High-precision electric quantity monitoring method and system for series battery based on adaptive algorithm

Abstract

The application relates to the technical field of battery health management, in particular to a series battery high-precision power monitoring method and system based on an adaptive algorithm, which comprises the following steps: acquiring multi-dimensional parameter data of a series battery pack; running a main state estimator based on an electro-thermal coupling equivalent circuit model, and obtaining a baseline state of charge value through online estimation according to the multi-dimensional parameter data; adopting a probability error prediction model to predict the estimation error of the baseline state of charge value based on the multi-dimensional parameter data, and outputting error probability distribution; generating a final high-fidelity state of charge estimation value by using the error probability distribution through a double-channel adaptive collaborative correction process; and constructing and driving a battery protection model by using the final high-fidelity state of charge estimation value and the multi-dimensional parameter data, and outputting a safety control instruction for controlling and protecting the series battery pack. The application realizes accurate power monitoring and active safety protection by using an adaptive algorithm.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of battery health management, in particular to a series battery high-precision power monitoring method and system based on an adaptive algorithm. BACKGROUND

[0002] The state of charge of a battery is a key core parameter in a battery management system, and accurate estimation thereof is crucial for guaranteeing the safety of the battery system and prolonging the service life.

[0003] At present, mainstream state of charge estimation methods for batteries include ampere-hour integration and model-based filtering algorithms (such as Kalman filtering). Although the ampere-hour integration method is simple to implement, the open-loop calculation method thereof is susceptible to current measurement noise and initial SOC value errors, leading to cumulative estimation errors over time. In addition, when establishing an equivalent model of a battery, the traditional model-based method is difficult to comprehensively and accurately reflect the complex electrochemical and thermodynamic dynamic characteristics of the battery in actual operation. Therefore, in the face of variable dynamic working conditions and significant temperature fluctuations, the accuracy and robustness of existing state of charge estimation methods for batteries still have room for improvement, and it is difficult to fully meet the application requirements of high-precision real-time monitoring of battery power.

[0004] Therefore, a series battery high-precision power monitoring method and system based on an adaptive algorithm are provided. SUMMARY

[0005] The application aims to provide a series battery high-precision power monitoring method and system based on an adaptive algorithm, which realizes accurate power monitoring and active safety protection by using an adaptive algorithm. Multi-dimensional parameter data of a series battery pack are acquired; a main state estimator is run based on an electro-thermal coupled equivalent circuit model, and a baseline state of charge value is obtained by online estimation according to the multi-dimensional parameter data; a probability error prediction model is used to predict the estimation error of the baseline state of charge value based on the multi-dimensional parameter data, and output an error probability distribution; a final high-fidelity state of charge estimation value is generated by using the error probability distribution through a double-channel adaptive collaborative correction process; and a battery protection model is constructed and driven by using the final high-fidelity state of charge estimation value and the multi-dimensional parameter data, and a safety control instruction is output for controlling and protecting the series battery pack.

[0006] To achieve the above-mentioned purpose, the application provides the following technical solutions:

[0007] The series battery high-precision power monitoring method based on an adaptive algorithm comprises the following steps:

[0008] Multi-dimensional parameter data of a series battery pack are acquired, and the multi-dimensional parameter data include the terminal voltage, current and temperature of each single battery;

[0009] a main state estimator running based on an electro-thermal coupled equivalent circuit model, the main state estimator adopting a Kalman filter algorithm to estimate a baseline state of charge value according to the multi-dimensional parameter data;

[0010] an error probability distribution including an error mean and a variance for representing a prediction confidence is outputted by predicting an estimation error of the baseline state of charge value based on the multi-dimensional parameter data using a probabilistic error prediction model, and a final high-fidelity state of charge estimation value is generated by using the error probability distribution through a double-channel adaptive collaborative correction process;

[0011] a battery protection model is constructed and driven using the final high-fidelity state of charge estimation value and the multi-dimensional parameter data, and a safety control instruction is outputted for controlling and protecting the series battery pack.

