Battery state estimation and dynamic control method, system, terminal and storage medium fusing electric heating coupling mechanism and residual correction

Abstract

This invention relates to the field of battery state estimation technology, and discloses a battery state estimation and dynamic control method, system, terminal, and storage medium that integrates electrothermal coupling mechanism and residual correction. The method includes: introducing the constant power feedback mechanism and electrothermal coupling mechanism of a mobile terminal to construct a continuous-time state-of-charge (SOC) ordinary differential equation; solving the SOC equation; using a data-driven model for residual compensation to obtain the final SOC; and constructing low-level hardware control instructions based on the final SOC for dynamic hardware control. This invention integrates constant power feedback decoupling, continuous-time SOC equation solving, residual-driven compensation, and multi-objective optimization to construct a full-stack framework from low-level state estimation to high-level hardware scheduling, enabling stable operation under complex dynamic loads and sensor noise conditions.

Description

Technical Field

[0001] This invention relates to the field of battery state estimation technology, and in particular to a battery state estimation and dynamic control method, system, terminal, and computer-readable storage medium that integrates electrothermal coupling mechanism and residual correction. Background Technology

[0002] Existing physical estimation methods (such as the ampere-hour integration method and the standard extended Kalman filter) are mostly based on the ideal assumptions of "constant current discharge" or "approximate constant voltage", and simulate battery dynamics by constructing a simple RC circuit (resistance-capacitance circuit).

[0003] In reality, mobile terminal power management chips (PMICs) operate in a "constant-power" mode to maintain hardware computing power. When the battery discharges to a low level (e.g., SOC < 15%, where SOC represents State of Charge), the battery's open-circuit voltage (OCV) drops rapidly. To maintain constant power, the actual physical discharge current must surge nonlinearly. Existing models fail to decouple this constant-power feedback mechanism from the underlying partial differential equations, leading to severe estimation distortion and algorithm collapse in the low-battery range. This, in turn, causes the "last 1% battery drop precipitously (voltage avalanche)" phenomenon, which severely impacts user experience.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main objective of this invention is to provide a battery state estimation and dynamic control method, system, terminal, and computer-readable storage medium that integrates electrothermal coupling mechanism and residual correction. This aims to solve the problems of existing technologies that ignore constant power feedback at low battery levels, leading to "cliff-like power loss" and rigid static power-saving strategies.

[0006] To achieve the above objectives, this invention provides a battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction. The method includes the following steps: Based on the relationship between the open-circuit voltage and the state of charge of the target battery, a semi-empirical polynomial model of the target battery is constructed, and the actual discharge current of the target battery is calculated based on the semi-empirical polynomial model. Acquire temperature change information, construct an electrothermal coupling model between the effective capacity of the target battery and the temperature change information, construct a state of charge ordinary differential equation based on the actual discharge current and the electrothermal coupling model, and predict the current predicted state of charge of the target battery based on the state of charge ordinary differential equation. The interaction features of the target battery at the current moment are constructed, the interaction features are input into the machine learning compensation model for prediction, the residual estimate of the target battery at the current moment is output, and the current predicted state of charge is updated according to the residual estimate to obtain the final state of charge. A time-varying model of the target battery is constructed, and all parameters in the time-varying model are randomly perturbed and sampled to obtain the current status of each parameter. Based on all the current statuses and the final state of charge, a corresponding power-saving decision is constructed. All power-saving decisions are comprehensively evaluated to obtain the final decision with the highest efficiency ratio. The final decision is then converted into a low-level hardware control instruction and issued to adjust the current state of the hardware.

[0007] Optionally, the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction, wherein the step of constructing a semi-empirical polynomial model of the target battery based on the relationship between the open-circuit voltage and state of charge of the target battery, and calculating the actual discharge current of the target battery based on the semi-empirical polynomial model, specifically includes: Determine the open-circuit voltage of the target battery and the relationship between the open-circuit voltage and the normalized state of charge to construct a semi-empirical polynomial model of the target battery: ; ; in, Indicates open-circuit voltage. Represents the normalized state of charge. Indicates the state of charge. Represents the natural constant; Based on the internal resistance and open-circuit voltage of the target battery, construct the terminal voltage of the target battery in a closed circuit, eliminate the terminal voltage, and then solve for the actual discharge current of the target battery: ; ; ; in, Indicates the first Terminal voltage at time , Indicates the battery's internal resistance. Indicates the actual discharge current. This indicates the constant power of the target battery.

[0008] Optionally, the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction, wherein the steps of acquiring temperature change information, constructing an electrothermal coupling model between the effective capacity of the target battery and the temperature change information, constructing a state-of-charge ordinary differential equation based on the actual discharge current and the electrothermal coupling model, and predicting the current predicted state of charge of the target battery based on the state-of-charge ordinary differential equation, specifically include: The temperature change information of the target battery is obtained, and the relationship between the effective capacity of the target battery and the temperature change information is established based on the Arrhenius equation to construct an electrothermal coupling model of the target battery. ; in, Indicates temperature The effective capacity below, Indicates the rated capacity of the target battery. Indicates an indicator function, Indicates activation energy. Represents the ideal gas constant. Indicates the reference temperature. Indicates the first Surface temperature at any given time; Based on the actual discharge current and the electrothermal coupling model, an ordinary differential equation for the state of charge is constructed, and the equation is solved to obtain the current predicted state of charge of the target battery at the current moment. ; in, Indicates the first The current predicted state of charge at time t, Indicates the first The actual discharge current at any given moment.

[0009] Optionally, the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction, wherein constructing the interaction features of the target battery at the current moment, inputting the interaction features into a machine learning compensation model for prediction, outputting the residual estimate of the target battery at the current moment, and updating the currently predicted state of charge based on the residual estimate to obtain the final state of charge, specifically includes: Extract the current current and current temperature of the target battery, construct cross-interaction terms based on the current current and current temperature, and construct the interaction features of the target battery at the current moment based on the current current, current temperature and cross-interaction terms; The interaction features are input into a pre-trained machine learning compensation model, which performs a weighted calculation on the interaction features based on a nonlinear mapping function, and outputs the residual estimate of the target battery at the current time. The current predicted state of charge is updated using the residual estimate to obtain the final state of charge of the target battery at the current moment; ; in, Indicates the first The final state of charge at time t, Indicates the first The current predicted state of charge at time t, Indicates the first The residual estimate at time step 1.

