Disclosure of Invention
The invention aims to provide an auxiliary equalizing charge device based on SOC estimation, which improves the equalizing charge efficiency of a battery pack. The device realizes self-adaptive auxiliary equalizing charge based on the feedback control of SOC (state of charge).
In order to achieve the above purpose, the present invention provides the following technical solutions:
an auxiliary equalizing charge device based on SOC estimation, comprising: the device comprises a rectifying module, a driving module, an output module, an acquisition module and a machine learning module;
the rectifying module is electrically connected with the driving module and is used for rectifying the input commercial power and then conveying the rectified commercial power to the driving module;
the driving module is electrically connected with the output module, converts current and transmits the converted current to the output module; the output module rectifies the voltage and the current and then outputs the rectified voltage and the rectified current as the final output of the device;
the input end of the acquisition module is electrically connected with the output module, and the voltage and the current of the output module are acquired and fed back;
the output end of the acquisition module is electrically connected with the machine learning module, and the acquired data is transmitted to the machine learning module after being converted;
the machine learning module predicts the SOC of the battery according to the obtained data, calculates to obtain optimal output waveform and frequency data according to the predicted SOC, and controls MCU (micro controller unit) to generate and output a corresponding waveform;
the output end of the machine learning module is electrically connected with the driving module and is used for transmitting the output waveform to the driving module and controlling the switching tube in the driving module to be switched on or off;
further, a data processing chip is arranged in the machine learning module, and the acquired data is learned and trained by using an artificial neural network;
further, training the data by adopting a circulating neural network mode is preferable, and taking the data in a time period as input to obtain the output of the current moment;
further, a main transformer is arranged in the driving module, and the secondary side of the main transformer is provided with two coil outputs which are respectively defined as a first coil and a second coil; the first coil is auxiliary output, the second coil is main output, and the acquisition module acquires the output of the second coil as feedback control;
further, the driving module is provided with a switching tube, and the conducting and cut-off frequency of the switching tube is limited between 50KHz and 110 KHz;
further, the acquisition module further comprises an error proportional amplifying circuit for scaling the received current and voltage signals;
further, the driving module is provided with a driving circuit, and the driving circuit comprises a pulse chip MCU1, a capacitor C2, a capacitor C3, a resistor R1, a resistor R2, a resistor R3, a resistor R4, a resistor R5, a resistor R6, a resistor R7, a resistor R8, a resistor R9, a resistor R10, a diode D1, a diode D2, a diode D3, a diode D4, a diode D5, a first transistor Q1, a second transistor Q2, a third transistor Q3, a fourth transistor Q4 and a transformer T1; pins 1, 2 and 7 of the pulse chip MCU1 are connected with a machine learning module; a pin 5 of the pulse chip MCU1 is connected with a gate of the third transistor Q3; the drain electrode of the third transistor Q3 is connected with the grid electrode of the first transistor Q1 and the grid electrode of the fourth transistor Q4; an emitter of the first transistor Q1 is connected with an emitter of the fourth transistor Q4, a cathode of the diode D4 and a resistor R8; the grid electrode of the second transistor Q2 is connected with the positive electrode of the diode D4 and the resistor R8, the drain electrode of the second transistor Q2 is connected with the positive electrode of the diode D2 and the negative end of the primary side of the transformer T1, and the source electrode of the second transistor Q2 is connected with the resistors R9 and R10; the pulse chip MCU1 generates corresponding waveforms by utilizing the data sent by the machine learning module, and outputs the waveforms from the pin 5 of the pulse chip MCU1 so as to drive the third transistor Q3 to be turned on and off; when the third transistor Q3 is turned on, the first transistor Q1 is turned on, and the fourth transistor Q4 is turned off, so the second transistor Q2 is turned on; when the third transistor Q3 is turned off, the first transistor Q1 is turned off, and the fourth transistor Q4 is turned on, so the second transistor Q2 is turned off; the driving current is amplified through the cooperative work of the first transistor Q1, the third transistor Q3 and the fourth transistor Q4, and finally the driving of the primary side current of the transformer T1 is realized through the second transistor Q2.
The beneficial effects of the invention include, but are not limited to:
(1) The invention provides an auxiliary equalizing charge device based on SOC estimation, which comprises the following components: the device comprises a rectifying module, a driving module, an output module, an acquisition module and a machine learning module;
the driving module converts the input current, and the acquisition module acquires and feeds back the voltage and the current of the output module; the machine learning module predicts the SOC of the battery according to the obtained data, calculates to obtain optimal output waveform and frequency data according to the predicted SOC, and controls the MCU to generate and output corresponding waveforms.
The driving module controls the switching tube in the driving module to be switched on or off based on the received waveform signal, and finally the purpose of stabilizing the output current is achieved.
(2) The invention also comprises a main transformer; the primary side of the main transformer is provided with a coil input; the secondary has two coil outputs, defined as a secondary first coil and a secondary second coil, respectively. The secondary side first coil is an auxiliary output, and the secondary side second coil is a main output. The output of the secondary side first coil supplies power to the machine learning module. The output of the secondary side second coil supplies power to the acquisition module, and the acquisition module acquires the output of the secondary side second coil as feedback control. The acquisition module further includes an error scaling circuit for scaling the received current and voltage signals.