[0012] Preferably, the electro-thermal coupled equivalent circuit model comprises:

[0013] a temperature-sensitive parameter adaptation layer, which internally contains a parameter library storing a preset multi-dimensional mapping relationship between circuit parameters and temperature, receives a real-time temperature as a query input, retrieves and outputs complete circuit parameters corresponding to the current temperature from the parameter library;

[0014] an electrical state calculation layer, which receives the circuit parameters, updates state variables through integral operation according to the input real-time current, and calculates a predicted terminal voltage value according to Kirchhoff's voltage law;

[0015] a coupled heat generation calculation layer, which is used for calculating total heat power generated by electrical activity, receives real-time current and internal resistance value and entropy heat coefficient corresponding to the current temperature, respectively calculates irreversible Joule heat and reversible entropy heat, sums the two, and outputs a total heat generation power value.

[0016] Preferably, the process in which the main state estimator adopts a Kalman filter algorithm to estimate a baseline state of charge value according to the multi-dimensional parameter data comprises: obtaining a state prior prediction value outputted after forward operation by the electro-thermal coupled equivalent circuit model, the state prior prediction value containing a prior state of charge and a model-predicted terminal voltage value; comparing the model-predicted terminal voltage with a real battery terminal voltage sampled synchronously by hardware and calculating a measurement residual, calculating an optimal Kalman gain according to the measurement residual and an error covariance matrix, and generating the baseline state of charge value by using the optimal Kalman gain to weight and correct the state prior prediction value.

[0017] Preferably, the probabilistic error prediction model comprises:

[0018] The state feedback and working condition fusion perception layer is used to receive real-time multi-dimensional parameter data synchronized with the main state estimator, and at the same time receive the diagonal elements of the optimal Kalman gain and error covariance matrix output by the main state estimator at the previous moment, and fuse them into a high-dimensional feature vector; the deep coupling error inference layer is used to model and predict the estimation error of the main state estimator at the current moment based on the high-dimensional feature vector; the confidence distribution generation layer is used to parse and encapsulate the prediction results of the deep coupling error inference layer, and output an error probability distribution containing the error mean and the variance used to characterize the prediction confidence.

[0019] Preferably, the process of generating a final high-fidelity state of charge estimation value by using the error probability distribution through a dual-channel adaptive collaborative correction process includes:

[0020] Based on the error probability distribution, which includes error variance and error mean, calculations of two correction channels are performed; in the output-end gated correction channel, the error variance is converted into a confidence fusion weight through a preset inverse function, the fusion weight is multiplied by the error mean to obtain a weighted error correction, and the weighted error correction is compensated to the baseline state of charge value to generate a final high-fidelity state of charge estimation value; in the model-end online adaptive channel, the error variance is used as a dynamic indicator of the mismatch degree of the electrothermal coupling equivalent circuit model, and is mapped to a process noise adjustment value through a preset monotonically increasing function, and the noise adjustment value is superimposed on the noise covariance matrix of the Kalman filter algorithm.

[0021] Preferably, the process of using the final high-fidelity state of charge estimation value and the multi-dimensional parameter data to construct and drive a battery protection model and output a safety control instruction includes:

[0022] The final high-fidelity state of charge estimate and the real-time temperature and battery health status information in the multi-dimensional parameter data are input as a combined feature vector into a neural network model for inferring safety boundaries to generate a safe operating boundary for the battery in the current state. The safe operating boundary includes a charging cutoff SOC threshold, a discharging cutoff SOC threshold, a maximum charge and discharge current allowed under the current operating conditions, and a maximum / minimum operating temperature that are dynamically adjusted according to the state. The final high-fidelity state of charge estimate and the real-time current and temperature are compared with the safe operating boundary. If it is detected that the real-time parameters have a tendency to cross the corresponding safety boundary, the corresponding protection logic is triggered, and a safety control instruction is generated to directly control the charging and discharging circuit.

