A fast charging method for emergency power supply of UAV based on fuzzy control

The emergency power supply of UAV refers to the power supply that provides electrical energy to the onboard electrical equipment when the main power supply fails during the flight of the UAV. The emergency power supply of a certain type of UAV in our school is composed of 20 nickel-cadmium battery units. The requirements of modern warfare for weapons and equipment are unprecedentedly harsh. Specifically for UAV systems, it includes many aspects such as greatly shortening the preparation time, improving the accuracy of system detection, improving the overall reliability of the system, and enhancing the quality of the system's intelligence information collection. However, most of the emergency power supplies of UAVs currently use timely maintenance and charging one day in advance to ensure their combat effectiveness, which is time-consuming. In addition, occasional omissions caused by manual maintenance are also inevitable, which are serious hidden dangers that affect the performance of combat effectiveness.

1 Charging characteristics of emergency power supply for drones
The reason why emergency power supply for drones should be charged with a small current one day in advance is that its charging and discharging is a complex electrochemical change process. In order to ensure the life span, the charging speed can only be sacrificed.
1.1 Multivariable
There are many factors that affect the charging process, such as electrolyte concentration, plate active material activity, and ambient temperature, which can all affect the charging speed. This makes it impossible for a simple control system to play a significant role in improving charging efficiency.
1.2 Nonlinearity
The optimal charging voltage of the emergency power supply for drones is a variable exponential function of time during the entire charging process, and will show different changing rules at different stages. The charging characteristic curve is shown in Figure 1.

As can be seen from Figure 1, the charging process of the emergency power supply of the drone can be divided into AB section, BC section, and CD section. Among them, the AB section is the initial stage of charging, and the power is basically used up. At this stage, a constant small current can be used for charging; the BC section is the section with the most drastic voltage change. If constant current charging is used, the current is large and the battery is easily damaged. If the current is small, the time cannot be fully explored; the voltage of the CD section starts to drop from the highest point, and trickle charging can be used.
1.3 Uniqueness
The charging current of the emergency power supply of the drone varies greatly according to its specific usage conditions, even if the same set of drone systems are equipped with power supplies of the same capacity and the same model.
The current charging technology used in the equipment does not take into account the nonlinear changes in the charging process, and cannot realize the adaptive charging of the emergency power supply of the drone. It can only be charged with a relatively small current, resulting in slow charging speed, and serious gassing in the later stage of charging, which causes damage to the internal power supply. Not only can it not be charged quickly, but it also greatly shortens the service life of the power supply, which must be improved. After comprehensive comparison, a fuzzy control strategy with low cost and strong adaptability can be selected.

2 Design of fuzzy control charger
Fuzzy control does not require the mastery of the mathematical model of the controlled object, and is particularly suitable for this nonlinear control. It has strong adaptability to changes in process parameters, and some artificial experience factors can be added to make the control process easier to implement according to human requirements. It can be expected that the working principle of this fuzzy control charger is to pre-design a control strategy table and store it in the ROM of the single-chip microcomputer. During control, the value of the input quantity is calculated according to the sampling result, and then it is fuzzified by the quantization factor to obtain its domain. Then the corresponding control quantity is obtained by looking up the table. The control quantity is multiplied by the proportional factor, which can be used as the output quantity to control the charging process.
2.1 Determination of input and output quantities
The selection of the input quantity of the fuzzy controller has a great influence on the performance of the system. The power supply temperature, power supply terminal voltage and charging current can all be used as input quantities, but these methods are difficult to implement in engineering and have poor effects. According to Figure 1, the BC segment is the segment with the most drastic voltage change and lasts for a long time. The voltage change rate △U/△t in this segment can be used as the input quantity of the fuzzy control, and then supplemented by the difference △E between the real-time voltage of the power supply and the maximum chargeable voltage, a more perfect control can be achieved. The output quantity is based on the charging current adjusted by the duty cycle increment △ton of the PWM wave.
2.2 Determination of language variables, domain and membership
Taking (-4, -3, -2, -1, 0, +1, +2, +3, +4) as the domain, when the basic domain of the input quantity AE is [0, 18], the quantization factor is:

Accordingly, the maximum value of △U/△t obtained from the experiment is 5.236 0×10-4V/s, and the quantization factor is:

When the output is charging current, the division method of the domain segment is the same as above, and I is used to represent the output control quantity, then the proportional factor is determined as:

For the domain (-4, -3, -2, -1, 0, +1, +2, +3, +4), 8 linguistic variable values ​​are defined: NB (negative large), NM (negative medium), NS (negative small), NO (zero negative), PO (zero positive), PS (positive small), PM (positive medium) and PB (positive large), and the normal function model μA(X) = e-(xa/b)2 is used to construct the membership function, as shown in Figure 2. Among them: the parameter a takes an appropriate value to map the set {-4, -3, -2, -1, 0, +1, +2, +3, +4} to the set {NB, NM, NS, NO, PO, PS, PM, PB}; the parameter 6 determines the shape of the membership function. The appropriate control sensitivity and stability can be obtained according to the adjustment of the system error. Here, the single-point fuzzy method is used when the fuzzification maps the exact value to the domain of the corresponding fuzzy quantity.

2.3 Establishment of fuzzy control strategy table
From the discussion in Section 1.3, we can know that different power sources or the same power source in different usage conditions have different specific charging performances. If only △E is used as the main criterion, it will inevitably bring a large deviation to the control result. Based on this understanding, we can improve it by increasing the weight of △U/△t. Its analytical formula is:

In the formula: α∈[0,1] is called the correction factor; <…> indicates rounding to the nearest integer; E and EC are the fuzzy quantization of △E and △U/△t respectively.
Considering the differences of batteries, the weight is appropriately increased, and α=0.6 is set. After the above calculation, the fuzzy control table can be obtained through the maximum membership judgment. The fuzzy control table is shown in Table 1.

In the table: 0 means maintaining the current charging current; +1 means the charging current increases by one level; -1 means the charging current decreases by one level. And so on, multiply the current level by the proportional factor KU to get the output current change value. Adding this change value to the current value at the previous moment is the current value that should be output this time.

3 Composition and Implementation
The fuzzy controller needs to complete the fuzzification of the input signal (the deviation and the rate of change of the deviation between the given signal and the feedback signal), perform fuzzy reasoning and fuzzy judgment (defuzzification) according to the fuzzy knowledge base, and obtain the precisely controlled variables. However, since the system adopts the method of pre-storing the fuzzy control table in ROM, the online reasoning operation is transformed into a table lookup operation, which greatly improves the response speed of the system. Its structural composition is shown in Figure 3.

3.1 Microprocessor module
The fuzzification, fuzzy decision and defuzzification in Figure 3 are all completed in the microprocessor module. Here, the Motorola microcontroller MC68HC05SR3 is selected. It has rich internal resources and large ROM and RAM space, which is convenient for implementing fuzzy control. In addition, it also has 4 A/D converters, which is very convenient for detecting analog quantities. The microcontroller and the corresponding interface circuit form the control core of the system.
3.2 Negative feedback circuit
The voltage and current detection circuit detects the system charging current through the A/D converter; the battery terminal voltage, battery temperature and other parameters are controlled by forming a negative feedback loop through the sampling circuit, thermistor, etc.
3.3 Charging current output circuit
First, the converter circuit converts the AC mains into the required DC voltage through pulse width modulation, and then calculates the optimal change amount through the microprocessor module according to the charging voltage and current obtained by the negative feedback circuit detection system, and adds this change amount to the charging circuit. After PWM output, the required optimal charging current is obtained.

4 Conclusion
The optimal charging voltage curve based on fuzzy control is obtained by monitoring the voltage change rate of the emergency power supply of the UAV. The fuzzy control strategy table is designed with this curve as the input, and the intelligent tracking fuzzy control is realized. By comparing with the traditional charging method, it is confirmed that this fast charging method of the emergency power supply of the UAV based on fuzzy control has the following advantages: greatly accelerated charging speed, low battery temperature rise, charging according to the optimal curve without damaging the power supply, etc. It can be seen that the use of this technology can realize the rapid and intelligent charging process of the emergency power supply of the UAV, which is of great significance to the stable performance of the combat capability of the UAV weapon system.