(3) The auxiliary equalizing charge device based on SOC estimation provided by the invention can estimate the SOC data of the battery according to the collected voltage and current data, and determine the equalizing charge parameters according to the SOC data, so that the efficiency of the power supply is improved, and meanwhile, the equalizing charge efficiency is improved. Meanwhile, in the charging process, the machine learning module continuously collects the charging current, the pulse duty ratio and the frequency, the battery pack voltage and the temperature change, and adopts unsupervised learning to optimize the estimated model of the SOC.
Detailed Description
The present invention is described in detail below with reference to examples, but the present invention is not limited to these examples.
Fig. 1 is a schematic flow chart of an auxiliary equalizing charge device based on SOC estimation provided in the present application, as shown in fig. 1, including: the device comprises a rectifying module, a driving module, an output module, an acquisition module and a machine learning module;
the rectifying module is electrically connected with the driving module and is used for rectifying the input commercial power and then conveying the commercial power to the driving module;
the driving module is electrically connected with the output module, and is used for converting the current and transmitting the converted current to the output module. The output module rectifies the voltage and the current and then outputs the rectified voltage and the rectified current as the final output of the device;
the input end of the acquisition module is electrically connected with the output module, and the voltage and the current of the output module are acquired and fed back;
the output end of the acquisition module is electrically connected with the machine learning module, and the acquired data is transmitted to the machine learning module after being converted;
the machine learning module predicts the SOC of the battery according to the obtained data, calculates to obtain optimal output waveform and frequency data according to the predicted SOC, and controls the MCU to generate and output corresponding waveforms.
The output end of the machine learning module is electrically connected with the driving module and is used for transmitting the output waveform to the driving module and controlling the switching tube in the driving module to be switched on or off;
further, a data processing chip is arranged in the machine learning module, and the acquired data is learned and trained by using an artificial neural network.
Further, it is preferable to train the data by using a cyclic neural network, and take the data in a period of time as input to obtain the output at the current time.
Further, a main transformer is arranged in the driving module; the primary side of the main transformer is provided with a coil input; the secondary side of the main transformer has two coil outputs, defined as a secondary side first coil and a secondary side second coil, respectively. The secondary first coil is auxiliary output, the secondary second coil is main output, and the acquisition module acquires the output of the secondary second coil as feedback control.
Further, the driving module is provided with a switching tube, and the conducting and cut-off frequency of the switching tube is limited between 50KHz and 110 KHz.
Further, the acquisition module also includes an error scaling circuit for scaling the received current and voltage signals.
In the invention, a driving module converts input current, and an acquisition module acquires and feeds back voltage and current of an output module; the machine learning module predicts the SOC of the battery according to the obtained data, calculates to obtain optimal output waveform and frequency data according to the predicted SOC, and controls the MCU to generate and output corresponding waveforms. The driving module controls the switching tube in the driving module to be switched on or off based on the received waveform signal, and finally the purpose of stabilizing the output current is achieved.
The invention also comprises a main transformer; the primary side of the main transformer is provided with a coil input; the secondary has two coil outputs, defined as a secondary first coil and a secondary second coil, respectively. The secondary side first coil is an auxiliary output, and the secondary side second coil is a main output. The output of the secondary side first coil supplies power to the machine learning module. The output of the secondary side second coil supplies power to the acquisition module, and the acquisition module acquires the output of the secondary side second coil as feedback control. The acquisition module further includes an error scaling circuit for scaling the received current and voltage signals.
The auxiliary equalizing charge device based on SOC estimation provided by the invention can estimate the SOC data of the battery according to the collected voltage and current data, and determine the equalizing charge parameters according to the SOC data, so that the efficiency of the power supply is improved, and meanwhile, the equalizing charge efficiency is improved. Meanwhile, in the charging process, the machine learning module continuously collects the charging current, the pulse duty ratio and the frequency, and the changes of the battery pack voltage and the temperature, adopts unsupervised learning, optimizes an estimated model of the SOC, automatically adjusts the pulse duty ratio and the frequency, and outputs proper current.
Fig. 2 is a circuit diagram of a driving circuit and a first coil output provided in this embodiment, and as shown in fig. 2, the driving circuit includes a pulse chip MCU1, a capacitor C2, a capacitor C3, a resistor R1, a resistor R2, a resistor R3, a resistor R4, a resistor R5, a resistor R6, a resistor R7, a resistor R8, a resistor R9, a resistor R10, a diode D1, a diode D2, a diode D3, a diode D4, a diode D5, a first transistor Q1, a second transistor Q2, a third transistor Q3, a fourth transistor Q4, and a transformer T1.