[0010] Optionally, the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction, wherein constructing a time-varying model of the target battery, randomly perturbing and sampling all parameters in the time-varying model to obtain the current state of each parameter, and constructing a corresponding power-saving decision based on all the current states and the final state of charge, specifically includes: Based on the multiple component-level loads of the target battery, a time-varying model of the instantaneous total power consumption of the target battery is constructed. Random perturbation sampling is performed on all parameters in the time-varying model to calculate the average image effect and nonlinear interaction effect for each parameter; Based on the average image effect and the nonlinear interaction effect of each parameter, a core driving factor for each parameter is constructed, wherein the core driving factor represents the factor that plays a decisive role in the power consumption of the target battery. Based on all the current core driving factors and the current final state of charge, construct a power-saving decision for each parameter.

[0011] Optionally, the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction, wherein the step of comprehensively evaluating all the power-saving decisions to obtain the final decision with the highest efficiency ratio, and converting the final decision into low-level hardware control instructions, and issuing the low-level hardware control instructions to adjust the current state of the hardware, specifically includes: All the core driving factors are mapped to a decision space, and the similarity degree corresponding to each parameter is calculated in the decision space; All the core driving factors are sorted according to their similarity, and the power-saving decision corresponding to the core driving factor with the highest similarity is defined as the final decision. The final decision is converted into underlying hardware control instructions, which are then sent to the hardware execution module to adjust the current state of multiple hardware devices.

[0012] Optionally, the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction, wherein mapping all the core driving factors to a decision space and calculating the proximity degree corresponding to each parameter in the decision space specifically includes: Define the expected energy-saving benefits and the user experience loss, and construct a two-dimensional decision space based on the expected energy-saving benefits and the user experience loss; Based on the average image effect and the nonlinear interaction effect of each core driving factor, each core driving factor is mapped into the decision space, and all core driving factors in the preset region below the decision space are removed. For all core driving factors within a preset region above the decision space, the degree of dispersion of each core driving factor is calculated using the entropy weight method, and a weight is assigned to each core driving factor based on the degree of dispersion. Based on each of the weights, the relative proximity of each core driving factor to the ideal solution is calculated.

[0013] Furthermore, to achieve the above objectives, the present invention also provides a battery state estimation and dynamic control system that integrates electrothermal coupling mechanism and residual correction, wherein the battery state estimation and dynamic control system that integrates electrothermal coupling mechanism and residual correction includes: The mechanism reconstruction module is used to construct a semi-empirical polynomial model of the target battery based on the relationship between the open-circuit voltage and the state of charge of the target battery, and to calculate the actual discharge current of the target battery based on the semi-empirical polynomial model. The initial state of charge prediction module is used to acquire temperature change information, construct an electrothermal coupling model between the effective capacity of the target battery and the temperature change information, construct a state of charge ordinary differential equation based on the actual discharge current and the electrothermal coupling model, and predict the current predicted state of charge of the target battery based on the state of charge ordinary differential equation. The final state of charge prediction module is used to construct the interaction features of the target battery at the current moment, input the interaction features into the machine learning compensation model for prediction, output the residual estimate of the target battery at the current moment, and update the current predicted state of charge based on the residual estimate to obtain the final state of charge. The decision-making construction module is used to construct a time-varying model of the target battery, randomly perturb and sample all parameters in the time-varying model to obtain the current status of each parameter, and construct a corresponding power-saving decision based on all the current statuses and the final state of charge. The decision positioning module is used to comprehensively evaluate all the power-saving decisions, obtain the final decision with the highest efficiency ratio, and convert the final decision into underlying hardware control instructions, and issue the underlying hardware control instructions to adjust the current state of the hardware.

[0014] Furthermore, to achieve the above objectives, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction, stored in the memory and executable on the processor. When the battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction is executed by the processor, it implements the steps of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction as described above.

[0015] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction, wherein when the battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction is executed by a processor, it implements the steps of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction as described above.

[0016] In this invention, a semi-empirical polynomial model of the target battery is constructed based on the relationship between its open-circuit voltage and state of charge (SCC). The actual discharge current of the target battery is calculated based on this semi-empirical polynomial model. Temperature change information is acquired, and an electrothermal coupling model between the effective capacity of the target battery and the temperature change information is constructed. An ordinary differential equation for the SCC is constructed based on the actual discharge current and the electrothermal coupling model. The current predicted SCC of the target battery is predicted based on this SCC. The interaction features of the target battery at the current moment are constructed, and these interaction features are input into machine learning compensation. The model makes predictions and outputs the residual estimate of the target battery at the current time. Based on the residual estimate, the current predicted state of charge (SOC) is updated to obtain the final SOC. A time-varying model of the target battery is constructed, and all parameters in the time-varying model are randomly perturbed and sampled to obtain the current state of each parameter. Based on all the current states and the final SOC, corresponding power-saving decisions are constructed. All power-saving decisions are comprehensively evaluated to obtain the final decision with the highest efficiency ratio. The final decision is then converted into low-level hardware control instructions, which are issued to adjust the current state of the hardware. This invention integrates constant power feedback decoupling, continuous-time SOC ordinary differential equation solving, residual-driven compensation, and multi-objective optimization to construct a full-stack framework from low-level state estimation to high-level hardware scheduling, enabling stable operation under complex dynamic loads and sensor noise conditions. Attached Figure Description

