretized state space expression of the first-order fractional equivalent circuit model is:

[0035]

[0036] Where x(k) = [SOC(k), U1(k)] T ; y(k) = [Ut]; Among them, N cut is the number of data truncation, which can greatly reduce the amount of system calculation while ensuring the accuracy of fractional-order calculation; D(k)=[-R0]; f[x(k)]=U OCV [SOC(k)]-U1(k) is the system nonlinear term; ω(k) is the system noise with covariance Q; υ(k) is the measurement noise with covariance R; Q N is the capacity measured under different conditions; I(k) is the input current; k is a discrete data sequence.

[0037] Step 2: Conduct capacity test of lithium iron phosphate battery of hybrid UAV under various working conditions;

[0038] The test bench used Figure 3As shown, it includes: a host computer, a charge and discharge test platform, and a high-altitude environmental test chamber. The host computer can control the charge and discharge mode and record the test data. The charge and discharge test platform can perform charge and discharge tests and voltage and current measurements on the battery. The high-altitude environmental test chamber can put the battery in a low-temperature and low-pressure environment to simulate the battery working environment when the hybrid UAV is flying. The test method is as follows: When the hybrid UAV is flying normally, its ambient temperature is about -50℃, and the highest temperature is at low altitude and ground state, and the highest temperature is 35℃. The lithium iron phosphate battery used in hybrid UAV is generally a low-temperature battery. When the ambient temperature is high, the battery capacity difference is small. Therefore, in order to reduce the test volume, the number of test nodes in the high-temperature environment can be reduced. Therefore, the test temperature nodes are selected as 35℃, 5℃, -5℃, -15℃, -25℃, -35℃, and -50℃; the highest air pressure of the hybrid UAV is 100Kpa of ground pressure, and the lowest air pressure is 25Kpa, which can cover the 10,000km high-altitude environment. The air pressure nodes are divided into intervals of 10Kpa. The maximum discharge rate of the battery is generally 4-5C, and the battery's power attenuation is more serious when it is discharged at a large rate, which is not conducive to the battery's cycle life. Therefore, large-rate discharge is rare, and the discharge rate intervals are divided into 0.1C, 0.3C, 0.5C, 1C, 2C, 3C, and 4C. In order to reduce the workload of the capacity test, the charging power is equal to the discharging power under the same conditions. The specific test steps are as follows: Before each capacity test, set the ambient temperature to 25°C and the air pressure to 100Kpa, charge the battery to the charging cut-off voltage at a C / 3 current, and then charge it at a constant voltage until the current drops to 0.05C, and let it stand for 1 hour until the battery returns to a balanced state. At the selected temperature and pressure nodes, perform constant current discharge at the selected discharge rate until the discharge cut-off voltage, record the discharged capacity, and after standing for 1 hour, change the environment to a temperature of 25°C and an air pressure of 100Kpa for constant current and constant voltage charging. After standing for 1 hour, change the environment and discharge conditions for discharge, and repeat this cycle until all temperature nodes, air pressure nodes, and discharge rate nodes are completed.

[0039] Step 3: Conduct an open circuit voltage hysteresis characteristic test on a lithium iron phosphate battery and establish a simplified open circuit voltage hysteresis model;

[0040] The open circuit voltage of lithium iron phosphate battery changes little with the external environment, but there is a hysteresis effect. In this example, the average temperature of the hybrid UAV flight environment is -20℃ and the average air pressure is 50Kpa as the test environment conditions, and the open circuit voltage test of the lithium iron phosphate battery in the charging state and the open circuit voltage test in the discharging state are carried out to obtain the battery open circuit voltage hysteresis characteristic curve. The specific test method is as follows:

[0041] The test battery is charged at a C / 3 constant current to the charging cut-off voltage, and then charged at a constant voltage until the current drops to 0.05C. At this time, the battery is fully charged (SOC=100%). After standing for 1 hour, the terminal voltage is recorded as the open circuit voltage in the discharge state of SOC=100%. Then, the C / 3 constant current is discharged until the SOC drops by 5%. After standing for 1 hour, the voltage is recorded as the open circuit voltage in the discharge state when SOC=95%. This cycle is repeated, each time discharging 5% SOC until SOC=0%, and the voltage is recorded when standing for 1 hour after each discharge. The test is terminated and the open circuit voltage value corresponding to each SOC point in the discharge state is obtained. The battery is charged with C / 3 constant current to the charging cut-off voltage, and then charged with constant voltage until the current drops to 0.05C. After standing for 1 hour, C / 3 constant current is discharged to the discharge cut-off voltage. After standing for 1 hour, the terminal voltage is recorded. Then, C / 3 constant current is charged until the SOC rises by 5%. After standing for 1 hour, the terminal voltage data is recorded as the open circuit voltage. This cycle is repeated until SOC=95%. After standing for 1 hour, the open circuit voltage is obtained. Finally, the battery is charged with C / 3 constant current to the charging cut-off voltage and then charged with constant voltage to 0.05C. After standing for 1 hour, the open circuit voltage value in the fully charged state is obtained, and the test is completed. The data is sorted to obtain the open circuit voltage value corresponding to each SOC point in the charging state.

[0042] Considering that the lithium iron phosphate battery of the hybrid UAV is in a discharge state most of the time during flight, in order to take into account the hysteresis characteristics of the open circuit voltage and make it biased towards the discharge state, the open circuit voltage data processing method is as follows:

[0043] U OCV =(U OCV_c -U OCV_d )*0.2+U OCV_d

[0044] Where U OCV_c U is the open circuit voltage in the charging state; OCV_d It is the open circuit voltage in the discharge state.

[0045] Then, an 8th-order polynomial was used to fit it, and a simplified open-circuit voltage model considering the hysteresis characteristics of lithium iron phosphate was obtained.

[0046] Step 4: Conduct parameter identification test and use optimization algorithm to identify parameters;

[0047] In order to ensure the cycle life of the hybrid UAV lithium iron phosphate battery, most of the time it works in the range of SOC = 10% to 90%, so SOC = 90%, 50% and 10% are selected as test nodes, and according to the test temperature and air pressure nodes in step 2, the above different node combinations are used as test conditions to carry out double pulse parameter identification tests under charge and discharge states. The specific test methods are as follows:

[0048] Discharge state double pulse parameter identification test: First, the battery is charged with C / 3 constant current to the charging cut-off voltage under 25℃ and 100Kpa, and then charged with constant voltage until the current drops to 0.05C, which is the full charge state. After standing for 1h, adjust the environment to the selected temperature and pressure node, discharge with C / 3 constant current to SOC=90%, stand for 1h and perform double pulse discharge (discharge with 2C for 10s constant current, stand for 40s and then discharge with 2C for 10s constant current, and then stand for 40s), then stand for 1h and discharge with C / 3 constant current to SOC=50%, stand for 1h and perform double pulse discharge, stand for 1h and then discharge with C / 3 constant current to SOC=10%, stand for 1h and then perform double pulse discharge, and then extract the battery terminal voltage data of three double pulse tests as the parameter identification test data under this temperature and pressure node. In this cycle, the discharge state double pulse parameter identification test data of all temperature and pressure node combinations are obtained.

[0049] Double pulse parameter identification test of charging state: First, discharge the battery with C / 3 constant current to the discharge cut-off voltage under the conditions of 25℃ and 100Kpa, and after standing for 1h, adjust the environment to the selected temperature and pressure node, charge with C / 3 constant current to SOC=10%, stand for 1h and perform double pulse charging (charge with 2C constant current for 10s, stand for 40s and then charge with 2C constant current for 10s, and then stand for 40s), then stand for 1h and charge with C / 3 constant current to SOC=50%, stand for 1h and perform double pulse charging, stand for 1h and then charge with C / 3 constant current to SOC=90%, stand for 1h and perform double pulse charging, and the test ends. Extract the battery terminal voltage data of three double pulse tests as the parameter identification test data under this temperature and pressure node. In this cycle, the double pulse parameter identification test data of charging state for all temperature and pressure node combinations are obtained.

[0050] The genetic algorithm is used to identify the parameters of all double pulse parameter identification test data, and the parameters of the fractional-order equivalent circuit model under different temperatures, air pressures, SOCs and charge and discharge states are obtained. The genetic algorithm process is as follows: Figure 4 As shown, its fitness function and calculation method are:

[0051]

[0052] U m (k)=U(k)+U OCV (SOC(k))

[0053] Among them, U m (k) is the terminal voltage of the fractional order model, U r (k) is the voltage at the measuring end.