Pins 1, 2 and 7 of the pulse chip MCU1 are connected with a machine learning module. A pin 5 of the pulse chip MCU1 is connected to a gate of the third transistor Q3. The drain of the third transistor Q3 is connected to the gate of the first transistor Q1 and the gate of the fourth transistor Q4. An emitter of the first transistor Q1 is connected to an emitter of the fourth transistor Q4, a diode D4 cathode, and a resistor R8. The grid electrode of the second transistor Q2 is connected with the positive electrode of the diode D4 and the resistor R8, the drain electrode of the second transistor Q2 is connected with the positive electrode of the diode D2 and the negative end of the primary side of the transformer T1, and the source electrode of the second transistor Q2 is connected with the resistors R9 and R10. The pulse chip MCU1 generates a corresponding type of waveform using the data transmitted from the machine learning module, and outputs the waveform from the pin 5 of the pulse chip MCU1, thereby driving the third transistor Q3 on and off. When the third transistor Q3 is turned on, the first transistor Q1 is turned on, and the fourth transistor Q4 is turned off, so the second transistor Q2 is turned on; when the third transistor Q3 is turned off, the first transistor Q1 is turned off, and the fourth transistor Q4 is turned on, so the second transistor Q2 is turned off; the driving current is amplified through the cooperative work of the first transistor Q1, the third transistor Q3 and the fourth transistor Q4, and finally the driving of the primary side current of the transformer T1 is realized through the second transistor Q2.
In this embodiment, the on/off frequency of the first transistor Q1, the second transistor Q2, the third transistor Q3, and the fourth transistor Q4 is limited to between 50KHz and 110 KHz.
Fig. 3 is a circuit diagram of the second coil output provided in this embodiment, and as shown in fig. 3, the circuit diagram includes a rectifying diode D6, a rectifying diode D7, a capacitor C4, a capacitor C5, a capacitor C6, a capacitor C7, a capacitor C8, a capacitor C9, a capacitor C10, a resistor R11, a resistor R12, a resistor R13, and an output terminal OUT1.
The cathode of the rectifying diode D6 is connected with the capacitor C6, the capacitor C4 and the resistor R11; and then output from the other end of the resistor R11 as a power supply of the machine learning module. The cathode of the rectifying diode D7 is connected with a capacitor C9, a capacitor C8 and a capacitor C7 and is used as the main output of the device;
the resistor R12 is a current sampling resistor and is connected in series in the output loop for collecting the output current.
Fig. 4 is a circuit diagram of a sampling feedback circuit provided in this embodiment, as shown in fig. 4, including a diode D8, a diode D9, a light emitting diode GREEN1, a light emitting diode RED1, a capacitor C11, a capacitor C12, a capacitor C13, a resistor R14, a resistor R15, a resistor R16, a resistor R17, a resistor R18, a resistor R19, a resistor R20, a resistor R21, a resistor R22, a resistor R23, a resistor R24, a resistor R25, a resistor R26, and a proportional amplifying chip LM1.
The circuit collects the output charging current and charging voltage, and transmits data to the machine learning module from the pins PH1, PH2, PH3 and PH4 after passing through the proportional amplifying chip.
The light emitting diodes GREEN1, RED1 are used to indicate the current charging mode.
Fig. 5 is an SOC estimation data diagram provided by an embodiment of the present invention, where, as shown in fig. 5, a dotted line is a true value of SOC, and a solid line is an estimated SOC value calculated by a machine learning module of the present apparatus.
The driving circuit in the auxiliary equalizing charge device based on SOC estimation provided by the invention is used for converting the primary side current of the main transformer T1; the acquisition module acquires and feeds back the voltage and the current of the output module; the machine learning module predicts the SOC of the battery according to the obtained data, calculates to obtain optimal output waveform and frequency data according to the predicted SOC, and controls the pulse chip MCU1 to generate and output a corresponding waveform. The waveform output by the pulse chip MCU1 is utilized to control the on and off of the first transistor Q1, the second transistor Q2, the third transistor Q3 and the fourth transistor Q4, so that the primary side current of the main transformer T1 is converted.
The invention also comprises a main transformer; the primary side of the main transformer is provided with a coil input; the secondary has two coil outputs, defined as a secondary first coil and a secondary second coil, respectively. The secondary side first coil is an auxiliary output, and the secondary side second coil is a main output. The output of the secondary side first coil supplies power to the machine learning module. The output of the secondary side second coil supplies power to the acquisition module, and the acquisition module acquires the output of the secondary side second coil as feedback control. The acquisition module further includes an error scaling circuit for scaling the received current and voltage signals. The output of the secondary side second coil is used as the main output of the device after passing through a rectifier diode D7, a capacitor C9, a capacitor C8 and a capacitor C7.
The auxiliary equalizing charge device based on SOC estimation provided by the invention can estimate the SOC data of the battery according to the collected voltage and current data, and determine the equalizing charge parameters according to the SOC data, so that the efficiency of the power supply is improved, and meanwhile, the equalizing charge efficiency is improved. Meanwhile, in the charging process, the machine learning module continuously collects the charging current, the pulse duty ratio and the frequency, and the changes of the battery pack voltage and the temperature, adopts unsupervised learning, optimizes an estimated model of the SOC, automatically adjusts the pulse duty ratio and the frequency, and outputs proper current.
The present invention is not limited to the above embodiments, but is not limited to the above embodiments, and any person skilled in the art will have obvious modifications and modifications equivalent to those of the equivalent embodiments, and can make various changes and modifications without departing from the scope of the present invention.