Position self-adaptive adjusting method for thin film of composite battery

Abstract

The invention provides a position self-adaptive adjustment method for a thin film of a composite battery, and the method comprises the steps: obtaining the working state of a target composite battery through a plurality of times of simulation of the working operation of the composite battery, and monitoring the battery temperature, the charge-discharge state and the internal resistance of the target composite battery in the working state in real time; in the design process of the composite battery, driving a film bracket through a servo motor, and adjusting an included angle between a film and the surface of the battery according to the first target inclination angle and the real-time temperature of the battery; the method comprises the following steps: training monitoring data of each film inclination angle and rotation angle adjustment and film inclination angle and rotation angle adjustment data to establish a mapping relation between a composite battery film position and a battery working state; in the design process of the composite battery, the adjustment amplitude and frequency of the inclination angle and the rotation angle of the thin film are dynamically adjusted, and under the condition of presetting battery safety and efficiency, the energy consumption of thin film position adjustment is targeted, so that the position of the composite battery thin film is adaptively adjusted.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to a method for adaptively adjusting the position of a film of a composite battery. Background Art

[0002] In the design process of composite batteries, the working environment and conditions of the battery are often complex and changeable. Moreover, during the charging and discharging process, the internal temperature, voltage, current and other parameters of the battery will change dynamically. At the same time, the external environment of the battery, such as ambient temperature, light intensity and other factors, is also constantly changing. The changes in these internal and external factors will directly affect the performance and life of the battery, so it is necessary to focus on these internal and external factors in the battery design process. First of all, the thin film material in the composite battery plays a key role. The adjustment of its position and angle is directly related to the heat dissipation effect and photoelectric conversion efficiency of the battery. However, in the battery design process, how to dynamically adjust the position and angle of the film in combination with the changes in the external environment, so as to improve the heat dissipation effect and photoelectric conversion efficiency of the battery, is a technical problem that needs to be solved urgently. The traditional film position adjustment method is usually fixed, or simply adjusted according to preset conditions, which cannot adapt to the real-time changes in the working state and environment of the battery. Therefore, how to achieve adaptive adjustment of the position of the composite battery film so that it can dynamically adjust the inclination and rotation angle of the film according to the real-time status of the battery and changes in the external environment, thereby achieving dynamic control of battery heat dissipation and optimization of photoelectric conversion efficiency, is a technical problem that needs to be solved urgently. It is also the key to achieving efficient and stable operation of composite batteries. Summary of the invention

[0003] The present invention provides a method for adaptively adjusting the position of a film of a composite battery, which mainly includes:

[0004] Through several simulations of the operation of the composite battery, the working state of the target composite battery is obtained, and the battery temperature, charge and discharge state and internal resistance of the target composite battery under the working state are monitored in real time;

[0005] Compare the battery temperature, charge-discharge state and internal resistance monitoring data of the target composite battery under the working state with the preset safety ranges of the battery temperature, charge-discharge state and internal resistance, respectively, to determine the first target inclination angle and the first target rotation angle of the composite battery film;

[0006] In the design process of the composite battery, the film support is driven by a servo motor to adjust the angle between the film and the battery surface according to the first target inclination angle and the real-time temperature of the battery;

[0007] According to the first target rotation angle, the film support is controlled to rotate, thereby adjusting the direction of the film, and detecting whether the film is kept on the target light-receiving surface or light-reflecting surface. If the rotation angle makes the film not on the target light-receiving surface or light-reflecting surface, the film direction is adjusted in time to keep the film on the target light-receiving surface or light-reflecting surface;

[0008] During the film inclination and rotation angle adjustment process, the battery temperature, charge and discharge state and internal resistance monitoring data are continuously obtained to determine the influence of the adjusted film position on the battery working state, thereby forming a closed-loop feedback control to adjust the second target inclination and second target rotation angle of the film;

[0009] By training the monitoring data of each film tilt and rotation angle adjustment and the film tilt and rotation angle adjustment data, a mapping relationship between the composite battery film position and the battery working state is established;

[0010] In the design process of composite batteries, the adjustment amplitude and frequency of the film inclination and rotation angle are dynamically adjusted. Under the conditions of preset battery safety and efficiency, the energy consumption of film position adjustment is targeted, so that the composite battery film position can be adaptively adjusted.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0012] The present invention discloses a method for adaptively adjusting the position of a film of a composite battery. The method obtains the working state of the composite battery through simulation, monitors the battery temperature, charge and discharge state and internal resistance in real time, and determines the target inclination and rotation angle of the film after comparing with the preset safety range. The film support is driven by a servo motor to adjust the angle and orientation between the film and the battery surface to form a closed-loop feedback control. By continuously acquiring monitoring data, the influence of the film position on the working state of the battery is judged, and a mapping model is established. According to the model output, the adjustment amplitude and frequency of the film position are dynamically adjusted under the premise of ensuring the safety and efficiency of the battery to achieve energy consumption optimization. The present invention can adaptively adjust the film position according to the working state of the battery, effectively improve the safety and working efficiency of the composite battery, reduce energy consumption, and has important practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The present invention is a flow chart of a method for adaptively adjusting the position of a thin film of a composite battery.

[0014] Figure 2 It is a schematic diagram of a method for adaptively adjusting the position of a thin film of a composite battery according to the present invention.

[0015] Figure 3 It is another schematic diagram of a method for adaptively adjusting the position of a thin film of a composite battery according to the present invention. DETAILED DESCRIPTION

[0016] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments. The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are only used to explain the relevant inventions, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0017] like Figure 1 -3. The method for adaptively adjusting the position of a thin film of a composite battery in this embodiment may specifically include:

[0018] Step S101 , simulating the operation of the composite battery several times, obtaining the working state of the target composite battery, and monitoring the battery temperature, charge and discharge state and internal resistance of the target composite battery in the working state in real time.

[0019] A temperature sensor array is used to obtain the surface temperature data of the composite battery, and the temperature distribution data is obtained through bicubic spline interpolation according to the temperature data; the heat flux density and thermal conductivity are calculated according to the temperature distribution data, and the temperature field data is obtained through the Laplace equation and the finite difference method; a convolutional neural network training model is established for the temperature field data, and the convolutional neural network model uses the temperature field data as the input layer and the current data as the output layer, and obtains the charge and discharge state prediction result through forward propagation and reverse gradient correction; the current and temperature characteristic vectors are extracted from the charge and discharge state prediction result, and feature matching is performed based on the characteristic vectors and the measured voltage data, and the voltage and current ratio is calculated by the least squares method to obtain the composite battery internal resistance change data.