[0023] High-precision battery capacity monitoring system for series-connected batteries based on adaptive algorithms, including:

[0024] A data acquisition module acquires multi-dimensional parameter data of the series battery pack, the multi-dimensional parameter data including terminal voltage, current and temperature of each single battery;

[0025] A state estimation module runs a main state estimator based on an electro-thermal coupled equivalent circuit model, the main state estimator adopting a Kalman filtering algorithm to obtain a baseline state of charge value through online estimation according to the multi-dimensional parameter data;

[0026] An error correction module adopts a probabilistic error prediction model to predict an estimation error of the baseline state of charge value based on the multi-dimensional parameter data, and outputs an error probability distribution including an error mean value and a variance used for representing a prediction confidence; through a double-channel adaptive collaborative correction process, the error probability distribution is used to generate a final high-fidelity state of charge estimation value;

[0027] A safety control module uses the final high-fidelity state of charge estimation value and the multi-dimensional parameter data to construct and drive a battery protection model, and outputs a safety control instruction, the safety control instruction being used for controlling and protecting the series battery pack.

[0028] Compared with the prior art, the present application has the following beneficial effects:

[0029] 1. By adopting an electro-thermal coupled equivalent circuit model, the influence of temperature on the internal characteristics of the battery is more accurately described, which helps to reduce the risk of model mismatch caused by temperature change, and provides a more reliable model basis for subsequent accurate estimation.

[0030] 2. By establishing a probabilistic error prediction model, the system has the ability to actively evaluate its estimation uncertainty, which can provide a quantitative reference for the reliability of the current SOC estimation result, thereby providing a clear decision basis for subsequent adaptive correction.

[0031] 3. By executing a double-channel adaptive collaborative correction process, the predicted error information is used to synchronously correct the output value and adjust the model parameters online, which helps to suppress the long-term accumulation of errors and improves the ability of the power monitoring system to maintain high precision in different environments.

[0032] 4. By constructing and driving a battery protection model linked with the high-fidelity SOC estimation value, a mechanism for dynamically generating a safety boundary is provided to replace the traditional fixed threshold protection method, which helps to better utilize the available capacity and performance of the battery under the premise of ensuring safety, thereby having a positive effect on improving the safety and operating efficiency of the battery throughout its life cycle. BRIEF DESCRIPTION OF DRAWINGS

[0033] Fig. 1 A flowchart of a series battery high-precision power monitoring method based on an adaptive algorithm provided by an embodiment of the present application;

[0034] Fig. 2 A double-channel adaptive collaborative correction process schematic diagram is provided for the embodiment of the present application.

[0035] Fig. 3 A high-precision electric quantity monitoring system structure schematic diagram based on an adaptive algorithm for a series battery is provided for the embodiment of the present application. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0037] Please refer to Figs. 1 to 3 The present application provides a high-precision electric quantity monitoring method and system for a series battery based on an adaptive algorithm, and the technical solutions are as follows:

[0038] Embodiment one:

[0039] The high-precision electric quantity monitoring method for a series battery based on an adaptive algorithm, the specific process is as shown in Fig. 1 , which comprises:

[0040] Obtaining multi-dimensional parameter data of a series battery pack, wherein the multi-dimensional parameter data comprises terminal voltage, current and temperature of each single battery;

[0041] Running a main state estimator based on an electro-thermal coupling equivalent circuit model, wherein the main state estimator adopts a Kalman filtering algorithm, and a baseline state of charge value is obtained by online estimation according to the multi-dimensional parameter data;

[0042] Using a probability error prediction model, predicting the estimation error of the baseline state of charge value based on the multi-dimensional parameter data, and outputting an error probability distribution containing error mean value and variance for representing prediction confidence; through a double-channel adaptive collaborative correction process, using the error probability distribution to generate a final high-fidelity state of charge estimation value;

[0043] Using the final high-fidelity state of charge estimation value and the multi-dimensional parameter data, constructing and driving a battery protection model, and outputting a safety control instruction, wherein the safety control instruction is used for controlling and protecting the series battery pack.

[0044] Further, the electro-thermal coupling equivalent circuit model comprises:

[0045] A temperature-sensitive parameter adaptation layer is configured to provide circuit parameters for an electrical state calculation layer according to a measured battery temperature. The temperature-sensitive parameter adaptation layer includes a parameter library storing a preset multi-dimensional mapping relationship between circuit parameters and temperature. The temperature-sensitive parameter adaptation layer receives a real-time temperature as a query input, retrieves and outputs complete circuit parameters corresponding to the current temperature from the parameter library through table lookup and interpolation calculation, including ohmic internal resistance, polarization resistance, polarization capacitance, and open-circuit voltage state-of-charge curve.