[0017] Figure 1 This is a flowchart of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction according to the present invention; Figure 2 This is a schematic diagram of the target battery abstraction of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction of the present invention; Figure 3 This is a schematic diagram of the multiphysics coupling framework of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction of the present invention. Figure 4 This is a diagram showing the battery physical model characteristic verification results of a preferred embodiment of the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction of the present invention. Figure 5 This is a schematic diagram of the full-scenario state of charge prediction of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction of the present invention. Figure 6 This is a schematic diagram of a continuous-time SOC model of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction according to the present invention. Figure 7 This is a correlation analysis result diagram of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction of the present invention; Figure 8 This is a schematic diagram of the three-dimensional response surface of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction according to the present invention. Figure 9 This is a schematic diagram illustrating the predictive performance and error propagation analysis of a preferred embodiment of the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction of the present invention. Figure 10 This is a schematic diagram of the sensitivity analysis of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction of the present invention. Figure 11 This is a schematic diagram of the scatter matrix and regression trend of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction according to the present invention. Figure 12 This is a comprehensive evaluation schematic diagram of a preferred embodiment of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction of the present invention; Figure 13 This is a residual diagnosis result diagram of a preferred embodiment of the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction of the present invention; Figure 14 This is a residual diagnosis result diagram based on robustness analysis of a preferred embodiment of the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction of the present invention. Figure 15 This is a comparative analysis of the physical sensitivity of a preferred embodiment of the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction of the present invention. Figure 16 This is a statistical result diagram of the global coverage of a preferred embodiment of the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction of the present invention. Figure 17 This is a structural diagram of a preferred embodiment of the battery state estimation and dynamic control system that integrates electrothermal coupling mechanism and residual correction according to the present invention. Figure 18 This is a structural diagram of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] Existing pure data-driven SOC estimation methods use deep neural networks to directly fit and predict the SOC trajectory from end-to-end historical voltage and current data. Pure black-box models are extremely sensitive to high-frequency sampling noise from sensors in real operating environments and lack time-series smoothness constraints, which can easily lead to violent fluctuations in the output power display. In addition, when the battery is deeply aged or encounters extreme high or low temperatures or other data distributions that the model has not seen, pure data models are prone to divergence, producing prediction errors that violate the common sense of electrochemical physics, and have extremely poor robustness.

[0020] Furthermore, when a device's battery is low, the operating system typically triggers a power-saving mode using a static, one-size-fits-all approach, such as forcibly reducing screen brightness or drastically limiting CPU frequency. This static transition lacks global sensitivity and quantification regarding the power consumption rate of each hardware component. For example, drastically dimming the screen can severely damage the user's visual experience (a very high cost for the experience), but the absolute power saving may be negligible. Existing technologies cannot perform multi-objective mathematical optimization between "expected energy-saving benefits" and "user experience sacrifices," making it difficult to adapt to diverse user scenarios. This results in power-saving strategies often either excessively sacrifice the user experience or have minimal power-saving effects.

[0021] To address the aforementioned issues, this invention proposes a battery state estimation and dynamic control method based on electrothermal coupling mechanism and residual correction, which can operate stably under complex dynamic loads and sensor noise conditions.

[0022] The preferred embodiment of the present invention describes a battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction, such as... Figure 1 As shown, the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction includes the following steps: Step S10: Based on the relationship between the open-circuit voltage and the state of charge of the target battery, construct a semi-empirical polynomial model of the target battery, and calculate the actual discharge current of the target battery based on the semi-empirical polynomial model.

[0023] In the embodiments disclosed in this invention, a full-stack framework from low-level state estimation to high-level hardware scheduling is constructed, such as... Figure 2 As shown, the physical model of the target battery can be simplified into a mathematical model (i.e., the equivalent circuit model ECM); where the open-circuit voltage is... This represents the theoretical voltage of the battery under no-load conditions, determined by the electrochemical characteristics of the battery's positive and negative electrode materials, and is the core reference voltage of the model; multiple branches are used to simulate the electrochemical polarization and diffusion polarization processes of the battery, among which the resistor... and Both represent the activation polarization resistance of an electrochemical reaction, reflecting the rate of reaction and capacitance. and This corresponds to the electric double-layer capacitance on the electrode surface, reflecting the energy storage characteristics of the electrode-electrolyte interface. and Represent and The voltage; This represents ohmic resistance, simulating ohmic losses within the battery, including the resistance of electrode materials, electrolyte, and connectors, and corresponding to the instantaneous voltage drop during charging and discharging. Indicates its voltage, This represents the current. The microscopic physical structure of the battery on the left (including the porous structure of the electrodes, lithium-ion diffusion paths, etc.) is too complex to be directly used for engineering calculations. By simplifying the model, the microscopic electrochemical reactions, ion diffusion, and other processes are transformed into quantifiable voltage, resistance, and capacitance parameters in the circuit. This retains the dynamic characteristics of voltage and current during battery charging and discharging while significantly reducing computational complexity, making it suitable for use in battery management systems for scenarios such as state estimation and lifetime prediction.

[0024] Among them, such as Figure 3 As shown, the Multi-Physics Coupled Equivalent Model (MPCEM) proposed in this invention is divided into three layers. In the physical modeling layer, the traditional constant current assumption is abandoned, and the constant power feedback mechanism and electrothermal coupling mechanism of the mobile terminal are introduced to construct a continuous-time Ordinary Differential Equation of Charge (ODE) to solve the benchmark SOC with physical constraints. In the data compensation layer, the systematic error of the physical model is extracted, and a data-driven model including current, temperature and their polarization cross terms is constructed to perform residual compensation and output a high-precision, adaptive final SOC. Finally, in the multi-objective control layer, based on the high-precision SOC, global sensitivity analysis is used to quantify the power consumption factor, and a Pareto front is constructed in the two-dimensional space of "expected energy saving benefit" and "user experience loss" to intelligently optimize and issue the most efficient underlying hardware control command.

[0025] Specifically, such as Figure 3 As can be seen, the SOC solver describes the rate of change of the target battery's state of charge over time. This represents the derivative of SOC with respect to time, i.e., the rate of change of SOC; express The current at any given moment, when the current flows out of the battery (discharging), It is a positive value at this time. A negative value means that the SOC decreases over time; The effective capacitance related to temperature T indicates that the rate of change of SOC is also related to the effective capacitance at the temperature of the battery. The effective capacitance of the battery is different at different temperatures, which will affect the rate of change of SOC. and The termination condition for the target battery's operation is defined as follows: when the battery's state of charge drops to 0 or lower, it indicates that the battery has been depleted and operation needs to be stopped. This indicates the termination voltage of the target battery. When the battery voltage drops to 3.0V or lower (lower is undervoltage lockout), the target battery needs to be stopped to protect the target battery and related equipment.