[0020] Exemplarily, the geometric structure of the composite battery is established through physical field simulation software, and a temperature sensor array is arranged on the surface of the composite battery with a grid spacing of 5 mm. The temperature data, charge and discharge current data, and voltage data are collected using a 1-second sampling period. The composite battery is simulated and run, and the temperature data is processed by bicubic spline interpolation to obtain temperature distribution data. The heat flux density and thermal conductivity of the composite battery surface are calculated based on the temperature distribution data, and the Laplace equation is used to solve the heat conduction process. Newton cooling boundary conditions are added at the boundary points, and the overall temperature field data is obtained by iterative calculation using the finite difference method. The formula for temperature distribution under Newton cooling conditions is:

[0021]

[0022] T(x) represents the temperature at position x, T∞ represents the ambient temperature far from the boundary, Ts represents the surface temperature, and L represents the characteristic length. A data table is established for the temperature field data and current data, with the temperature field data as the input layer and the current data as the output layer. A convolutional neural network is used for training. The training samples are divided into a training set and a validation set according to an 8-to-2 ratio. The charge and discharge state prediction results are obtained through forward propagation and reverse gradient correction. The current and temperature feature vectors are extracted from the charge and discharge state prediction results, and the features are matched with the measured voltage data. The ratio of voltage to current is calculated using the least squares method. The internal resistance change data of the composite battery is derived in combination with Ohm's law, and the internal resistance data is standardized. The temperature field simulation of the composite battery depends on accurate geometric structure modeling. The structure of the battery usually contains multiple layers of composite materials such as positive and negative electrodes, diaphragms, and electrolytes. Each layer of material has different thermal conductivity properties. In practical applications, the positive electrode material is mostly lithium nickel cobalt manganese oxide, the negative electrode is graphite, and the diaphragm is polypropylene or polyethylene. The thermal conductivity coefficients of these materials are 1.48, 25.0, and 0.22 watts per meter Kelvin, respectively. The temperature sensor array is arranged in a matrix with a spacing of 5 mm. The number of sensors is determined according to the battery size, and the measurement error of a single sensor is within 0.1 degrees Celsius. When bicubic spline interpolation is used to process temperature data, the continuity of the first-order derivative is maintained at the interpolation node, and the interpolation function forms a smooth transition in adjacent intervals. In the calculation of heat flux density, the convection heat transfer coefficient between the composite battery surface and the air is taken as 15 watts per square meter Kelvin, and the ambient temperature is set at 25 degrees Celsius. In the process of solving the Laplace equation, the five-point difference format is used to discretize the partial differential equation, the grid division accuracy is 1 mm, and the boundary conditions use the third-class boundary conditions. The finite difference iterative calculation sets the convergence threshold to 0.001 degrees Celsius and the maximum number of iterations is 1000 times. The charge and discharge state prediction adopts a convolutional neural network structure. The input layer contains 32 temperature field feature maps, the hidden layer contains 3 convolutional layers and 2 fully connected layers, and the output layer corresponds to the three states of charging, discharging, and static. The training data set contains 10,000 sets of temperature field and current data pairs, of which 8,000 sets are used for training and 2,000 sets are used for verification. The training process uses a stochastic gradient descent optimizer and the learning rate is set to 0.001. The internal resistance calculation uses the least squares method to process the voltage and current data. The sampling time window is 10 seconds, and 100 voltage and current data points are collected in each time window. The internal resistance data is normalized using the Z-score method to convert the data to a distribution with a mean of 0 and a standard deviation of 1. The internal resistance of the battery changes with the increase in the number of charge and discharge cycles. The internal resistance of a new battery is about 10 milliohms. After 500 charge and discharge cycles, the internal resistance rises to 15 milliohms. When the internal resistance increases by more than 50%, the battery life is determined to be terminated. In actual measurements, the internal resistance is negatively correlated with temperature. For every 10 degrees Celsius increase in temperature, the internal resistance decreases by about 5%.During the fast charging process, the temperature distribution of the composite battery is high in the center and low at the edge. The temperature in the center is 3 to 5 degrees Celsius higher than that in the edge. During the discharge process, the temperature distribution is relatively uniform, and the temperature gradient is less than 2 degrees Celsius per centimeter. The temperature field data is significantly correlated with the charge and discharge state. The temperature rise rate is 0.5 to 1 degrees Celsius per minute in the charging state, and the temperature rise rate is 0.3 to 0.8 degrees Celsius per minute in the discharge state.

[0023] Step S102, comparing the battery temperature, charge and discharge state and internal resistance monitoring data of the target composite battery in the working state with the preset safety ranges of battery temperature, charge and discharge state and internal resistance, to determine the first target tilt angle and the first target rotation angle of the composite battery film.

[0024] Acquire battery temperature monitoring data, and obtain the initial value of the film inclination angle according to the deviation between the temperature monitoring data and the preset temperature safety range; acquire the current density excess value according to the charge and discharge current monitoring data, and obtain the film rotation angle reference value based on the current density excess value; collect the battery internal resistance value, and convert the part of the internal resistance value exceeding the preset internal resistance safety threshold into an angular coordinate component, and obtain the film inclination correction value by performing trigonometric function conversion on the temperature deviation and the angular coordinate component; use a back propagation neural network to perform weighted calculation on the film inclination initial value and the inclination correction value, and obtain the film first target inclination value and the first target rotation angle value from the output layer of the neural network.

[0025] Exemplarily, the battery temperature monitoring data is obtained through a real-time acquisition device, and the upper limit of the temperature safety interval is set to 45 degrees Celsius and the lower limit is set to 10 degrees Celsius. If the monitored temperature exceeds the upper limit or is lower than the lower limit, the temperature deviation value is obtained by subtracting the boundary value from the over-limit value, and the temperature deviation value is converted into the initial value of the film inclination angle according to the linear mapping formula, in degrees. For the charge and discharge current monitoring data, the upper limit of the charging current density is set to 2 amperes per square centimeter, and the upper limit of the discharge current density is set to 3 amperes per square centimeter, and the over-limit value of the charge and discharge state is obtained. The film adjustment weight is calculated according to the ratio of the over-limit value to the temperature deviation value, and the film rotation angle reference value is obtained by multiplying the weight value by the standardization coefficient. The real-time data detection device is used to collect the battery internal resistance value, and the preset internal resistance safety threshold of 20 milliohms is compared. The part exceeding the threshold is converted into the radius component in the angular coordinate, and the temperature deviation value is converted into the angle component. The film inclination correction value is obtained by trigonometric conversion, in degrees. The back propagation neural network is used to weight the initial and corrected values ​​of the film inclination. The input layer contains three neurons: temperature deviation value, over-limit value, and internal resistance deviation value. The hidden layer uses a hyperbolic tangent activation function. The output layer corresponds to the target film inclination value and rotation angle value. The final adjustment parameters of the composite battery film are determined according to the inclination value and the rotation angle value. Composite battery temperature monitoring involves multi-level safety threshold setting. The upper limit of the temperature safety range of 45 degrees Celsius represents the critical point of battery thermal runaway. At this time, the electrochemical reaction inside the battery accelerates violently, while the lower limit of 10 degrees Celsius corresponds to the critical temperature at which the battery material performance decreases. In practical applications, the temperature sensitivity of different types of composite batteries varies. Lithium-ion batteries begin to show performance degradation at 35 degrees Celsius, while sodium-ion batteries remain stable at 40 degrees Celsius. The charge and discharge current density limit is set based on the battery material tolerance. The upper limit of the charge current density is 2 amperes per square centimeter to avoid lithium ion deposition. Higher charge rates will cause lithium dendrite growth. The upper limit of the discharge current density is set to 3 amperes per square centimeter. At this value, the battery heat generation is maintained within a safe range. The measured data show that when the charging current exceeds the limit by 0.5 amperes per square centimeter, the film inclination angle is adjusted by 15 degrees, and when the discharge current exceeds the limit by 0.8 amperes per square centimeter, the rotation angle increases by 20 degrees. Battery internal resistance monitoring reflects the health status of the battery. The internal resistance baseline value of a new battery is 15 milliohms, and the safety threshold is set at 20 milliohms. The internal resistance growth rate varies with the number of cycles. The internal resistance growth rate for the first 100 cycles is 0.01 milliohms per time, and the growth rate for 100 to 300 cycles is increased to 0.02 milliohms per time. The internal resistance over-limit value is converted to polar coordinates. The radius component represents the degree of over-limit, and the angle component corresponds to the temperature effect. After conversion, the film inclination correction value is obtained. The neural network weight adjustment adopts the error back propagation mechanism. The three neurons in the input layer receive the temperature deviation value, the current over-limit value, and the internal resistance deviation value respectively. The hidden layer uses the hyperbolic tangent function to process the nonlinear relationship, and the output layer gives the combined value of the film inclination angle and the rotation angle.The training data is collected from the battery operating status under different working conditions, including abnormal scenarios such as rapid temperature rise, drastic current fluctuations, and sudden changes in internal resistance. In network training, the temperature deviation weight coefficient is 0.4, the current overlimit weight coefficient is 0.35, and the internal resistance deviation weight coefficient is 0.25. In the dynamic response process of the film adjustment parameters, the three parameters of temperature, current, and internal resistance restrict each other. The increase in temperature causes the internal resistance to decrease, which in turn affects the charge and discharge current distribution. High current charging and discharging aggravates the temperature rise, forming a positive feedback effect. The film inclination adjustment breaks this positive feedback by changing the heat dissipation area. When the temperature exceeds the limit by 5 degrees Celsius, the inclination increases by 10 degrees and the heat dissipation area increases by 15%. The rotation angle adjustment changes the direction of heat flow. When the current exceeds the limit, the heat flow intensity decreases by 20% for every 30 degrees increase in the rotation angle.