[0046] The electrical state calculation layer is configured to solve a state space equation of the battery based on the current circuit parameters and real-time current, and output a model-predicted terminal voltage. The electrical state calculation layer has a topology of a second-order RC equivalent circuit fixed therein, and state variables include state-of-charge and polarization voltage of two RC networks. The electrical state calculation layer receives the circuit parameters provided by the temperature-sensitive parameter adaptation layer, and updates the state variables through integral calculation according to the input real-time current, and calculates the predicted terminal voltage value according to Kirchhoff's voltage law.

[0047] The coupled heat generation calculation layer is configured to calculate total heat power generated by electrical activity. The coupled heat generation calculation layer has calculation logic based on Joule's law and thermodynamic principles integrated therein. The coupled heat generation calculation layer receives real-time current, internal resistance value corresponding to the current temperature, and entropy heat coefficient, respectively calculates irreversible Joule heat and reversible entropy heat, sums the two, and outputs a total heat power value.

[0048] Specifically, the temperature-sensitive parameter adaptation layer is configured to provide accurate circuit parameters that dynamically change with temperature for the entire model. The internal parameters of the battery, such as internal resistance and capacitance, are not constant, but are closely related to temperature. In order to capture this characteristic, the method pre-establishes a multi-dimensional mapping table storing the corresponding relationship between circuit parameters and temperature, i.e., a parameter library, through experimental calibration.

[0049] During system operation, the layer level receives the battery temperature measured by the external sensor in real time. The temperature is used as a query index to retrieve and output a complete set of circuit parameters at the current temperature from the parameter library through table lookup and interpolation calculation when the temperature is between two calibration points. The set of parameters includes:

[0050] Ohmic internal resistance: representing the instantaneous voltage response of the battery interior;

[0051] Polarization resistance: describing the impedance of electrochemical polarization and concentration polarization process;

[0052] Polarization capacitance: describing the dynamic characteristics and relaxation time of the polarization process;

[0053] Open-circuit voltage state-of-charge curve: describing the nonlinear relationship between open-circuit voltage and state-of-charge of the battery in the equilibrium state.

[0054] The layer level ensures that subsequent calculations of the model are always based on parameters that best match the current thermodynamic state, and is the basis for the adaptive ability of the model.

[0055] The electrical state calculation layer is the core of the model, responsible for simulating the electrical dynamic response of the battery. The topology of the second-order RC equivalent circuit model is fixed in this layer, which can better balance the model accuracy and computational complexity. Its state variables specifically include the state of charge (SOC) of the battery and the polarization voltage on the two RC networks representing fast and slow polarization processes, respectively. The workflow of this layer is as follows: first, it receives real-time updated circuit parameters from the temperature-sensitive parameter adaptation layer and real-time charging and discharging current from the current sensor. Based on the input real-time current, the state of charge (SOC) of the battery is updated in real time by integrating the current. At the same time, according to the state equation of the RC network, the two polarization voltage predictions are updated. According to Kirchhoff's voltage law, the open-circuit voltage corresponding to the current SOC, the voltage drop on the ohmic internal resistance, and the two polarization voltages are algebraically summed to calculate the model-predicted battery terminal voltage. Finally, the predicted terminal voltage output by this layer will be used in the subsequent state estimator (such as Kalman filter) to compare with the actual measured value to correct and optimize the state estimation.

[0056] The coupled heat generation calculation layer is responsible for quantifying the heat generated by the internal electrical activity of the battery during operation. The heat generated by the battery is mainly composed of two parts: irreversible Joule heat and reversible entropy heat. This layer integrates calculation logic based on Joule's law and thermodynamic principles, receives real-time current, and obtains the ohmic internal resistance, polarization resistance, and entropy heat coefficient corresponding to the current temperature from the temperature-sensitive parameter adaptation layer. Then, the Joule heat is calculated, which is the heat generated when the current flows through the internal resistance and is the main part of the heat generation. Further, the entropy heat is calculated, which is the reversible heat absorption or release phenomenon accompanied by the electrochemical reaction itself and is related to the current direction, temperature, and entropy heat coefficient. Finally, the total heat power is output, which is the sum of the calculated Joule heat and entropy heat, obtaining the total heat generation power value of the battery at the current time. This output value can not only be used to evaluate the thermal safety state of the battery, but also as an input for more complex electro-thermal coupled models to further improve the model accuracy.