[0026] Specifically, the open-circuit voltage of the target battery is determined, and the relationship between the open-circuit voltage and the normalized state of charge is determined to construct a semi-empirical polynomial model of the target battery: ; ; in, Indicates open-circuit voltage. Represents the normalized state of charge. Indicates the state of charge. Represents the natural constant; Based on the internal resistance and open-circuit voltage of the target battery, construct the terminal voltage of the target battery in a closed circuit, eliminate the terminal voltage, and then solve for the actual discharge current of the target battery: ; ; ; in, Indicates the first Terminal voltage at time , Indicates the battery's internal resistance. Indicates the actual discharge current. This indicates the constant power of the target battery.

[0027] The present invention, by constructing a semi-empirical polynomial model, aims to abandon the constant current assumption of traditional equivalent circuits and deeply mathematically couple this nonlinear open-circuit voltage model with the "constant power maintenance characteristic" of mobile terminals. Figure 4 As shown, Figure 4 (a) shows the characteristic curves of OCV-SOC. Figure 4(b) shows the nonlinear rise of current under constant power; based on this, it can be seen that when the SOC drops below 15%, the algorithm will automatically calculate that the discharge current will be forced to surge by more than 40%. This mechanism accurately reproduces the real polarization phenomenon from the underlying partial differential mathematics, and completely eliminates the "cliff-like power drop" and estimation collapse problems caused by ignoring constant power feedback when the traditional algorithm is low in power.

[0028] Step S20: Obtain temperature change information, construct an electrothermal coupling model between the effective capacity of the target battery and the temperature change information, construct a state of charge ordinary differential equation based on the actual discharge current and the electrothermal coupling model, and predict the current predicted state of charge of the target battery based on the state of charge ordinary differential equation.

[0029] In the embodiments disclosed in this invention, an electrothermal coupling model is established based on the Arrhenius equation to determine the change in the effective capacity of the battery caused by temperature changes. In order to obtain specific predicted values, the fourth-order Runge-Kutta (RK4) numerical integration algorithm with adaptive step size can be applied to solve the ordinary differential equation of the state of charge over time, thereby outputting the physical predicted state of charge at the current moment (i.e., the current predicted state of charge).

[0030] Specifically, the temperature change information of the target battery is obtained, and the relationship between the effective capacity of the target battery and the temperature change information is established based on the Arrhenius equation to construct an electrothermal coupling model of the target battery. ; in, Indicates temperature The effective capacity below, Indicates the rated capacity of the target battery. Indicates an indicator function, Indicates activation energy. Represents the ideal gas constant. Indicates the reference temperature. Indicates the first Surface temperature at any given time; Based on the actual discharge current and the electrothermal coupling model, an ordinary differential equation for the state of charge is constructed, and the equation is solved to obtain the current predicted state of charge of the target battery at the current moment. ; in, Indicates the first The current predicted state of charge at time t, Indicates the first The actual discharge current at any given moment.

[0031] Among them, such as Figure 5 As shown, the SOC prediction results for the entire scenario are presented. Figure 5 (a) shows the SOC prediction results for standby mode. Figure 5 (b) shows the SOC prediction results for music playback status. Figure 5 (c) shows the SOC prediction results for the video playback status. Figure 5 (d) shows the SOC prediction results for web browsing status. Figure 5 (e) shows the SOC prediction results for the navigation state. Figure 5 (f) in the figure shows the SOC prediction results for the game state; such as Figure 6 As shown, Figure 6 (a) shows a comparison of SOC predictions for video playback status. Figure 6 (b) shows the time series of prediction errors. Figure 6 (c) in the figure shows the power consumption time series. Figure 6 (d) in the figure shows the rate of change of SOC.

[0032] pass Figure 5 and Figure 6 It is evident that although the purely physical ODE model provides a smooth, anti-divergence foundation, due to the complex nonlinear polarization effects in actual batteries, the purely physical open-loop integral will inevitably produce systematic errors that accumulate linearly over time (such as...). Figure 6 (The error area in (b) increases over time). This objective physical phenomenon directly proves that the ultimate dynamic tracking accuracy cannot be achieved by relying solely on a pure physical model. Therefore, in the embodiments disclosed in this invention, the current predicted state of charge is further used as a "smooth reference base" with high robustness, and in subsequent steps, a data-driven model is used to accurately "compensate" the systematic cumulative error of the system, thereby achieving the complementary advantages of physics and data.

[0033] Step S30: Construct the interaction features of the target battery at the current moment, input the interaction features into the machine learning compensation model for prediction, output the residual estimate of the target battery at the current moment, and update the current predicted state of charge based on the residual estimate to obtain the final state of charge.

[0034] In this process, the main steps are to correct the error in the current predicted state of charge; extract the real-time discharge current and surface temperature at the current moment; and construct a physical cross-interaction term to characterize the degree of concurrent polarization. Then, real-time discharge current, surface temperature, and physical cross-interaction terms are constructed as interactive features. As input, it is fed into a pre-trained machine learning compensation model (such as a multidimensional response surface model or a deep neural network).

[0035] Specifically, the current current and current temperature of the target battery are extracted, cross-interaction terms are constructed based on the current current and current temperature, and the interaction features of the target battery at the current moment are constructed based on the current current, current temperature and cross-interaction terms. The interaction features are input into a pre-trained machine learning compensation model, which performs a weighted calculation on the interaction features based on a nonlinear mapping function, and outputs the residual estimate of the target battery at the current time. The current predicted state of charge is updated using the residual estimate to obtain the final state of charge of the target battery at the current moment; ; in, Indicates the first The final state of charge at time t, Indicates the first The current predicted state of charge at time t, Indicates the first The residual estimate at time step 1.

[0036] In the machine learning compensation model, the input single features and cross features are weighted and calculated using an internal weight matrix or nonlinear mapping function. Utilizing cross-interaction terms, the machine learning compensation model can accurately extract and quantify the complex electrothermal coupling effect of "high-current discharge inducing high Joule heating, leading to a nonlinear abrupt change in internal resistance." After this mapping process, the model outputs a residual estimate for the current moment. This residual estimate is then superimposed on a physical reference to obtain a highly accurate final state of charge.

