Summary of the invention
The above problem existing for the identification method and system of aging and service life for battery in the prior art, the present invention mention
Cell decay is quickly obtained by a neural network model for a kind of lithium battery degeneration discrimination method and degeneration alarm system
Index to identify cell degradation degree, and reaches threshold value when degenerating, alarms.
Technical scheme is as follows:
A kind of lithium battery degeneration discrimination method is made several with reference to physical quantity using the analysis method of array dispersion degree
For an array, dispersion degree identification is carried out to the array, the physical quantity includes the open-circuit voltage of battery and the environment of battery
Temperature, the discharge-rate of battery, the capacity that can release.
Further, in application dispersion analysis, consider the coefficient of variation for participating in data, carried out each such as with reference to physical quantity
Lower variation:
Time frame variation coefficient=actual discharge time/0.2C multiplying power lower discharge time;
The capacity coefficient of variation=actual capacity/nominal capacity;
The open-circuit voltage coefficient of variation=practical open-circuit voltage/voltage rating;
Temperature variations coefficient=actual temperature/20 DEG C;
On this basis, the effect according to each physical quantity played in degeneration, multiplied by corresponding coefficient.
Further, it establishes neural network model and carries out degeneration factor estimation, include the following steps:
Step 1: establishing neural network model;
Fuzzy inference system in the neural network model uses Sugeno pattern fuzzy model, establishes according to experimental data
FIS;According to lithium ion battery external behavior Parameter analysis, discharge time t is filtered outIt puts, release capacity QIt puts, open-circuit voltage Uk, ring
Input of the border temperature T as neural network model, the output of model are the degree of degeneration characterising parameter of lithium ion battery, that is, are degenerated
Factor beta:
β=f (tIt puts, QIt puts, Uk, T)
In formula, tIt putsIndicate discharge time;QIt putsIt indicates to release capacity;UkIndicate open-circuit voltage;T indicates environment temperature;β be
Numerical value between 0~1, when β is close to 0, then cell degradation lesser extent;When β is close to 1, then cell degradation degree is serious.
Step 2: model training and simulating, verifying;
Experimental data is divided into two groups, i.e. training group and check groups, uses training data to input as model training, with training
System model is arranged trained step-length and is trained, builds simulation model;Using Data Processing in Experiment result as model
Input, respectively obtains the degeneration factor of corresponding date lithium ion battery.
A kind of lithium battery degeneration alarm system, including host computer, communication interface and slave computer, system is by detection device to work
The battery or the electric current of battery pack, voltage, temperature of work are detected, and are calculated according to electric discharge average current and discharge time accumulation
Open-circuit voltage, environment temperature before releasing capacity, electric discharge;It send host computer to carry out degeneration factor calculating, calculated value is sent back to down
Position machine issues alarm if degeneration factor is more than threshold value.
Further, the peripheral circuit of slave computer includes current detecting unit, voltage detection unit, A/D converting unit, temperature
Detection unit, clock acquisition unit, serial communication unit, power supply unit, display unit, alarm unit;The current detecting list
The current information of member acquisition and the information of voltage of voltage detection unit detection are sent to slave computer by A/D converting unit, institute
The temporal information of the temperature information and the acquisition of clock acquisition unit of stating temperature detecting unit detection is sent to slave computer, the serial ports
Communication unit and power supply unit are not connected to slave computer, and the information of slave computer is respectively sent to display unit and alarm unit.
Further, including current detecting is calculated with calculation of capacity and communication with degeneration factor;
The current detecting includes the following steps: with calculation of capacity
The first step, current detecting subprogram start;
Second step returns to upper level main program when electric current is more than the upper limit;When electric current is less than the upper limit, electricity is carried out
Cumulative calculation returns to upper level main program after calculating;
The communication includes the following steps: with degeneration factor calculating
The first step, initialization of (a) serial ports;
Second step sends data acquisition command to slave computer;
Third step, if data receiver terminates, data loading, the degradation model program of Calling MATLAB judges to degenerate
Whether coefficient is not less than threshold value, if it is threshold value is greater than or equal to, then sends alarm command to slave computer, if not being greater than or
Equal to threshold value, then second step is back to;If still receiving data, data are continued to;
4th step returns to second step, continues to data until communication is finished with degeneration factor calculating.