[0057] The electro-thermal coupled equivalent circuit model is constructed, the temperature-sensitive parameter adaptation layer is used to adjust the internal parameters according to the real-time temperature to address the model parameter drift problem in a wide temperature range, and the second-order RC network structure is used to more accurately describe the polarization and other dynamic responses of the battery. The coupled heat generation calculation layer in the model quantifies the heat generated by the battery, providing data support for evaluating the thermal state of the battery and implementing active thermal management, and making up for the lag of traditional methods that rely only on temperature measurement.

[0058] Further, the main state estimator adopts a Kalman filtering algorithm, and a process of obtaining the baseline state of charge value according to the multi-dimensional parameter data online estimation includes: obtaining a state prior prediction value output by the electro-thermal coupled equivalent circuit model after forward operation, the state prior prediction value containing a prior state of charge and a model-predicted terminal voltage; comparing and calculating a measurement residual by comparing the model-predicted terminal voltage of the electro-thermal coupled equivalent circuit model with a real battery terminal voltage synchronously sampled by hardware; calculating an optimal Kalman gain according to the measurement residual and an error covariance matrix; and generating the baseline state of charge value by weighting and correcting the state prior prediction value by using the optimal Kalman gain.

[0059] Specifically, the main state estimator receives an output of the electro-thermal coupled equivalent circuit model, adopts a Kalman filtering algorithm, and obtains a baseline state of charge value by performing optimal fusion of model prediction and hardware measurement data through a periodically iterative prediction-correction closed loop process, so as to online estimate a baseline state of charge value closer to a real value.

[0060] The first stage of the process is a prediction stage, in which the estimator first performs state prediction, and obtains a state prior prediction value at the current time (k) by using a battery state space equation defined in the electrical state calculation layer, according to an optimal state estimation value (containing a SOC and a polarization voltage) at the last time and a real-time current input at the current time (k).

[0061] Then, in the prediction stage, the estimator performs covariance prediction for updating an error covariance matrix, which describes an uncertainty degree of the state prediction value. The uncertainty of the prediction increases with time. After the prediction stage is completed, a state prior prediction value containing a prior state of charge and a model-predicted terminal voltage, and an uncertainty description corresponding thereto are obtained.

[0062] The second stage of the process is a correction stage, in which the estimator corrects the prior prediction value given by the prediction stage by using a hardware actual measurement value, to obtain an optimal estimation at the current time. The first step of the stage is to calculate a measurement residual by calculating a difference between the model-predicted terminal voltage contained in the state prior prediction value and a real battery terminal voltage obtained by synchronously sampling the hardware circuit, to obtain the measurement residual. The measurement residual directly reflects a deviation between the model prediction and the actual situation.

[0063] Subsequently, the optimal Kalman gain is calculated according to the measurement residual, which is a weight coefficient for determining how much the measurement residual should be believed in the correction of the state. The calculation of the gain integrates two aspects of uncertainty: one is the state prediction uncertainty calculated in the prediction stage (represented by the error covariance matrix P), and the other is the noise uncertainty inherent in the hardware measurement system itself (represented by the measurement noise covariance matrix R). When the model prediction uncertainty is large and the measurement noise is small, the Kalman gain will be large, meaning that the measurement value is more believed; otherwise, the model prediction is more believed.

[0064] After obtaining the optimal Kalman gain, the state is weighted and corrected. The measurement residual is multiplied by the calculated optimal Kalman gain to obtain a correction amount, and then the correction amount is added to the prior state of charge in the state prior prediction value, and the final baseline state of charge value is output.