[0037] Through this process, the present invention compensates for the simplification errors of the lumped parameter physical model, such as... Figure 5 As shown, in multiple dynamic polarity scenarios (such as games and navigation), after combining residual compensation, the mean absolute error is steadily controlled within 1.5%, and it maintains extremely high generalization ability across decay states.

[0038] In another embodiment of the present invention, a correlation analysis was performed on the feature variables, such as... Figure 7 The above, Figure 7 (a) shows a bar chart illustrating the importance of multiple feature variables. Figure 7 (b) shows a pie chart illustrating the importance of multiple feature variables. Figure 7 (c) shows a heatmap illustrating the importance of multiple feature variables. Figure 7(a) shows a three-scatter plot of the importance of multiple characteristic variables; it can be seen that the current is the most important influencing factor, while the differential of the current (i.e., dI / dt) accounts for only 5%.

[0039] Furthermore, such as Figure 8 The diagram shows the three-dimensional response surface plot of the SOC-current-temperature coupling, which provides a more intuitive understanding of the importance of factors affecting the state of charge. Figure 8 (a) and Figure 8 (b) shows the three-dimensional response surfaces of different faces.

[0040] Further as Figure 9 As shown, Figure 9 (a) shows the total current sequence results. Figure 9 (b) shows the temperature series results. Figure 9 (c) shows the results of the charged state sequence. Figure 9 (d) in the figure shows the error sequence results.

[0041] In training the machine learning compensation model, the training dataset comes from real charge-discharge aging tests of batteries under standardized constant temperature chambers and multiple operating conditions. The real battery SOC measured by high-precision equipment is used as the real label. The predicted physical SOC output by the physical ODE solver is subtracted from the real label to obtain the systematic residual label. Using collected historical current, temperature, and cross-polarization terms as feature inputs, the gradient descent algorithm is used to continuously optimize the model weights until the loss function (Loss) between the predicted residual and the residual label converges, completing the model deployment.

[0042] Step S40: Construct a time-varying model of the target battery, randomly perturb and sample all parameters in the time-varying model to obtain the current status of each parameter, and construct a corresponding power-saving decision based on all the current statuses and the final state of charge.

[0043] In another embodiment of the present invention, for a mobile terminal, a time-varying model of instantaneous total power consumption is first constructed, which is represented as the superposition of component-level loads (such as screen, CPU, and network), and a random fluctuation term is added; for example, for the core component CPU, its dynamic power consumption physical model (i.e., the time-varying model mentioned above) is represented as: ; in, This indicates the CPU's dynamic power consumption. Indicates activity factor, This represents the equivalent capacitance. Indicates the operating frequency. Indicates the core power supply voltage. This indicates the power consumption due to error.

[0044] Among them, a clear mathematical mapping was established between the "battery discharge current" of the black box and the specific "mobile smart terminal user interaction tasks (such as adjusting brightness, CPU frequency reduction)", which provides the underlying boundary conditions for subsequent accurate evaluation of the actual physical benefits of power saving strategies.

[0045] Specifically, a time-varying model of the instantaneous total power consumption of the target battery is constructed based on multiple component-level loads of the target battery. Random perturbation sampling is performed on all parameters in the time-varying model to calculate the average image effect and nonlinear interaction effect for each parameter; Based on the average image effect and the nonlinear interaction effect of each parameter, a core driving factor for each parameter is constructed, wherein the core driving factor represents the factor that plays a decisive role in the power consumption of the target battery. Based on all the current core driving factors and the current final state of charge, construct a power-saving decision for each parameter.

[0046] Specifically, the scheduling parameters defined in the time-varying model (including CPU operating frequency, screen display brightness, network operation mode, and number of background processes, etc.) are extracted and randomly perturbed. Using the global sensitivity analysis method, the average impact effect and nonlinear interaction effect of each parameter are calculated, thereby separating the core driving factors that directly and linearly determine power consumption (such as CPU load, which is characterized by a high average impact effect value) and nonlinear interference terms that are complexly coupled with other factors (such as battery temperature, which is characterized by a high nonlinear interaction effect value).

[0047] Furthermore, based on the core driving factors, it can be determined which factor is causing the current operational anomaly and requires adjustment, thereby generating power-saving decisions corresponding to each core driving factor. When generating power-saving decisions, it is not necessary to consider the importance weight of each factor; that is, each factor needs to generate a corresponding power-saving decision. For example, even if screen display brightness has the lowest importance, a power-saving decision to adjust "screen display brightness" is still generated and subsequently mapped into the decision space to provide users with more comprehensive choices.

[0048] Step S50: Perform a comprehensive evaluation of all the power-saving decisions to obtain the final decision with the highest efficiency ratio, and convert the final decision into a low-level hardware control instruction, and issue the low-level hardware control instruction to adjust the current state of the hardware.

[0049] Specifically, all the core driving factors are mapped into a decision space, and the similarity degree corresponding to each parameter is calculated in the decision space; All the core driving factors are sorted according to their similarity, and the power-saving decision corresponding to the core driving factor with the highest similarity is defined as the final decision. The final decision is converted into underlying hardware control instructions, which are then sent to the hardware execution module to adjust the current state of multiple hardware devices.

[0050] After obtaining power-saving decisions for all factors, a two-dimensional decision space is constructed, incorporating both "expected energy-saving benefits" and "user experience loss." Multiple candidate power-saving strategies, such as "reducing CPU frequency," "limiting display module luminous intensity," and "dynamically adjusting RF antenna transmission power," are mapped into this decision space. A Pareto front is fitted to the outer edge to evaluate all power-saving decisions, automatically eliminating dominated, inferior strategies (such as blindly and drastically reducing screen brightness, which results in significant user experience loss but minimal actual power-saving benefits) that fall below the front. Only the set of non-dominated optimal solutions on the front is extracted, and a quantitative evaluation is performed using an entropy-weighted algorithm. Specifically, the evaluation process involves: first, objectively assigning weights based on the dispersion of each evaluation index using the entropy-weighted method; then, calculating the relative proximity of each candidate strategy to the ideal solution (i.e., maximizing energy saving while minimizing user experience loss) using the TOPSIS method, and prioritizing them according to their proximity.