[0054] Step 5: Fractional-order unscented Kalman filter SOC estimation;

[0055] An unscented Kalman filter algorithm based on a fractional-order equivalent circuit model is established. Based on the results of capacity tests, parameter identification tests, and an open-circuit voltage model considering hysteresis characteristics, the SOC of a hybrid UAV is estimated. The specific calculation method is as follows:

[0056] 51. Get Sigma sampling point:

[0057]

[0058]

[0059]

[0060]

[0061]

[0062] Among them, J(1) and J(2) represent the first and second columns of the J matrix. L is the state dimension, L=2; λ is the scaling parameter.

[0063] 52. One-step prediction of Sigma point set state:

[0064]

[0065] 53. Calculate the state one-step prediction and covariance:

[0066] State one-step prediction:

[0067]

[0068] One-step covariance forecast:

[0069]

[0070] The corresponding weight of the sampling point is:

[0071]

[0072]

[0073] α and β are parameters to be selected, which control the distribution of sampling points.

[0074] 54. Calculate the observed quantity:

[0075] The one-step state prediction is transformed again into:

[0076]

[0077] in,

[0078] The Sigma point of the observed quantity is:

[0079]

[0080] The observed quantity is:

[0081]

[0082] 55. Kalman gain:

[0083] K(k+1)=PzzPxz -1

[0084] The observed value covariance is:

[0085]

[0086] The covariance between the state and the observation is:

[0087]

[0088] Where Y(k+1) is the measured voltage at time k+1.

[0089] 56. System status and covariance update:

[0090] The difference between the measured value and the observed value is:

[0091]

[0092] The system status is updated as follows:

[0093]

[0094] The covariance is updated as:

[0095] P(k+1)=P(k+1|k)-K(k+1)P zz K(k+1) T .

[0096] The present invention has many specific application paths. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements can be made without departing from the principle of the present invention. These improvements should also be regarded as the protection scope of the present invention.

Claims

1. A hybrid power system battery traceless Kalman SOC estimation method, characterized in that: Here are the steps: 1) Establish an equivalent circuit model based on fractional-order theory; 2) Conduct capacity tests of lithium iron phosphate batteries for hybrid UAVs under various operating conditions; 3) Conduct open circuit voltage hysteresis characteristic test of lithium iron phosphate battery and establish a simplified open circuit voltage hysteresis model; 4) Conduct parameter identification tests and use optimization algorithms for parameter identification; 5) Fractional-order unscented Kalman filter SOC estimation; The step 2) specifically includes: setting the ambient temperature to 25°C and the air pressure to 100Kpa, charging the battery to the charging cut-off voltage at a C / 3 current, and then charging at a constant voltage until the current drops to 0.05C, and standing for 1 hour until the battery returns to a balanced state; using a high-altitude environment simulation box, at the selected temperature and air pressure nodes, performing constant current discharge at the selected discharge rate until the discharge cut-off voltage, and recording the discharged capacity; and repeating this cycle until all temperature nodes, air pressure nodes, and discharge rate nodes are completed, and the battery discharge capacity is considered equal to the battery charging capacity; The method for processing the open circuit voltage data in step 3) is as follows: IN OCV =(U OCV_c -IN OCV_d )*0.2+U OCV_d Where U OCV_c U is the open circuit voltage in the charging state; OCV_d is the open circuit voltage in the discharge state; The step 4) specifically includes: selecting SOC=90%, 50% and 10% as test nodes respectively, and according to the test temperature and air pressure nodes in step 2), using the above different node combinations as test conditions, respectively performing double pulse parameter identification tests under charge and discharge states, and using the obtained test terminal voltage data to perform parameter identification using a genetic algorithm to obtain the fractional-order equivalent circuit model parameters under different temperatures, air pressures, SOCs and charge and discharge states; The step 5) specifically includes: establishing an unscented Kalman filter algorithm based on a fractional-order equivalent circuit model, and estimating the SOC of the hybrid UAV based on the results of capacity tests, parameter identification tests, and a simplified open-circuit voltage hysteresis model.

2. The hybrid system battery traceless Kalman SOC estimation method according to claim 1, characterized in that: The step 1) specifically includes: based on a first-order integer RC model, establishing a first-order fractional equivalent circuit model including internal resistance, polarization resistance, CPE fractional-order elements and corresponding orders, and deriving a corresponding mathematical model based on the selected model.

Images