[0065] Finally, at the end of the correction stage, the error covariance is updated. According to the Kalman gain and the correction process this time, the error covariance matrix is updated to reduce its uncertainty and provide more accurate initial values for the prediction stage next time. Specifically, the update of the error covariance is built-in Kalman filter standard algorithm and Joseph form algorithm. Under normal stable working conditions, the system defaults to the standard algorithm to save computing resources, and continuously monitors the size of the measurement residual in this correction process. Once it is detected that the residual exceeds the preset normal range, it indicates that it is affected by external impact or the model has a transient and severe deviation, and the Joseph form algorithm is switched to update the covariance this time to ensure the positive definiteness and convergence of the error covariance matrix. This method effectively prevents the risk of divergence of the filter factor value calculation, greatly improving the robustness of the estimator under dynamic impact.

[0066] Through the prediction and correction closed loop of Kalman filtering, the cumulative error of the electro-thermal coupling model caused by simplification or aging is corrected by the measured voltage, and the dynamic characteristics of the model are used to suppress the interference of sensor noise on the estimation result. The core of this method is to dynamically weigh the confidence of the model and the measurement value through the Kalman gain, to obtain a baseline state of charge value that is statistically better, more stable and reliable than a single information source, providing a solid foundation for subsequent accurate correction and protection strategies.

[0067] Further, the probability error prediction model comprises: a state feedback and working condition fusion perception layer, configured to receive real-time multi-dimensional parameter data synchronized with the main state estimator, and simultaneously receive diagonal elements of a Kalman gain and an error covariance matrix P output by the main state estimator at a previous time, and fuse into a high-dimensional feature vector; a deep coupling error inference layer, configured to model and predict an estimation error of the main state estimator at a current time based on the high-dimensional feature vector; and a confidence distribution generation layer, configured to analyze and package a prediction result of the deep coupling error inference layer, and output an error probability distribution containing an error mean value and a variance used to represent a prediction confidence.

[0068] Specifically, a function of the state feedback and working condition fusion perception layer is to construct a high-dimensional feature vector for error prediction. First, the state feedback and working condition fusion perception layer is synchronized with the main state estimator to receive real-time multi-dimensional parameter data (terminal voltage, current, temperature), and simultaneously, diagonal elements of a Kalman gain and an error covariance matrix P output by the main state estimator at a previous calculation time are obtained. Finally, the real-time working condition data and internal state feedback data are fused to form a high-dimensional feature vector.

[0069] Further, the deep coupling error inference layer is configured to model and predict an estimation error of the main state estimator at a current time based on an input high-dimensional feature vector. The high-dimensional feature vector is received from a previous layer, and a forward propagation calculation is performed through an internal nonlinear mapping model to output a prediction value of the estimation error at the current time.

[0070] Further, the confidence distribution generation layer parses and encapsulates the original prediction result of the error inference layer to output in the form of a probability distribution. Specifically, the prediction result output by the deep-coupled error inference layer is parsed into an error mean and an error variance, and the error mean and the error variance are encapsulated to output a complete error probability distribution. The distribution includes a prediction of the error size (mean) and a credibility representation of the prediction result (variance). The deep-coupled error inference layer is implemented as a time-series feature enhanced attention network. The internal structure of the attention network includes a time convolution module for capturing the dynamic change trend of the input high-dimensional feature vector in a short time window to generate local working condition features containing time dimension information. The high-dimensional feature vector in this embodiment includes voltage, current, and temperature. Subsequently, the high-dimensional feature vector is sent to an attention weighting module that dynamically adjusts the attention weights of different features according to the criticality of the working condition by learning the feature weight distribution. For example, when the battery is close to full charge or full discharge, the attention module automatically increases the feature weights related to the terminal voltage and the internal polarization voltage of the model. In this way, the model realizes the transition from passively receiving all information to actively focusing on key information, making its prediction logic closer to the real physical and chemical process of the battery, thereby significantly improving the accuracy and robustness of error prediction under complex and variable working conditions.

[0071] By constructing a probability error prediction model, real-time working conditions and internal states of the estimator are fused to actively predict potential errors of the main estimator using nonlinear mapping. Not only is the error prediction value output, but also a credibility quantification of the prediction is provided to provide a key decision basis for the subsequent correction process, so that intelligent compensation and adaptive adjustment can be performed according to the prediction credibility, improving the effectiveness of correction and the robustness of the system as a whole.