Beneficial effects of the present invention are as follows:
The present invention sets up battery and declines by establishing the degradation function model of lithium battery using neural network and discrete theory
Subtract coefficient.By the analysis to battery relevant parameter, the degeneration factor of battery is calculated, to analyze cell degradation and aging
Degree.For the quick-charging circuit of battery, equalizing circuit and management system evaluation provide quickly evaluation and identification, also for
The cascade utilization of battery, lifetime limitation replacement provide foundation.The present invention is suitble to the model insertion to other electricity existing at present
In the management system of pond, to evaluate battery status and prediction aging and alarm to avoid user's loss.
Specific embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
The basic ideas of design of the invention are as follows:
Cell degradation degree State of Degeneration (english abbreviation SOD) usually can be from the open circuit electricity of battery
The parametric synthesis such as pressure and the environment temperature of battery, the discharge-rate of battery, the capacity that can release analyze and determine.But it is above several
The kind mutual coupling of factor is very strong, and is difficult to remove.Therefore being considered as adaptive neural network fuzzy system establishes them
Relationship between SOD.
In view of corresponding to a battery completely filled, had under certain open-circuit voltage, environment temperature and discharge-rate
It is corresponding to release capacity, and this releasing capacity can embody the degree of degeneration of battery, it may be assumed that the capacity of releasing is more, cell degradation
Degree is lighter;Conversely, cell degradation is more serious.
The present invention uses the analysis method of array dispersion degree, using above each parameter physical quantity as an array,
Under certain condition, dispersion degree identification is carried out to the array.Such as: when environment temperature increases, then correspond to full charge pond its
Open-circuit voltage is higher, under certain discharge-rate, it will keeps discharge time longer, it is more to release capacity.It is namely relevant
Physical quantity can synchronize raising.Conversely, related physical quantity can synchronize reduction.The battery more serious for degree of aging, above situation
Under, there will be institute's differences for the variation of each physical quantity, and therefore, the dispersion degree of array will increase.Therefore, pass through array dispersion point
Analysis can judge cell degradation degree.
In application dispersion analysis, it is necessary to consider to participate in the coefficient of variation of data, therefore, the present invention refers to physics for each
Amount has all carried out corresponding variation, to reduce influence of the absolute figure to array dispersion.
1, Time frame variation coefficient=actual discharge time/0.2C multiplying power lower discharge time;
2, the capacity coefficient of variation=actual capacity/nominal capacity;
3, the open-circuit voltage coefficient of variation=practical open-circuit voltage/voltage rating;
4, temperature variations coefficient=actual temperature/20 DEG C.
(2) it establishes neural network model and carries out degeneration factor estimation
Below by the explanation for establishing model, three model training, simulating, verifying aspect progress SOD discrimination methods.
1) neural network model is established
Fuzzy inference system (Fuzzy Inference System, FIS) in this model pastes mould using Sugeno pattern
Type establishes FIS according to a large amount of reliable experimental datas, so that the model of building is more objective, to also avoid because items are joined
Number, which is highly coupled, is difficult to the problem of removing.
According to lithium ion battery external behavior Parameter analysis, finishing screen selects discharge time tIt puts, release capacity QIt puts, open circuit electricity
Press Uk, input of the environment temperature T as neural network model, the output of model is that the degree of degeneration of lithium ion battery describes ginseng
Number --- degeneration factor β is shown in formula (1).
β=f (tIt puts, QIt puts, Uk, T) and (1)
T in formulaIt putsIndicate discharge time, unit is hour (h);
QIt putsIt indicates to release capacity, unit is ampere-hour (Ah);
UkIndicate open-circuit voltage, unit is volt (V);
T indicates that environment temperature, unit are degree Celsius (DEG C).
Using the adaptive neural network fuzzy system in MATLAB, Sugeno pattern fuzzy model is established automatically as shown in Figure 1, four
A input quantity is respectively discharge time, discharge capacity, open-circuit voltage, environment temperature.Output variable, that is, degeneration factor β.β is 0
Numerical value between~1, when β is close to 0, then cell degradation lesser extent;When β is close to 1, then cell degradation degree is serious.