[0026] The adjustment scale of the battery inclination angle is determined according to the value of the battery temperature exceeding the preset safety range and the value of the charge and discharge state exceeding the preset safety range; the adjustment scale of the rotation angle under the internal resistance condition is determined according to the value of the internal resistance exceeding the preset safety range; based on the adjustment scale of the battery inclination angle and the adjustment scale of the rotation angle under the internal resistance condition, the first target inclination angle and the first target rotation angle of the composite battery film are judged.

[0027] According to the over-limit value of the battery temperature, a three-point linear interpolation calculation of the temperature interval is used to obtain the inclination adjustment amount corresponding to the temperature, and the inclination adjustment amount and the over-limit value of the charge and discharge state are combined with the rate segmentation point calculation to obtain the total inclination adjustment value; for the over-limit value of the internal resistance, a standard resistance reference value is used to obtain the rate difference, and the rate difference is mapped to the angle interval through a hyperbolic tangent function to obtain a rotation angle reference value, and the rotation angle reference value and the total inclination adjustment value construct an adjustment amount matrix; a long short-term memory network is used to train the correspondence between the over-limit values ​​of the battery temperature exceeding the preset safety range, the charge and discharge state exceeding the preset safety range, and the internal resistance exceeding the preset safety range to the adjustment amount, and the correspondence is limited to the adjustment range by an activation function to obtain a inclination rotation angle mapping matrix; a dot product operation is performed on the values ​​of the mapping matrix and the adjustment amount matrix to obtain the initial values ​​of the adjustment parameters, and the initial values ​​of the adjustment parameters are iteratively calculated by the least squares method to obtain the first target inclination angle and the first target rotation angle of the composite battery film.

[0028] Exemplarily, evenly distributed interpolation nodes are established for the collected battery temperature over-limit values, and the tilt angle adjustment corresponding to the temperature is calculated by three-point linear interpolation with a temperature interval of 5 degrees. The corresponding tilt angle correction is calculated according to the over-limit value of the charge and discharge state combined with the piecewise linear function of the 0.5 rate segmentation point. The temperature weight coefficient 0.6 and the charge and discharge weight coefficient 0.4 are respectively multiplied by the adjustment value and then added to obtain the total adjustment value of the battery tilt angle. The standard resistance reference value is used to compare the over-limit value of the internal resistance, and the obtained rate difference is mapped to the angle range of 0 to 90 degrees through the hyperbolic tangent function. The rate difference is converted into a rotation angle reference value in combination with the sine transform, and the adjustment amount matrix is ​​constructed from the total adjustment value of the battery tilt angle and the rotation angle reference value. The correspondence between the three over-limit values ​​and the adjustment amount is trained through a long short-term memory network. The number of neurons in the hidden layer is set to twice the input dimension. The input data is processed by standard deviation normalization. The output is limited to the effective adjustment range through the activation function to obtain the tilt rotation angle mapping matrix. Dot product operation is performed based on the mapping matrix value and the adjustment matrix to obtain the initial value of the adjustment parameter, and the square sum of the product of the inclination adjustment amount and the angular velocity is set as the optimization objective function. The first target inclination angle and rotation angle of the composite battery film are obtained by iterative calculation using the least squares method. When the film inclination is calculated by linear interpolation, the adjustment interval is divided according to the degree of temperature exceeding the limit. When the temperature exceeds the preset range by 5 degrees Celsius, the corresponding inclination angle is adjusted by 10 degrees, when it exceeds 10 degrees Celsius, it is adjusted by 20 degrees, and when it exceeds 15 degrees Celsius, it is adjusted by 30 degrees. The adjustment angle corresponding to the intermediate temperature value is calculated through the interpolation point. The over-limit of the charge and discharge state is also calculated in segments. Taking the nominal rate 1C as the benchmark, the inclination angle increases by 15 degrees when the charging current exceeds 0.5C, and increases by 25 degrees when it exceeds 1C. The discharge current uses similar segmentation points. The weight coefficient reflects the influence ratio of temperature and charge and discharge state. The higher temperature weight reflects the priority of temperature control. The abnormal degree of the internal resistance exceeding the limit value is quantified by ratio comparison. The standard internal resistance value is 15 milliohms. The ratio of the measured value to the standard value is mapped to the angle space through the hyperbolic tangent function. When the internal resistance is 1.2 times the standard value, the rotation angle is 30 degrees, 1.5 times is 60 degrees, and 2 times is close to 90 degrees. The sine transform converts the linear proportional relationship into a nonlinear adjustment characteristic to avoid excessive angle adjustment. The adjustment matrix contains two dimensions: inclination angle and rotation angle. The matrix elements represent the coupling relationship between different parameters. During the training process of the long short-term memory network, the input layer receives three normalized over-limit values ​​of temperature, charge and discharge state, and internal resistance. The hidden layer contains 6 neurons, and the timing characteristics are processed through the triple structure of forget gate, input gate, and output gate. The network training samples come from measured data under different working conditions, including extreme scenarios such as fast charging, fast discharging, and temperature changes. The training data is normalized by standard deviation to eliminate the influence of dimension. The dot product operation of the mapping matrix and the adjustment matrix comprehensively considers the influence of various parameters. The initial adjustment value is optimized by the least squares method, and the angular velocity term in the objective function limits the stability of the adjustment process.In actual applications, when the battery temperature suddenly rises by 8 degrees Celsius, the charging rate exceeds 0.8C, and the internal resistance increases by 40%, the calculated target inclination angle is 35 degrees and the rotation angle is 45 degrees. The two angles work together to change the heat dissipation conditions. The increase in the film inclination angle increases the heat dissipation area, and the change in the rotation angle adjusts the direction of heat flow. Under the joint action, the temperature limit value drops by 80% within 300 seconds, and the charging rate drops to a safe range. There are differences in the adjustment characteristics of different types of batteries. The temperature tolerance of lithium iron phosphate batteries is stronger than that of ternary lithium batteries. The inclination angle adjustment amount is reduced by 20% under the same limit value. The internal resistance change of sodium ion batteries is more sensitive. When the internal resistance exceeds the limit, the rotation angle adjustment amount increases by 30%. In the dynamic response process of adjusting the parameters, the inclination angle change and the rotation movement constrain each other. The balance point is found through the least squares method to ensure the cooling effect while avoiding mechanical vibration.

[0029] Step S103, during the design process of the composite battery, the film support is driven by a servo motor to adjust the angle between the film and the battery surface according to the first target inclination angle and the real-time temperature of the battery.