[0072] Further, the double-channel adaptive cooperative correction process, specifically Fig. 2As shown, the process of generating the final high-fidelity state of charge estimation value using the error probability distribution includes: obtaining the error probability distribution containing the error mean and error variance output by the probability error prediction model, and simultaneously starting and executing the calculation of two correction channels according to the error probability distribution; in the output end gated correction channel, the error variance is converted into a confidence fusion weight through a preset inverse ratio function, the fusion weight is multiplied by the error mean to obtain a weighted error correction amount, and the weighted error correction amount is compensated to the baseline state of charge value to generate a corrected state of charge value; in the model end online adaptive channel, the same error variance is taken as a dynamic index representing the current mismatch degree of the electro-thermal coupling equivalent circuit model, is mapped to a process noise adjustment value through a preset monotonically increasing function, and the noise adjustment value is superimposed on the basic process noise covariance matrix Q of the Kalman filtering algorithm.

[0073] Specifically, the core of the dual-channel adaptive collaborative correction process is to simultaneously start and execute two parallel and complementary correction channels, including the output end gated correction channel and the model end online adaptive channel, according to the error probability distribution.

[0074] The role of the output end gated correction channel is to compensate and correct the baseline state of charge value output by the main state estimator to eliminate the prediction error. The channel first obtains the baseline state of charge value output by the main state estimator and the error probability distribution output by the probability error prediction model, then maps the error variance value in the error probability distribution through a preset inverse ratio function to convert it into a confidence fusion weight ranging from 0 to 1. The characteristic of the inverse ratio function is that the smaller the input error variance, the higher the credibility of the error prediction, and the closer the output weight value to 1. Conversely, the larger the error variance, the closer the weight value to 0. Subsequently, the calculated confidence fusion weight is multiplied by the error mean to obtain the final weighted error correction amount. Finally, the weighted error correction amount is compensated to the input baseline state of charge value, and the final high-fidelity state of charge estimation value is output. The inverse ratio function is specifically implemented as a nonlinear mapping function with adjustable threshold and sensitivity, specifically a parameterized tuned Sigmoid function, and the specific formula is:

[0075] ;

[0076] wherein, represents the final generated confidence fusion weight, represents the Sigmoid function, represents the sensitivity, which is obtained by combining offline experimental analysis and manual engineering tuning, represents the error variance value, The maximum error variance acceptable to the system is pre-calibrated according to a large amount of offline experimental data.

[0077] This design allows developers to accurately define the extent to which error prediction is trusted according to battery type and application scenario, so that the compensation behavior of the correction amount is no longer fixed but configurable and optimized, making the entire adaptive correction process more intelligent.

[0078] The role of the model-side online adaptive channel is not to directly correct the output result, but to indirectly optimize the estimation performance of the main state estimator at the next moment by adjusting its internal parameters in real time. The model-side online adaptive channel first obtains the error variance value in the error probability distribution and uses it as a dynamic indicator representing the current mismatch degree of the electro-thermal coupling equivalent circuit model. Then, through a pre-set monotonically increasing function, which is Softplus function in this embodiment, the error variance value is mapped to a process noise adjustment value. The greater the error variance, the more serious the model mismatch, and the greater the output process noise adjustment value. Subsequently, the obtained process noise adjustment value is superimposed on the basic process noise covariance matrix Q of the Kalman filter algorithm. In this way, the increased process noise covariance matrix Q makes the Kalman filter reduce the trust degree of model prediction and increase the trust weight of actual hardware measurement value in the next calculation period, thereby realizing online adaptive adjustment of the main state estimator.

[0079] By establishing a double-channel adaptive cooperative correction process, the output-side gated correction channel uses the confidence of error prediction to intelligently and non-blindly compensate the current state of charge value, ensuring the instant accuracy of each output. At the same time, the model-side online adaptive channel uses error variance as an indicator of model mismatch to dynamically adjust the process noise of the Kalman filter, so that the estimator can trust the measured data more when the model performs poorly, significantly improving the robustness and estimation accuracy of the system under complex working conditions throughout the battery's life cycle.