[0051] Furthermore, we define the expected energy-saving benefits and the user experience loss, and construct a two-dimensional decision space based on the expected energy-saving benefits and the user experience loss; Based on the average image effect and the nonlinear interaction effect of each core driving factor, each core driving factor is mapped into the decision space, and all core driving factors in the preset region below the decision space are removed. For all core driving factors within a preset region above the decision space, the degree of dispersion of each core driving factor is calculated using the entropy weight method, and a weight is assigned to each core driving factor based on the degree of dispersion. Based on each of the weights, the relative proximity of each core driving factor to the ideal solution is calculated.

[0052] In another embodiment of this invention, based on the priority ranking result, the highest-ranked strategy is converted into specific underlying hardware control instructions and executed. For example, when the terminal is in a "low-battery survival scenario" with less than 20% battery, the system, after the aforementioned Pareto optimization and ranking, determines that "limiting the CPU" and "switching the network" are the optimal solutions in terms of overall performance. The system then generates control instructions to dynamically adjust the dynamic voltage frequency adjustment level of the CPU or GPU through the power management chip (PMIC) to reduce the core power supply voltage, and simultaneously switches the radio frequency antenna from the high-power 5G network mode to the low-power 4G or Wi-Fi mode through the baseband chip. The hardware instructions also include, but are not limited to, limiting the maximum refresh rate and peak nits brightness of the screen through the display driver IC (Display Driver IC).

[0053] Among them, such as Figure 10 As shown, Figure 10 (a) in the diagram shows the sensitivity ranking. Figure 10 (b) shows a bubble chart of the sensitivity mean. Figure 10 (c) in the figure shows the significance test plot for the parameters; through Figure 10 It can be seen that this method can accurately identify that CPU load is the absolute main factor determining power consumption (with an extremely high average impact effect), while temperature is a nonlinear interference factor that causes system instability (with an extremely high nonlinear interaction effect), providing scientific data support for subsequent hardware scheduling.

[0054] Furthermore, such as Figure 11 As shown, Figure 11 (a) in the image shows the screen brightness distribution; Figure 11 (b) shows the relationship between processor load and SOC; Figure 11 (c) in the diagram illustrates the relationship between network patterns and SOC; Figure 11 (d) in the figure shows the relationship between screen brightness and SOC; Figure 11 (e) in the diagram shows the distribution of processor load; Figure 11 (f) in the diagram illustrates the relationship between network patterns and SOC; Figure 11 (g) in the figure shows the relationship between screen brightness and SOC; Figure 11 The relationship between processor load and SOC in (h); Figure 11 (i) in the diagram shows the distribution of network patterns; the distribution of these measured data further validates the above-mentioned sensitivity results.

[0055] Furthermore, such as Figure 12 As shown, Figure 12 (a) shows the policy priority ranking. Figure 12 (b) shows a multi-dimensional evaluation comparison of the top three strategies. Figure 12(c) shows the results of the cost-benefit analysis. Figure 12 (d) in the diagram shows the scenario-based strategy recommendation matrix; Figure 12 The Pareto front constructed in this invention is clearly demonstrated. Through this front, the system can automatically filter out the disadvantageous strategies of "high consumption and low benefit" located in the lower left corner, ensuring that every instruction finally issued to the hardware is at the optimal balance point of "energy saving and experience".

[0056] Furthermore, in the set of non-dominated optimal solutions on the Pareto front, the EWM-TOPSIS (entropy weighted approximation ideal solution sorting) algorithm is used for comprehensive evaluation to extract the strategy with the highest performance ratio. This strategy is then transformed into specific low-level hardware control instructions (such as dynamically adjusting the CPU's dynamic voltage and frequency adjustment levels via PMIC, switching network modes, etc.) and issued for execution. Through this process, the present invention completes the entire logical chain from state estimation to closed-loop control, always distributing the hardware instructions with the best performance, maximizing battery life without sacrificing the core experience.

[0057] To further verify the reliability of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction proposed in this invention, a variety of verification experiments were conducted.

[0058] like Figure 13 As shown, Figure 13 (a) in the figure shows the cumulative distribution function of the relative error. Figure 13 (b) shows the error versus the disturbance intensity. Figure 13 (c) in the diagram shows a scatter plot of the correlation; in Figure 13 Based on this, after adding 10% noise, the following was obtained: Figure 14 The robustness analysis results shown are as follows: Figure 14 (a) shows the predicted trend of current disturbance SOC. Figure 14 (b) shows the distribution of MSE (Mean-Square Error) under different perturbations. Figure 14 (c) in the diagram shows the relative error distribution. Figure 14 (d) in the figure shows the composite perturbation error surface. Figure 14 (e) in the figure shows a comparison of time series errors. Figure 14 (f) in the diagram shows a comparison of radar charts with multiple indicators; Figure 13 and Figure 14 The results all indicate that the model disclosed in this invention can operate stably under complex dynamic loads and sensor noise conditions.

[0059] Furthermore, such as Figure 15 As shown, this further verifies the model's physical sensitivity to extreme temperatures and aging; Figure 15(a) shows the effect of temperature changes on model accuracy. Figure 15 (b) shows the impact of battery aging on model accuracy. Figure 15 (c) in the figure shows a comprehensive sensitivity comparison; based on Figure 15 ,like Figure 16 As shown, the global coverage statistics of the validation dataset are presented. Figure 16 (a) shows the statistical results of the data used. Figure 16 (b) in the diagram shows the coverage of the inspection dimensions.

[0060] This invention integrates constant power feedback decoupling, continuous-time ordinary differential equation solving of the state of charge, residual-driven compensation, and multi-objective optimization to construct a full-stack framework from bottom-level state estimation to top-level hardware scheduling, which can operate stably under complex dynamic loads and sensor noise conditions.