[0030] A film angle adjustment value is calculated based on the first target inclination angle and the battery temperature signal, and the adjustment value changes according to a preset increment when the temperature exceeds a safe range; an incremental encoder is used to obtain a servo motor rotor position signal, and an angle control amount is obtained through the position signal and a preset resolution parameter, and the angle control amount is used to generate a pulse width modulation signal; a deviation value between an actual torque and an expected torque is calculated for the servo motor rotor position signal, and a motor compensation signal is obtained through the torque deviation value and a reference speed parameter; Kalman filtering is performed on the position data collected by the film angle sensor, and a real-time calibration value is obtained by comparing the filtered position data with the target angle, and the calibration value is output to the servo driver.

[0031] Exemplarily, the film angle adjustment value is calculated based on the first target inclination angle and the real-time temperature of the battery, the upper limit of the temperature safety range is set to 45 degrees Celsius and the lower limit is set to 10 degrees Celsius, and the temperature change rate is calculated using a 0.1 second sampling period. If the temperature exceeds the safety range, the angle increases by 5 degrees per second, and when the temperature drops to the safety range, the angle decreases by 3 degrees per second. The speed control amount of the servo motor is obtained through kinematic conversion. A 2000-line incremental encoder is used to obtain the servo motor rotor position signal, and the angle resolution of 0.045 degrees per pulse is obtained through quadruple frequency decoding. The speed compensation amount is calculated according to the proportional gain coefficient of 1.5, and a pulse width signal with a frequency of 20 kHz and an adjustable duty cycle is generated to drive the servo motor to operate. For the servo motor rotor position signal, the gradient descent algorithm is combined to calculate the deviation between the actual torque and the expected torque, and a closed-loop feedback controller is constructed for the torque deviation. The motor control compensation signal is generated based on the reference speed of 1000 revolutions per minute, and the controller sampling period is 1 millisecond. According to the position data collected by the film angle sensor, the Kalman filter is used to eliminate the mechanical vibration noise below 50 Hz, and the position feedback data is compared with the target angle to output the real-time calibration value of the film angle to the servo driver. During the film angle adjustment process, the temperature change rate reflects the heat accumulation rate of the battery. When the temperature rises rapidly from 25 degrees Celsius to 40 degrees Celsius, the temperature change rate reaches 1.5 degrees Celsius per second. At this time, the angle adjustment speed is correspondingly increased to 7.5 degrees per second, and the heat dissipation area increases rapidly. On the contrary, when the temperature drops from 40 degrees Celsius to 30 degrees Celsius, the change rate is -1 degrees Celsius per second, and the angle adjustment speed is reduced to 3 degrees per second, maintaining moderate heat dissipation. The corresponding relationship between the motor speed and the angle change rate in the kinematic conversion is 120 revolutions per minute corresponding to an angle change of 6 degrees per second. The incremental encoder realizes position detection through orthogonal signals. The A and B phase signals produce a 90-degree phase difference, and the resolution is increased to 8000 pulses per revolution through quadrature frequency technology. In actual applications, when the motor speed is 600 rpm, the encoder output frequency is 80 kHz, and the position resolution reaches 0.045 degrees. The proportional gain coefficient is dynamically adjusted with the speed change, the gain in the low-speed section is 2.0, and the high-speed section is reduced to 1.2 to avoid system oscillation. The pulse width signal has a duty cycle range of 10% to 90% at a carrier frequency of 20 kHz, corresponding to a motor speed of 0 to 3000 rpm. The torque control uses a gradient descent algorithm to achieve fast convergence, and an iterative calculation is completed within a sampling period of 1 millisecond. The measured data shows that the torque peak reaches 2 Nm when the motor starts, drops to 0.5 Nm during steady-state operation, and the torque fluctuation amplitude is controlled within 0.1 Nm. The reference speed is set to 1000 rpm, corresponding to a film angle change rate of 50 degrees per minute, which meets the temperature dynamic response requirements. Mechanical vibration mainly comes from motor operation and bracket movement, and spectrum analysis shows that it is mainly concentrated in the range of 20 to 50 Hz.The state equation of the Kalman filter contains two components: position and velocity. The standard deviation of the observation noise is set to 0.1 degrees, and the standard deviation of the process noise is 0.05 degrees per second. After filtering, the smoothness of the position data is improved, and the residual vibration amplitude is reduced to 0.02 degrees. In a typical heat dissipation scenario, when the battery temperature rises from 25 degrees Celsius to 42 degrees Celsius, the film angle gradually increases from the initial 15 degrees to 45 degrees. The rapid dissipation of heat causes the temperature rise rate to drop from 1.2 degrees Celsius per second to 0.3 degrees Celsius per second. During the temperature recovery stage, the angle slowly decreases and falls back to 20 degrees within 300 seconds. The servo motor speed is stable throughout the process, and the position control accuracy is better than 0.1 degrees. There are differences in the temperature characteristics of different types of batteries. The temperature response of ternary lithium batteries is more sensitive. Under the same working conditions, the angle adjustment speed is increased by 20%, while the temperature stability of lithium iron phosphate batteries is better, and the angle adjustment speed is reduced by 30%.

[0032] Step S104, according to the first target rotation angle, controls the film bracket to rotate, thereby adjusting the orientation of the film, and detecting whether the film remains on the target light-receiving surface or light-reflecting surface. If the rotation angle makes the film not on the target light-receiving surface or reaction surface, the film orientation is adjusted in time to keep the film on the target light-receiving surface or reaction surface.

[0033] The deviation value is calculated according to the target rotation angle and the feedback signal of the rotary encoder, the three-phase current value of the driving motor is detected by a current sampler, and the rotation compensation amount is calculated according to the amplitude and phase of the three-phase current value; the operation of the photoelectric sensor array is controlled according to the rotation compensation amount, and the photoelectric sensor array receives the surface reflected light data, and obtains the light receiving characteristic map through convolution neural network processing; the value of the film backlight area is extracted according to the light receiving characteristic map, and the orientation deviation angle is obtained according to the backlight area value processed by a proportional integral operator; the orientation deviation angle is processed by a sliding mean filter, and it is judged according to the value after filtering and the preset threshold value. If the light intensity is lower than the preset standard value, the correction amount is increased, and the bracket rotation compensation angle is output through the correction amount.

