A kind of state-of-charge evaluation method based on optimizing initial value
Technical field
The invention belongs to automobile power cell technical field, relate in particular to the method for automobile power cell SOC estimation.
Background technology
SOC is one of important parameter of battery, refers to the dump energy of battery, and the running status that SOC is battery provides important evidence.But SOC is affected by several factors, its value can only be estimated by measuring these factor values.
The method of estimation of electrokinetic cell SOC mainly contains discharge test method, open-circuit voltage method, Ah counting method, internal resistance measurement method, Kalman filtering method, load method and neural network etc. both at home and abroad at present.The most frequently used: open-circuit voltage method, Ah counting method, Kalman filtering method.
Open-circuit voltage method: determine SOC size by the funtcional relationship of open circuit electromotive force and SOC.In electric discharge latter stage, the effect of test is better, but uses and to have a recovery stage later at battery, at the corresponding relation of this stage voltage and SOC, is not very obvious.
Ah counting method: Ah counting method is to estimate the SOC of battery by calculating the accumulation electric weight of battery when the charge or discharge, but inaccurate current measurement will increase SOC evaluated error, and through accumulating for a long time, it is increasing that this error can become.
Kalman filtering method: Kalman filtering method is a kind of minimum variance estimate method of recursion, utilized the value in a upper moment and the parameter value of measurement in real time to estimate.When Kalman filtering method is estimated for electrokinetic cell SOC, be a discrete system, need an accurate model, and calculated amount be very large.In the research of electrokinetic cell SOC, there is the correction algorithm of a lot of Kalman filtering methods, these algorithms have been optimized estimation precision to a great extent.
Because Kalman filtering method has error correcting action, so the SOC algorithm of great majority based on Kalman filtering all tends to its correction factor improve or use in conjunction with other algorithms and use, and ignore the precise decreasing that initial value error is brought.The SOC algorithm (application number 201110142292.9) of speeding scientific and technological such as Zhenjiang perseverance is exactly to carry out Integrated using by Ah counting method and Kalman filtering method, to improve its robustness, convergence, and adaptivity.In the patent of LG, (application number 200680016312.5) revised efficiency factor.But in the arranging of SOC initial value, or use experience method is weighted estimation.Its precision is not very high.
Summary of the invention
The object of the present invention is to provide a kind of state-of-charge evaluation method based on optimizing initial value, comprise the steps:
S1, the parameters initialization to battery;
S2, estimate battery in a upper operational phase because the quantity of electric charge of polarization effect accumulation;
S3, battery is is repeatedly discharged and recharged, then carry out standing, and detect battery described discharge and recharge standing process in magnitude of voltage, the time t (off) that recording voltage is stable;
S4, t described in S3 (off) is carried out to linearization process;
S5, draw initial value
The quantity of electric charge computing method of accumulating described in S2 further, are as follows:
Under steady temperature:
Charging: power consumption=SOC increment+stored charge amount
?
by Δ SOC corresponding to different current values, draw
by matlab to different current values and α corresponding to different current value
1(I) carry out matching, draw α
1(I)=p
1i
4+ p
2i
3+ p
3i
2+ p
4i
1+ p
5, α
1(I) be that charge efficiency is about the coefficient of electric current;
Electric discharge: SOC reduction=power consumption+stored charge amount
?
by Δ SOC corresponding to different current values, draw
by matlab to different current values and α corresponding to different current value
2(I) carry out matching, draw α
2(I)=p
1i
4+ p
2i
3+ p
3i
2+ p
4i
1+ p
5, α
2(I) be discharging efficiency about the coefficient of electric current,
Wherein
for emitting the quantity of electric charge, Δ SOC is SOC reduction, and Q (up) is stored charge amount, p
1i
4for being matching parameter value out, current i is independent variable, and p is exactly the coefficient of independent variable,
Because η=α (I) β (T), so, in the charging stage
at discharge regime
wherein, η is the function about electric current I and temperature T.
Further, Δ t≤t (off).
The invention has the beneficial effects as follows: within release time, use battery, can provide a comparatively accurate SOC initial value to battery, make more rapid convergence of SOC value that EKF method estimates.
Accompanying drawing explanation
Fig. 1 is evaluation method process flow diagram of the present invention.
Fig. 2 is model of the present invention.
When Fig. 3 is charging, efficiency is about the figure of electric current
Initial value estimation figure when Fig. 4 is charging.
Fig. 5 be under discharge scenario α about the function of I.
Wherein, E is standard voltage source, and R is ohmic internal resistance, and U (up) is polarizing voltage.
Embodiment
Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described:
As shown in Fig. 2 model, set up Kalman filtering algorithm.Wherein, E is standard voltage source, and R is ohmic internal resistance, and U (up) is polarizing voltage.
S1, the parameters initialization to battery;
S2, estimate battery in a upper operational phase because the quantity of electric charge of polarization effect accumulation comprises:
Under steady temperature:
Charging: power consumption=SOC increment+stored charge amount
?
by Δ SOC corresponding to different current values, draw
by matlab to different current values and α corresponding to different current value
1(I) carry out matching, draw α
1(I)=p
1i
4+ p
2i
3+ p
3i
2+ p
4i
1+ p
5, α
1(I) be that charge efficiency is about the coefficient of electric current;
Electric discharge: SOC reduction=power consumption+stored charge amount
?
by Δ SOC corresponding to different current values, draw
by matlab to different current values and α corresponding to different current value
2(I) carry out matching, draw α
2(I)=p
1i
4+ p
2i
3+ p
3i
2+ p
4i
1+ p
5, α
2(I) be discharging efficiency about the coefficient of electric current,
Wherein
for emitting the quantity of electric charge, Δ SOC is SOC reduction, and Q (up) is stored charge amount, p
1i
4for being matching parameter value out, current i is independent variable, and p is exactly the coefficient of independent variable,
Because η=α (I) β (T), so, in the charging stage
at discharge regime
wherein, η is the function about electric current I and temperature T.
S3, battery is is repeatedly discharged and recharged, then carry out standing, and detect battery described discharge and recharge standing process in magnitude of voltage, the time t (off) that recording voltage is stable;
S4, t described in S3 (off) is carried out to linearization process;
S5, draw initial value
Δt≤t(off);
S6, by initial value SOC (t described in S5
0+ Δ t) for expanded Kalman filtration algorithm model and carry out SOC estimation.
According to matlab models fitting, show that under discharge scenario α is about the function of electric current I:
F (x)=p1*i^4+p2*i^3+p3*i^2+p4*i+p5, wherein:
p1=-5.23e+011(-5.469e+011,-4.992e+011)
p2=7.233e+011(6.904e+011,7.563e+011)
p3=-3.751e+011(-3.922e+011,-3.58e+011)
p4=8.647e+010(8.253e+010,9.041e+010)
p5=-7.474e+009(-7.815e+009,-7.133e+009)
According to matlab models fitting, show that under charge condition α is about the function of electric current I:
F (x)=p1*x^4+p2*x^3+p3*x^2+p4*x+p5, wherein:
p1=-0.09137(-0.4133,0.2306)
p2=0.979(-2.161,4.119)
p3=-3.527(-13.52,6.468)
p4=4.833(-7.766,17.43)
P5=-1.144 (6.546,4.258), wherein, because fitting function selection is 4 rank, so most significant digit is 4 rank, also can use 5 rank, 6 rank, but conventionally use 4 rank.