[0061] Furthermore, such as Figure 17 As shown, based on the above-mentioned battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction, the present invention also provides a battery state estimation and dynamic control system integrating electrothermal coupling mechanism and residual correction, wherein the battery state estimation and dynamic control system integrating electrothermal coupling mechanism and residual correction includes: The mechanism reconstruction module 51 is used to construct a semi-empirical polynomial model of the target battery based on the relationship between the open-circuit voltage and the state of charge of the target battery, and to calculate the actual discharge current of the target battery based on the semi-empirical polynomial model. The initial state of charge prediction module 52 is used to acquire temperature change information, construct an electrothermal coupling model between the effective capacity of the target battery and the temperature change information, construct a state of charge ordinary differential equation based on the actual discharge current and the electrothermal coupling model, and predict the current predicted state of charge of the target battery based on the state of charge ordinary differential equation. The final state of charge prediction module 53 is used to construct the interaction features of the target battery at the current time, input the interaction features into the machine learning compensation model for prediction, output the residual estimate of the target battery at the current time, and update the current predicted state of charge according to the residual estimate to obtain the final state of charge. The decision building module 54 is used to build a time-varying model of the target battery, randomly perturb and sample all parameters in the time-varying model to obtain the current status of each parameter, and build a corresponding power-saving decision based on all the current statuses and the final state of charge. The decision positioning module 55 is used to comprehensively evaluate all the power-saving decisions, obtain the final decision with the highest efficiency ratio, convert the final decision into a low-level hardware control instruction, and issue the low-level hardware control instruction to adjust the current state of the hardware.

[0062] Furthermore, such as Figure 18 As shown, based on the above-mentioned battery state estimation and dynamic control method and system that integrates electrothermal coupling mechanism and residual correction, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 18 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0063] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc. Further, the memory 20 may include both internal and external storage devices. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a battery state estimation and dynamic control program 40 that integrates electrothermal coupling mechanism and residual correction. This battery state estimation and dynamic control program 40 that integrates electrothermal coupling mechanism and residual correction can be executed by the processor 10, thereby realizing the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction in this application.

[0064] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction.

[0065] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface. The components of the terminal communicate with each other via a system bus.

[0066] In one embodiment, when the processor 10 executes the battery state estimation and dynamic control program 40 that integrates electrothermal coupling mechanism and residual correction in the memory 20, it implements the steps of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction as described above.

[0067] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction, and the battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction implements the steps of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction as described above when executed by a processor.

[0068] In summary, this invention provides a battery state estimation and dynamic control method and related equipment that integrates electrothermal coupling mechanism and residual correction. The method includes: acquiring development data and decision data of all cities within a preset area within a preset time period; quantifying the decision data to obtain a quantization result; determining multiple nodes based on the quantization result and inputting the development data and decision data into the corresponding nodes; constructing a simulation network based on the nodes; analyzing the development data and decision data in all the nodes to obtain analysis results; obtaining correlation information between the multiple nodes based on the analysis results; determining associated nodes based on the correlation information; determining multiple sets of node connection relationships based on the multiple associated nodes to obtain the number of node connection relationship groups; calculating the degree of correlation between the multiple nodes based on the correlation information and the number of node connection relationship groups; calculating the average betweenness centrality of the simulation network based on the degree of correlation; constructing a result analysis model; inputting the development data, the number of nodes, the number of node connection relationship groups, and the average betweenness centrality into the result analysis model; and outputting the simulation results of urban development differences. This invention integrates constant power feedback decoupling, continuous-time ordinary differential equation solving of the state of charge, residual-driven compensation, and multi-objective optimization to construct a full-stack framework from bottom-level state estimation to top-level hardware scheduling, which can operate stably under complex dynamic loads and sensor noise conditions.

[0069] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0070] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The computer-readable storage medium can be a memory, magnetic disk, optical disk, etc.

[0071] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A battery state estimation and dynamic control method that fuses electro-thermal coupling mechanism and residual correction, characterized in that, The battery state estimation and dynamic control method that integrates the electrothermal coupling mechanism and residual correction includes: Based on the relationship between the open-circuit voltage and the state of charge of the target battery, a semi-empirical polynomial model of the target battery is constructed, and the actual discharge current of the target battery is calculated based on the semi-empirical polynomial model. Acquire temperature change information, construct an electrothermal coupling model between the effective capacity of the target battery and the temperature change information, construct a state of charge ordinary differential equation based on the actual discharge current and the electrothermal coupling model, and predict the current predicted state of charge of the target battery based on the state of charge ordinary differential equation. The interaction features of the target battery at the current moment are constructed, the interaction features are input into the machine learning compensation model for prediction, the residual estimate of the target battery at the current moment is output, and the current predicted state of charge is updated according to the residual estimate to obtain the final state of charge. A time-varying model of the target battery is constructed, and all parameters in the time-varying model are randomly perturbed and sampled to obtain the current status of each parameter. Based on all the current statuses and the final state of charge, a corresponding power-saving decision is constructed. All power-saving decisions are comprehensively evaluated to obtain the final decision with the highest efficiency ratio. The final decision is then converted into a low-level hardware control instruction and issued to adjust the current state of the hardware.

2. The battery state estimation and dynamic control method of claim 1, wherein The step of constructing a semi-empirical polynomial model of the target battery based on the relationship between its open-circuit voltage and state of charge, and calculating the actual discharge current of the target battery based on the semi-empirical polynomial model, specifically includes: Determine the open-circuit voltage of the target battery and the relationship between the open-circuit voltage and the normalized state of charge to construct a semi-empirical polynomial model of the target battery: ； ； wherein, represents open circuit voltage, represents normalized state of charge, represents state of charge, represents natural constant; Based on the internal resistance and open-circuit voltage of the target battery, construct the terminal voltage of the target battery in a closed circuit, eliminate the terminal voltage, and then solve for the actual discharge current of the target battery: ； ； ； in, Indicates the first Terminal voltage at time , Indicates the battery's internal resistance. Indicates the actual discharge current. This indicates the constant power of the target battery.