[0034] Exemplarily, the deviation value is calculated based on the first target rotation angle and the feedback signal of the rotary encoder, and the current sampler with a sampling frequency of 10 kHz is used to detect the three-phase current value of the drive motor. The rotation compensation amount is calculated from the current amplitude and phase, and the motor control amount is generated by combining the calibrated speed of 1000 revolutions per minute, and the motor rotation angle is controlled by closed-loop feedback. An 8×8 photoelectric sensor array is arranged around the surface of the film, and the photoelectric sensor array spacing is 10 mm. The surface reflected light data is received, and the light receiving feature map is constructed through a convolutional neural network. The network contains 3 convolutional layers and 2 fully connected layers, and the output film surface light intensity distribution data matches the preset light receiving surface standard data. The film backlight area value is extracted from the light intensity distribution data, and the orientation deviation angle is obtained by proportional integral operator processing. The proportional coefficient is set to 1.5, and the integral time constant is set to 0.5 seconds. The film orientation correction amount is calculated based on the orientation deviation angle and the upper limit of the bracket speed of 90 degrees per second. For the orientation correction, a 16-point sliding mean filter is used to remove high-frequency jitter. The adaptive threshold judgment rule is set to increase the correction by 20% when the light intensity is lower than the preset standard value of 80%, and the correction remains unchanged when the light intensity is higher than the standard value. The compensation angle of the bracket rotation is output from the correction to achieve closed-loop control of the film orientation. The film rotation accuracy control involves the coordination of multiple links, the first of which is the motor drive control. The three-phase current is detected at a sampling frequency of 10 kHz, which can accurately capture the subtle changes in the current waveform. The measured data shows that the current phase deviation is controlled within 2 degrees. When the target angle is 45 degrees, the motor runs at a calibrated speed of 1000 revolutions per minute, and the positioning error is limited to 0.1 degrees through closed-loop control. The photoelectric sensor array adopts a matrix layout, with an 8×8 array covering an 80×80 mm range, and a field of view of 30 degrees for each sensor. In actual application, when the film is on the light-receiving surface, the light intensity in the central area reaches 1000 lux and that in the edge area is 800 lux; when it turns to the reflective surface, the central area drops to 200 lux and the edge area is 150 lux. The convolutional neural network extracts light intensity features through a 3-layer 16-channel convolution structure, and the fully connected layer outputs the matching value. When the matching degree exceeds 90%, the film is judged to be in the correct orientation. The feature extraction of the backlight area focuses on the gradient change of light intensity. When facing normally, the light intensity in the backlight area does not exceed 100 lux and is evenly distributed. The parameters of the proportional-integral controller have been optimized, the proportional coefficient of 1.5 ensures the response speed, and the integral time constant of 0.5 seconds suppresses overshoot. The measured data shows that in the process of switching from the light-receiving surface to the reflective surface, the 90-degree rotation is completed within 1 second, and the overshoot is less than 2 degrees. The sliding mean filter uses a 16-point data window, corresponding to a time span of 1.6 milliseconds, which effectively filters out mechanical vibrations above 50 Hz. The adaptive threshold judgment is dynamically adjusted under different working conditions. When the ambient light intensity decreases, the photoelectric signal is attenuated by 20% overall. At this time, the threshold is reduced accordingly to maintain the detection sensitivity.Under cloudy conditions, the light intensity reference value automatically drops to 600 lux, and the judgment rules are adjusted accordingly. In the dynamic scene test, the film direction switches every 10 seconds, and a total of 1,000 cycles are run without failure. During a single switching process, the peak starting current of the motor is 2 amperes, the steady-state current is 0.5 amperes, and the rotation process is smooth and jitter-free. Photoelectric detection feedbacks angle information in real time, and the positioning accuracy is better than 0.5 degrees. Under different temperature conditions, in the range of -10 to 50 degrees Celsius, the sensitivity change of the photoelectric device does not exceed 5%, and the detection stability is maintained by adaptive threshold compensation. Repeated positioning accuracy tests show that after switching the light-receiving surface and the reflective surface 100 times, the end point angle error is less than 1 degree, which meets the positioning requirements.

[0035] Step S105, during the process of adjusting the film inclination angle and rotation angle, continuously obtain the battery temperature, charge and discharge status and internal resistance monitoring data, judge the influence of the adjusted film position on the battery working state, form a closed-loop feedback control, and adjust the second target inclination angle and the second target rotation angle of the film.

[0036] A high-speed sampling device is used to obtain battery temperature monitoring data, charge and discharge state data and internal resistance detection data, and a temperature change rate value, a charge and discharge state change rate value and an internal resistance change rate value are obtained through a central difference operation based on the battery temperature monitoring data, the charge and discharge state data and the internal resistance detection data; a three-dimensional response space is established based on the temperature change rate value, the charge and discharge state change rate value and the internal resistance change rate value, and a radial basis kernel function support vector regression is used to calculate the influence of film adjustment on the correction of three response values, and a position adjustment correction amount is obtained from the combined change of the three-axis response values; a variable step gain controller is used to generate an adjustment signal for the position adjustment correction amount, and a second target inclination value of the film is obtained by multiplying the response weight and the correction amount; if there is an angle deviation between the second target inclination value and the current inclination value, the angle deviation and the three response values ​​are used to form an input vector, and the second target rotation angle value is obtained by mapping the output layer of a feedforward neural network.

[0037] Exemplarily, the battery temperature monitoring data, charge and discharge state data, and internal resistance detection data are collected every 0.1 seconds by a high-speed sampling device, and the temperature change rate value, charge and discharge state change rate value, and internal resistance change rate value are obtained by using a 5-point central difference operation. The change rate values ​​obtained before and after the film adjustment are paired according to the time sequence before and after, and the parameter adjustment response value is calculated. A three-dimensional response space is established according to the parameter adjustment response value, with the temperature change rate as the X-axis, the charge and discharge state change rate as the Y-axis, and the internal resistance change rate as the Z-axis. The support vector regression of the radial basis kernel function is used to calculate the correction influence of the film adjustment on the three response values, and the position adjustment correction amount is output from the combined change of the three-axis response value. A variable step gain controller is used to generate an adjustment signal for the position adjustment correction amount, and the temperature response weight is set to 0.5, the charge and discharge response weight is set to 0.3, and the internal resistance response weight is set to 0.2. The second target inclination value of the film is obtained by multiplying the response weight and the correction amount in combination with the real-time response data. The angle deviation is obtained by subtracting the current inclination value from the second target inclination value. The angle deviation and the three response values ​​form an input vector and are input into a feedforward neural network with 4 layers of hidden neurons. The hidden layer uses a hyperbolic tangent activation function, and the output layer is mapped to the range of 0 to 360 degrees to obtain the second target rotation angle value. Real-time monitoring of the battery parameter change rate reflects the immediate effect of film adjustment, and a sampling period of 0.1 seconds can capture rapid changes. In actual applications, when the film inclination is adjusted from 30 degrees to 45 degrees, the temperature change rate decreases from 0.8 degrees Celsius per second to 0.3 degrees Celsius per second, the charging current change rate decreases from 0.5 amperes per second to 0.2 amperes per second, and the internal resistance change rate decreases from 0.05 milliohms per second to 0.02 milliohms per second. The central difference calculation uses data from the two points before and after to eliminate the influence of single-point fluctuations. During the construction of the three-dimensional response space, the temperature change rate range is set at -2 to 2 degrees Celsius per second, the charge and discharge change rate range is -1 to 1 ampere per second, and the internal resistance change rate range is -0.1 to 0.1 milliohms per second. The support vector regression uses the radial basis kernel function, the kernel parameter is set to 0.5, and the penalty factor is 1.0. When the temperature drops rapidly, the charging current is stable, and the internal resistance rises slowly, the correction influence shows that the temperature response is dominant, and the correction amount is mainly determined by the temperature change. In the variable step gain control, different response weights reflect the priority relationship of the parameters. The temperature response weight of 0.5 reflects the importance of heat dissipation control, the charge and discharge response weight of 0.3 focuses on current stability, and the internal resistance response weight of 0.2 monitors the battery health status. The measured data shows that when the temperature change rate decreases by 50%, the tilt angle adjustment amount is 15 degrees, and when the charging current change rate decreases by 30%, the tilt angle adjustment amount is 8 degrees. The input vector of the neural network contains 4 components, which are processed by 4 hidden layers, each containing 16 neurons. The hyperbolic tangent activation function limits the output to the range of -1 to 1, and then converts it to the actual angle value through linear mapping.During the dynamic adjustment process, when the inclination deviation is 20 degrees and the temperature change rate is 0.6 degrees Celsius per second, the rotation angle of the network output is 90 degrees, achieving rapid adjustment. In actual operation, differences in the response characteristics of different types of batteries were observed. The ternary lithium battery is more sensitive to temperature changes, and the temperature change rate is reduced by 60% under the same inclination adjustment, while the lithium iron phosphate battery has a slower temperature response and the change rate is reduced by 40%. The charging and discharging characteristics also show obvious differences. The temperature of high-rate batteries rises rapidly when charged with large currents, requiring a larger inclination adjustment. The change in internal resistance is related to the battery life stage. The internal resistance of new batteries is stable, while the internal resistance of aging batteries fluctuates greatly. The controller adapts to different working conditions through weight adaptive adjustment.