2) lithium battery group is tested
Using four groups of Li-ion batteries piles samples, (every group of battery pack is connected in series with three sheet lithium ion batteries, single for this experiment
Body battery size is INCMP58145155N-I, voltage rating 3.7V, rated capacity 10Ah) implement four kinds simultaneously and different fills
Discharge system, specific charge-discharge parameter setting are as shown in table 1.
1 battery set charge/discharge parameter of table
Experiment combines acquisition relevant experimental data using upper computer and lower computer, and slave computer is by the new University of Science and Technology's energy in Shijiazhuang
The battery comprehensive parameters automatic test equipment (model BTS-M 300A/12V) of source development corporation, Ltd. production, related experiment
Data are recorded by uploading to host computer computer terminal.
3) data training and emulation
The present invention is considered as the relative discrete degree between input variable parameter to consider the degeneration journey of lithium ion battery
It spends (State of Degeneration), degree of degeneration is smaller, and the dispersion degree between input quantity will be smaller, opposite to degenerate
Degree is bigger, and dispersion degree will be bigger.
It is as shown in Figure 2 to establish Artificial Neural Network Structures.
Experimental data is divided into two groups --- training group and check groups.Training data is used to input as model training, with instruction
Practice system model.The step-length that training is arranged is 30, and training result is as shown in Figure 3.
Can see trained error from the lower left corner of Fig. 3 is 0.00021859, tentatively judges that the system performance is good.Point
Not with test data set and inspection data group come test macro, test data set mean error is as the result is shown
0.00082952, inspection data group mean error are as follows: 0.0025974, it is good that such error display goes out system performance.
According to above-mentioned training, simulation model is built as shown in figure 4, input data collection variable is e, output variable h.
Using Data Processing in Experiment result as the input of model, corresponding date lithium ion battery can be respectively obtained
Degeneration factor.Using the model to the degeneration factor calculated result of certain specific battery parameter.Fig. 5 is to degenerate in four groups of Cell Experimentation Ans
The tracking situation of coefficient, from the figure, it can be seen that the variation of other three groups degeneration factors is slower, and 4# group cell degradation system
Number variation is than comparatively fast, illustrating that its catagen speed is accelerated.And 4# group battery is using 0.5C charge and discharge.Available conclusion: big
Current charging and discharging can accelerate cell degradation speed.
(3) degeneration alarm system
1) composition of system
Degeneration alarm system overall construction design is as shown in Figure 6.
According to Fig.6, System Working Principle is as follows: system is by the portions such as upper computer and lower computer and detection display alarm
Divide and constitutes.Detected by electric current, voltage, temperature of the detection device to the battery (group) of work, according to electric discharge average current with
Discharge time accumulation calculates releasing capacity, the open-circuit voltage before discharging, environment temperature.Host computer is sent to carry out degeneration factor meter
It calculates, sends calculated value back to slave computer, if degeneration factor is more than threshold value, issue alarm.
2) slave computer and peripheral circuit
Slave computer chooses STC89C52, and peripheral circuit is by crystal oscillating circuit, reset circuit, voltage and current Acquisition Circuit, temperature
The part such as detection circuit, warning circuit, clock acquisition is spent to form.The structure of slave computer and peripheral circuit is as shown in Figure 7.
3) software design
A. calculation of capacity
The design carries out actual capacity detection using ampere-hour method, if the capacity of lithium battery is indicated with Q, its unit is
Ah.Capacity can be calculated with formula (2):
In formula: i is battery discharge current;
T is battery discharge time.
Integral operation may be implemented in the algorithm of cumulative summation.
QIt puts=∑ Ia×Δt (3)
In formula: IaIt is load current;Δ t is the sampling period (i.e. 1 millisecond) of main control chip.
Current detecting and calculation of capacity flow chart are as shown in Figure 8.
B. communication is calculated with degeneration factor and is alarmed
Fig. 9 is communication and degeneration factor calculating alarm flow figure.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention.It is all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.