3. The battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction according to claim 1, characterized in that, The process of acquiring temperature change information, constructing an electrothermal coupling model between the effective capacity of the target battery and the temperature change information, constructing a state-of-charge (SOC) ordinary differential equation based on the actual discharge current and the electrothermal coupling model, and predicting the current predicted SOC of the target battery based on the SOC ordinary differential equation specifically includes: The temperature change information of the target battery is obtained, and the relationship between the effective capacity of the target battery and the temperature change information is established based on the Arrhenius equation to construct an electrothermal coupling model of the target battery. ； in, Indicates temperature The effective capacity below, Indicates the rated capacity of the target battery. Indicates an indicator function, Indicates activation energy. Represents the ideal gas constant. Indicates the reference temperature. Indicates the first Surface temperature at any given time; Based on the actual discharge current and the electrothermal coupling model, an ordinary differential equation for the state of charge is constructed, and the equation is solved to obtain the current predicted state of charge of the target battery at the current moment. ； in, Indicates the first The current predicted state of charge at time t, Indicates the first The actual discharge current at any given moment.

4. The battery state estimation and dynamic control method of claim 1, wherein The process of constructing the interaction features of the target battery at the current moment, inputting the interaction features into a machine learning compensation model for prediction, outputting the residual estimate of the target battery at the current moment, and updating the current predicted state of charge based on the residual estimate to obtain the final state of charge, specifically includes: Extract the current current and current temperature of the target battery, construct cross-interaction terms based on the current current and current temperature, and construct the interaction features of the target battery at the current moment based on the current current, current temperature and cross-interaction terms; The interaction features are input into a pre-trained machine learning compensation model, which performs a weighted calculation on the interaction features based on a nonlinear mapping function, and outputs the residual estimate of the target battery at the current time. The current predicted state of charge is updated using the residual estimate to obtain the final state of charge of the target battery at the current moment; ； in, Indicates the first The final state of charge at time t, Indicates the first The current predicted state of charge at time t, Indicates the first The residual estimate at time step 1.

5. The battery state estimation and dynamic control method of claim 1, wherein The construction of the time-varying model of the target battery involves randomly perturbing and sampling all parameters in the time-varying model to obtain the current state of each parameter. Based on all the current states and the final state of charge, a corresponding power-saving decision is constructed, specifically including: Based on the multiple component-level loads of the target battery, a time-varying model of the instantaneous total power consumption of the target battery is constructed. Random perturbation sampling is performed on all parameters in the time-varying model to calculate the average image effect and nonlinear interaction effect for each parameter; Based on the average image effect and the nonlinear interaction effect of each parameter, a core driving factor for each parameter is constructed, wherein the core driving factor represents the factor that plays a decisive role in the power consumption of the target battery. Based on all the current core driving factors and the current final state of charge, construct a power-saving decision for each parameter.

6. The battery state estimation and dynamic control method of claim 5, wherein the residual correction is performed by using a Kalman filter. The process involves comprehensively evaluating all power-saving decisions to obtain the final decision with the highest efficiency ratio, and then converting the final decision into underlying hardware control instructions. These instructions are then issued to adjust the current state of the hardware. Specifically, this includes: All the core driving factors are mapped to a decision space, and the similarity degree corresponding to each parameter is calculated in the decision space; All the core driving factors are sorted according to their similarity, and the power-saving decision corresponding to the core driving factor with the highest similarity is defined as the final decision. The final decision is converted into underlying hardware control instructions, which are then sent to the hardware execution module to adjust the current state of multiple hardware devices.

7. The battery state estimation and dynamic control method of claim 6, wherein the battery state estimation and dynamic control method is characterized by: The step of mapping all the core driving factors to a decision space and calculating the proximity degree corresponding to each parameter in the decision space specifically includes: Define the expected energy-saving benefits and the user experience loss, and construct a two-dimensional decision space based on the expected energy-saving benefits and the user experience loss; Based on the average image effect and the nonlinear interaction effect of each core driving factor, each core driving factor is mapped into the decision space, and all core driving factors in the preset region below the decision space are removed. For all core driving factors within a preset region above the decision space, the degree of dispersion of each core driving factor is calculated using the entropy weight method, and a weight is assigned to each core driving factor based on the degree of dispersion. Based on each of the weights, the relative proximity of each core driving factor to the ideal solution is calculated.

8. A battery state estimation and dynamic control system that fuses electro-thermal coupling mechanism and residual correction, characterized by, The battery state estimation and dynamic control system integrating electrothermal coupling mechanism and residual correction is used to implement the battery state estimation and dynamic control method integrating electrothermal coupling mechanism and residual correction as described in any one of claims 1-7. The battery state estimation and dynamic control system integrating electrothermal coupling mechanism and residual correction includes: The mechanism reconstruction module is used to construct a semi-empirical polynomial model of the target battery based on the relationship between the open-circuit voltage and the state of charge of the target battery, and to calculate the actual discharge current of the target battery based on the semi-empirical polynomial model. The initial state of charge prediction module is used to acquire temperature change information, construct an electrothermal coupling model between the effective capacity of the target battery and the temperature change information, construct a state of charge ordinary differential equation based on the actual discharge current and the electrothermal coupling model, and predict the current predicted state of charge of the target battery based on the state of charge ordinary differential equation. The final state of charge prediction module is used to construct the interaction features of the target battery at the current moment, input the interaction features into the machine learning compensation model for prediction, output the residual estimate of the target battery at the current moment, and update the current predicted state of charge based on the residual estimate to obtain the final state of charge. The decision-making construction module is used to construct a time-varying model of the target battery, randomly perturb and sample all parameters in the time-varying model to obtain the current status of each parameter, and construct a corresponding power-saving decision based on all the current statuses and the final state of charge. The decision positioning module is used to comprehensively evaluate all the power-saving decisions, obtain the final decision with the highest efficiency ratio, and convert the final decision into underlying hardware control instructions, and issue the underlying hardware control instructions to adjust the current state of the hardware.

9. A terminal, characterized by comprising: The terminal includes: a memory, a processor, and a battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction, stored in the memory and executable on the processor. When the battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction is executed by the processor, it implements the steps of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction. When the battery state estimation and dynamic control program that integrates electrothermal coupling mechanism and residual correction is executed by a processor, it implements the steps of the battery state estimation and dynamic control method that integrates electrothermal coupling mechanism and residual correction as described in any one of claims 1-7.

Images