[0038] Step S106, establishing a mapping relationship between the composite battery film position and the battery working state by training the monitoring data of each film tilt angle and rotation angle adjustment and the film tilt angle and rotation angle adjustment data.

[0039] The temperature data, charge and discharge state data and internal resistance detection data of the battery are obtained, and the film inclination value and rotation angle value at the corresponding moment are recorded according to the temperature data, charge and discharge state data and internal resistance detection data of the battery, and the normalized data are obtained by maximum and minimum value normalization processing; a recurrent neural network model is established according to the normalized data, and the inclination value and rotation angle value are input into the recurrent neural network model to obtain the temperature value, charge and discharge state value and internal resistance detection value; a matrix database is established for the temperature value, charge and discharge state value and internal resistance detection value, and the matrix database uses the inclination value and the rotation angle value as row and column indexes to filter and obtain the filtered matrix data; a bidirectional linear interpolation calculation is performed on the filtered matrix data to obtain the battery state parameters under any combination of inclination value and rotation angle value, and the battery state parameters are subjected to Gaussian smoothing processing, and a state prediction model is established in combination with the matrix database.

[0040] Exemplarily, for the battery temperature data, charge and discharge state data, and internal resistance detection data collected every 0.1 seconds, the film inclination value and rotation angle value at the corresponding moment are recorded, and the maximum and minimum value normalization processing is used to map all data to the interval of 0 to 1. The training data of a single adjustment process is formed by timestamp association, and the training data is determined to be valid when the data fluctuation amplitude is less than 0.1. A recurrent neural network is constructed based on the single adjustment training data. The input layer contains two neurons of inclination value and rotation angle value, the hidden layer contains 4 layers of 32 neurons per layer, and the output layer contains three neurons of temperature value, charge and discharge state value, and internal resistance value. The learning rate is set to 0.01, the number of training iterations is 1000, and the gradient is updated through the mean square error function. A matrix database is established based on the trained data records, with inclination value and rotation angle value as row and column indexes, and the matrix elements store battery state parameters. Quartiles are used for outlier filtering, and missing data are supplemented by cubic spline functions to construct a continuous mapping data structure. The battery state parameters under any combination of inclination and rotation angles are calculated by bidirectional linear interpolation, and the interpolation results are Gaussian smoothed to eliminate spikes. The state prediction function is established in combination with the historical adjustment sequence data to realize the mapping conversion from the film position to the battery working state. During the training data acquisition process, the battery state parameters form an accurate correspondence with the film position adjustment. Taking the typical adjustment process as an example, when the film inclination increases from 15 degrees to 45 degrees, the temperature decreases from 42 degrees Celsius to 35 degrees Celsius, the charging current decreases from 2.5 amperes to 1.8 amperes, and the internal resistance decreases from 18 milliohms to 15 milliohms. The normalization process maps the temperature to the range of 0.4-0.8, the charging current to the range of 0.3-0.9, and the internal resistance to the range of 0.2-0.7 to eliminate the dimension effect. The architecture design of the recurrent neural network is optimized for timing characteristics. The input layer receives the normalized position data and extracts deep features through 4 hidden layers. 32 neurons in each layer provide sufficient feature extraction capabilities to avoid underfitting caused by too small a network. During the training process, the learning rate of 0.01 strikes a balance between convergence speed and stability, and 1000 iterations ensure sufficient learning. The measured data show that the prediction error after training is controlled within 5%. The database uses a matrix structure to store mapping relationships, with an inclination range of 0-90 degrees and a rotation angle range of 0-360 degrees, divided into 1-degree intervals to form a 90×360 matrix. Quartile filtering removes outliers, and data is judged as abnormal when it deviates from the median by 1.5 times the interquartile range. Cubic spline interpolation maintains the smoothness of the curve and ensures a smooth transition between adjacent data points. In practical applications, when the inclination angle is 30 degrees and the rotation angle is 180 degrees, the corresponding temperature is 38 degrees Celsius, the charging current is 2.0 amperes, and the internal resistance is 16 milliohms. Bidirectional linear interpolation realizes the calculation of state parameters at any position, and interpolation operations are performed in a 1-degree matrix grid. Gaussian smoothing uses a kernel function with a standard deviation of 0.5 to effectively suppress data fluctuations introduced by interpolation.The historical sequence shows that the state parameters of the same position adjustment under different working conditions fluctuate by 10%, which is dynamically adjusted through the prediction function. The mapping relationships of different types of batteries show significant differences. The temperature response of ternary lithium batteries is more sensitive. The same inclination change causes a 15% temperature change, while the temperature change of lithium iron phosphate batteries is only 8%. The charging and discharging characteristics also show obvious differences. High-rate batteries require greater position adjustment when charged at high currents. The internal resistance mapping relationship changes with battery life. The internal resistance of new batteries changes smoothly. After 500 cycles, the sensitivity of internal resistance to position adjustment increases by 30%.

[0041] Step S107, in the design process of the composite battery, dynamically adjust the adjustment amplitude and frequency of the film inclination angle and rotation angle, and target the energy consumption of the film position adjustment under the conditions of preset battery safety and efficiency, so that the composite battery film position is adaptively adjusted.

[0042] The temperature safety threshold, charge and discharge safety threshold and internal resistance safety threshold are obtained according to the battery working state parameters, and the position adjustment amount is calculated by a proportional integral differential controller; the position adjustment amount is received, and a position adjustment optimizer is constructed by a deep neural network, and the deep neural network outputs an inclination adjustment step value and an adjustment interval value according to the battery state parameters and the motor energy consumption data; for the inclination adjustment step value and the adjustment interval value, a numerical integrator based on the Euler method is used to construct a speed sequence, and the total energy consumption of the adjustment process is obtained by accumulating the voltage and current products, and a displacement curve is generated in combination with the temperature safety threshold, the charge and discharge safety threshold and the internal resistance safety threshold; a state transfer matrix is ​​constructed according to the displacement curve, and a position sensor is used to obtain a real-time position deviation, and a compensation amount is calculated by multiplying the position deviation with a gain matrix, and a film position adjustment instruction is output in combination with the compensation amount and the battery state update value.

[0043] Exemplarily, according to the battery working state parameters obtained by the mapping operation, the temperature safety threshold is set to 45 degrees Celsius, the charge and discharge safety threshold is set to 3 times, and the internal resistance safety threshold is set to 20 milliohms. The proportional integral differential controller is used to calculate the position adjustment amount, and the servo motor voltage and current values ​​are monitored in real time through the power detection circuit. The single adjustment energy consumption is obtained from the voltage and current product, and the minimum adjustment step length is calculated by combining the safety threshold and the energy consumption value. The position adjustment optimizer is constructed by a deep neural network. The input layer contains battery state parameters and motor energy consumption data. The hidden layer contains 4 layers of 64 neurons per layer. The output layer corresponds to the inclination adjustment step value and the adjustment interval value. The reward function is set to the weighted sum of the safety margin and the inverse of the energy consumption. The gradient ascent method is used to update the network parameters to obtain the adjustment law of the inclination and rotation angle. For the inclination adjustment value and the rotation angle value, a numerical integrator based on the Euler method is used to construct a speed sequence. The integration step size varies adaptively in the range of 0.1 to 1 second. The total energy consumption of the adjustment process is calculated by accumulating the voltage and current product. The speed curve is generated in combination with the battery safety boundary constraint, and the displacement curve is obtained from the speed curve integration. According to the displacement curve, an 8×8 state transfer matrix is ​​constructed. The controller with a proportional coefficient of 1.5 and a differential time of 0.5 seconds is set to perform trajectory tracking. The position sensor is used to feedback the real-time position deviation. The compensation amount is calculated by multiplying the deviation by the gain matrix. The film position adjustment instruction is output by combining the compensation amount and the battery state update value. The battery working state and the film position adjustment form a closed-loop optimization system. The safety threshold setting is based on the physical properties of the battery. The temperature threshold of 45 degrees Celsius corresponds to the material stability boundary. Exceeding this temperature accelerates the decomposition of the electrolyte. The charge and discharge rate of 3C is the balance point between capacity and life. Higher rates lead to the growth of lithium dendrites. The internal resistance threshold of 20 milliohms reflects the health status of the battery. Exceeding this value means that the performance is significantly attenuated. The energy consumption optimization process is achieved by real-time monitoring of motor parameters. Under typical working conditions, the servo motor has an operating voltage of 24 volts, a no-load current of 0.5 amperes, and a load current of 2.5 amperes. A single adjustment process lasts for 2 seconds, and the total energy consumption is about 120 joules. The minimum adjustment step size is dynamically adjusted with the change of energy consumption. When the energy consumption exceeds 150 joules, the step size is reduced to 5 degrees. When the energy consumption is less than 50 joules, the step size is increased to 15 degrees. The training of the deep neural network uses multi-condition data, including three types of scenarios: standard conditions, fast charging conditions, and high temperature conditions. The input data is normalized, the temperature range is mapped to 0-1, the current is mapped to 0-1.5, and the internal resistance is mapped to 0-2. During the network training process, the reward function weight is set to a safety margin of 0.7 and an energy consumption item of 0.3, reflecting the principle of safety priority. The training results show that the adjustment interval is 10 seconds under standard conditions and shortened to 5 seconds under fast charging conditions. The numerical integration adopts an adaptive step size strategy. The step size is 0.1 second when the temperature changes drastically and increases to 1 second when the temperature is stable. The measured data shows that during the 45-degree inclination adjustment process, the trajectory smoothness is high under a step size of 0.1 seconds, and the position overshoot is less than 1 degree, but the calculation amount increases by 5 times.The state transfer matrix reflects the conversion relationship between 8 typical position points, each of which contains two components: inclination and rotation angle. Different types of batteries exhibit unique regulation characteristics. Ternary lithium batteries have fast temperature response and short regulation intervals, with a typical value of 3 seconds; lithium iron phosphate batteries have good temperature stability, and the regulation interval can be extended to 15 seconds. Large-capacity batteries have large thermal capacity, so the regulation step is usually set above 10 degrees; small-capacity batteries have small thermal capacity, and the regulation step is reduced to about 5 degrees. The internal resistance of new batteries is stable, and the regulation law is mainly determined by temperature; the internal resistance of cycle-aged batteries fluctuates greatly, and the regulation law considers internal resistance constraints more. Controller parameters also change with operating conditions. During fast charging, the proportional coefficient is increased to 2.0 to speed up the response speed; under low temperature conditions, the differential time is extended to 0.8 seconds to reduce overshoot. Position deviation compensation uses nonlinear gain, with a gain of 1.0 for small deviations and an increase to 1.8 for large deviations, to achieve rapid convergence.

[0044] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for adaptively adjusting the position of a composite battery film, characterized in that: The method comprises: Through several simulations of the operation of the composite battery, the working state of the target composite battery is obtained, and the battery temperature, charge and discharge state and internal resistance of the target composite battery under the working state are monitored in real time; Compare the battery temperature, charge-discharge state and internal resistance monitoring data of the target composite battery under the working state with the preset safety ranges of the battery temperature, charge-discharge state and internal resistance, respectively, to determine the first target inclination angle and the first target rotation angle of the composite battery film; In the design process of the composite battery, the film support is driven by a servo motor to adjust the angle between the film and the battery surface according to the first target inclination angle and the real-time temperature of the battery; According to the first target rotation angle, the film support is controlled to rotate, thereby adjusting the direction of the film, and detecting whether the film is kept on the target light-receiving surface or light-reflecting surface. If the rotation angle makes the film not on the target light-receiving surface or light-reflecting surface, the film direction is adjusted in time to keep the film on the target light-receiving surface or light-reflecting surface; During the film inclination and rotation angle adjustment process, the battery temperature, charge and discharge state and internal resistance monitoring data are continuously obtained to determine the influence of the adjusted film position on the battery working state, thereby forming a closed-loop feedback control to adjust the second target inclination and second target rotation angle of the film; By training the monitoring data of each film tilt and rotation angle adjustment and the film tilt and rotation angle adjustment data, a mapping relationship between the composite battery film position and the battery working state is established; In the design process of composite batteries, the adjustment amplitude and frequency of the film inclination and rotation angle are dynamically adjusted. Under the conditions of preset battery safety and efficiency, the energy consumption of film position adjustment is targeted, so that the composite battery film position can be adaptively adjusted.

2. The method according to claim 1, characterized in that The method of simulating the operation of the composite battery several times to obtain the working state of the target composite battery and monitoring the battery temperature, charge and discharge state and internal resistance of the target composite battery in the working state in real time includes: Using a temperature sensor array to obtain composite battery surface temperature data, and obtaining temperature distribution data through bicubic spline interpolation based on the temperature data; Calculating heat flux and thermal conductivity according to the temperature distribution data, and obtaining temperature field data by Laplace equation and finite difference method; A convolutional neural network training model is established for the temperature field data, wherein the convolutional neural network model uses the temperature field data as an input layer and the current data as an output layer, and obtains a charge and discharge state prediction result through forward propagation and reverse gradient correction; The current and temperature characteristic vectors are extracted from the charge and discharge state prediction results, characteristic matching is performed based on the characteristic vectors and the measured voltage data, and the ratio of voltage to current is calculated by the least square method to obtain the internal resistance change data of the composite battery.

3. The method according to claim 1, characterized in that The method of comparing the battery temperature, charge-discharge state and internal resistance monitoring data of the target composite battery under the working state with the preset safety ranges of the battery temperature, charge-discharge state and internal resistance to determine the first target inclination angle and the first target rotation angle of the composite battery film includes: Acquire battery temperature monitoring data, and obtain an initial value of the film inclination angle according to a deviation between the temperature monitoring data and a preset temperature safety range; Obtaining a current density over-limit value according to the charge and discharge current monitoring data, and obtaining a film rotation angle reference value based on the current density over-limit value; Collecting the internal resistance value of the battery, converting the portion of the internal resistance value exceeding the preset internal resistance safety threshold into an angular coordinate component, and obtaining a film inclination correction value by performing trigonometric function conversion on the temperature deviation and the angular coordinate component; A back propagation neural network is used to perform weighted calculation on the initial tilt angle of the film and the tilt angle correction value, and a first target tilt angle value and a first target rotation angle value of the film are obtained from the output layer of the neural network; It also includes: determining the adjustment scale of the battery inclination angle according to the numerical value of the battery temperature exceeding the preset safety range and the numerical value of the charge and discharge state exceeding the preset safety range; determining the adjustment scale of the rotation angle under the internal resistance condition according to the numerical value of the internal resistance exceeding the preset safety range, and judging the first target inclination angle and the first target rotation angle of the composite battery film based on the adjustment scale of the battery inclination angle and the adjustment scale of the rotation angle under the internal resistance condition.

4. The method according to claim 3, characterized in that The method comprises: determining the adjustment scale of the battery inclination angle according to the value of the battery temperature exceeding the preset safety range and the value of the charge and discharge state exceeding the preset safety range; determining the adjustment scale of the rotation angle under the internal resistance condition according to the value of the internal resistance exceeding the preset safety range; and judging the first target inclination angle and the first target rotation angle of the composite battery film based on the adjustment scale of the battery inclination angle and the adjustment scale of the rotation angle under the internal resistance condition, including: According to the battery temperature over-limit value, a three-point linear interpolation calculation of the temperature interval is used to obtain the tilt angle adjustment value corresponding to the temperature, and the tilt angle adjustment value is combined with the charge and discharge state over-limit value and the rate segmentation point calculation to obtain the total tilt angle adjustment value; A standard resistance reference value is used to obtain a ratio difference value for an over-limit value of internal resistance, the ratio difference value is mapped to an angle interval through a hyperbolic tangent function to obtain a rotation angle reference value, and the rotation angle reference value and the total tilt angle adjustment value form an adjustment amount matrix; A long short-term memory network is used to train the correspondence between the over-limit values ​​of the battery temperature exceeding the preset safety range, the charge and discharge state exceeding the preset safety range, and the internal resistance exceeding the preset safety range and the adjustment amount, and the corresponding relationship is limited within the adjustment range by an activation function to obtain a tilt rotation angle mapping matrix; A dot product operation is performed on the values ​​of the mapping matrix and the adjustment amount matrix to obtain initial values ​​of the adjustment parameters, and the initial values ​​of the adjustment parameters are iteratively calculated by the least square method to obtain a first target inclination angle and a first target rotation angle of the composite battery film.

5. The method according to claim 1, characterized in that In the design process of the composite battery, the film support is driven by a servo motor to adjust the angle between the film and the battery surface according to the first target inclination angle and the real-time temperature of the battery, including: Calculating a film angle adjustment value according to the first target tilt angle and the battery temperature signal, wherein the adjustment value changes according to a preset increment when the temperature exceeds a safe range; An incremental encoder is used to obtain a servo motor rotor position signal, and an angle control amount is obtained through the position signal and a preset resolution parameter, and the angle control amount is used to generate a pulse width modulation signal; Calculating the deviation between the actual torque and the expected torque according to the servo motor rotor position signal, and obtaining a motor compensation signal through the torque deviation and a reference speed parameter; Kalman filtering is performed on the position data collected by the thin film angle sensor, and a real-time calibration value is obtained by comparing the filtered position data with the target angle, and the calibration value is output to the servo driver.

6. The method according to claim 1, characterized in that According to the first target rotation angle, the film support is controlled to rotate, thereby adjusting the direction of the film, detecting whether the film is kept on the target light-receiving surface or light-reflecting surface, and if the rotation angle makes the film not on the target light-receiving surface or light-reflecting surface, the film direction is adjusted in time to make the film remain on the target light-receiving surface or light-reflecting surface, including: The deviation value is calculated according to the target rotation angle and the feedback signal of the rotary encoder, the three-phase current value of the driving motor is detected by a current sampler, and the rotation compensation amount is calculated according to the amplitude and phase of the three-phase current value; Controlling the operation of the photoelectric sensor array according to the rotation compensation amount, wherein the photoelectric sensor array receives surface reflected light data and obtains a light receiving characteristic map through convolutional neural network processing; Extracting the film backlight area value according to the light receiving characteristic diagram, and obtaining the orientation deviation angle according to the backlight area value processed by a proportional integral operator; A sliding mean filter is used to process the orientation deviation angle. According to the value after filtering and the preset threshold value, if the light intensity is lower than the preset standard value, the correction amount is increased, and the bracket rotation compensation angle is output through the correction amount.

7. The method according to claim 1, characterized in that During the film inclination and rotation angle adjustment process, the battery temperature, charge and discharge state and internal resistance monitoring data are continuously obtained to determine the influence of the adjusted film position on the battery working state, thereby forming a closed-loop feedback control to adjust the second target inclination and second target rotation angle of the film, including: A high-speed sampling device is used to obtain battery temperature monitoring data, charge and discharge state data, and internal resistance detection data, and a temperature change rate value, a charge and discharge state change rate value, and an internal resistance change rate value are obtained by performing a central difference operation based on the battery temperature monitoring data, the charge and discharge state data, and the internal resistance detection data; A three-dimensional response space is established according to the temperature change rate value, the charge-discharge state change rate value and the internal resistance change rate value, and the influence of film adjustment on the correction of the three response values ​​is calculated by radial basis kernel function support vector regression, and the position adjustment correction amount is obtained from the combined change of the three-axis response values; A variable step gain controller is used to generate an adjustment signal for the position adjustment correction amount, and a second target inclination angle value of the film is obtained by multiplying the response weight and the correction amount; If there is an angle deviation between the second target inclination value and the current inclination value, the angle deviation and the three response values ​​are used to form an input vector, and the second target rotation angle value is obtained through feedforward neural network output layer mapping.

8. The method according to claim 1, characterized in that The mapping relationship between the position of the composite battery film and the working state of the battery is established by training the monitoring data of each film tilt angle and rotation angle adjustment and the film tilt angle and rotation angle adjustment data, including: Acquire temperature data, charge and discharge state data and internal resistance detection data of the battery, record the film inclination value and rotation angle value at the corresponding moment according to the temperature data, charge and discharge state data and internal resistance detection data of the battery, and obtain normalized data by maximum and minimum value normalization processing; Establishing a recurrent neural network model according to the normalized data, inputting the inclination angle value and the rotation angle value into the recurrent neural network model, and obtaining a temperature value, a charge-discharge state value, and an internal resistance detection value; A matrix database is established for the temperature value, the charge-discharge state value and the internal resistance detection value, wherein the matrix database uses the inclination value and the rotation angle value as row and column indexes, and filters to obtain filtered matrix data; Bidirectional linear interpolation calculation is performed on the filtered matrix data to obtain battery state parameters under any combination of inclination angle value and rotation angle value, Gaussian smoothing is performed on the battery state parameters, and a state prediction model is established in combination with the matrix database.

9. The method according to claim 1, characterized in that: In the design process of the composite battery, the adjustment amplitude and frequency of the film inclination angle and the rotation angle are dynamically adjusted, and the energy consumption of the film position adjustment is targeted under the conditions of preset battery safety and efficiency, so that the composite battery film position is adaptively adjusted, including: The temperature safety threshold, charge and discharge safety threshold and internal resistance safety threshold are obtained according to the battery working state parameters, and the position adjustment amount is calculated by the proportional integral differential controller; The position adjustment amount is received, and a position adjustment optimizer is constructed using a deep neural network, wherein the deep neural network outputs an inclination angle adjustment step value and an adjustment interval value according to battery state parameters and motor energy consumption data; According to the tilt angle adjustment step value and the adjustment interval value, a numerical integrator based on the Euler method is used to construct a speed sequence, the total energy consumption of the adjustment process is obtained by accumulating the voltage and current products, and a displacement curve is generated by combining the temperature safety threshold, the charge and discharge safety threshold and the internal resistance safety threshold; A state transfer matrix is ​​constructed according to the displacement curve, a position sensor is used to obtain a real-time position deviation, a compensation amount is calculated by multiplying the position deviation by a gain matrix, and a film position adjustment instruction is output in combination with the compensation amount and a battery state update value.

Images