[0080] Further, the process of constructing and driving a battery protection model using the final high-fidelity state of charge estimate and the multi-dimensional parameter data to output safety control instructions comprises: inputting the final high-fidelity state of charge estimate, real-time temperature and battery health state information in the multi-dimensional parameter data as a combined feature vector into a neural network model for inferring a safety boundary, dynamically generating a safety operation boundary of the battery under the current state in real time by forward propagation calculation of the neural network, wherein the safety operation boundary includes a charging cutoff SOC threshold, a discharging cutoff SOC threshold, a maximum charging and discharging current allowed under the current working condition, and a highest / lowest working temperature dynamically adjusted with the state; comparing the final high-fidelity state of charge estimate and real-time current and temperature with the dynamically generated dynamic safety operation boundary, and if it is detected that the real-time parameters have a trend of crossing the corresponding safety boundary, triggering the corresponding protection logic to generate a safety control instruction to directly control the charging and discharging circuit.

[0081] Specifically, the battery protection model structure comprises three layers, namely a hardware protection layer, a state-aware software protection layer, and a dynamic safety boundary prediction protection layer.

[0082] In terms of overcurrent or short circuit protection, the hardware protection layer monitors the total current in real time, and once the current exceeds a preset hardware threshold (for example, a short circuit detection threshold much larger than the normal working current), the protection circuit is instantly triggered to directly disconnect the MOSFET switch of the main circuit without waiting for software intervention.

[0083] In terms of overvoltage or undervoltage protection, the hardware also monitors the battery voltage, and when extreme overcharge or overdischarge occurs, causing the voltage to reach the limit voltage threshold set by the hardware, the same hardware instantaneous shutdown action is performed.

[0084] The state-aware software protection layer is executed by the main controller software, and instead of relying only on fixed thresholds, it uses intermediate data such as the high-fidelity state of charge estimate output by the previous process to achieve more intelligent protection based on the state of the battery.

[0085] For overcharge protection, if the real-time voltage of any single battery cell exceeds the preset charging cutoff voltage, or the high-fidelity state of charge estimate output by the dual-channel adaptive collaborative correction process reaches or exceeds its upper threshold, either condition is met, and protection is triggered to stop charging.

[0086] For overdischarge protection, if the real-time voltage of any single battery cell is lower than the preset discharging cutoff voltage, or the high-fidelity state of charge estimate reaches or is lower than its lower threshold, either condition is met, and protection is triggered to stop discharging.

[0087] For over-current protection, the software level sets state-based charge and discharge over-current thresholds, which are dynamically adjusted according to the current high-precision state of charge estimate and real-time temperature. For example, when the SOC is in a high or low range, or when the temperature is high, the allowed continuous current threshold is adjusted accordingly. When the real-time current continuously exceeds the dynamically adjusted threshold for a certain period of time, the protection is triggered.

[0088] For temperature protection, when the battery temperature exceeds the set maximum working temperature or is lower than the minimum working temperature, the system will further judge in combination with the current high-precision state of charge estimate. For example, in a low-temperature environment (such as below 5°C), if the SOC is at a high level, a more stringent charging restriction or a prohibition of charging protection will be triggered to prevent low-temperature lithium precipitation.

[0089] The dynamic safety boundary prediction protection layer is the most core intelligent protection layer of the method. It uses the obtained high-precision state cognition to achieve predictive protection of safety risks, rather than simply triggering a threshold.

[0090] First, the dynamic safety boundary is generated. The final high-precision state of charge estimate output by the dual-channel adaptive collaborative correction process, the real-time temperature in the multi-dimensional parameter data, and the estimated battery health state (SOH) information are jointly combined into a combined feature vector and input into a pre-trained neural network model for inferring the safety boundary.

[0091] The neural network model can dynamically and in real time generate the complete safety running boundary of the battery under the current state through forward propagation calculation. The safety running boundary is a multi-dimensional threshold set, the specific content of which will be dynamically adjusted according to the battery state and working conditions, including the charge cut-off SOC threshold, the discharge cut-off SOC threshold, the maximum charge and discharge current allowed under the current working condition, and the highest and lowest working temperature allowed. These dynamic boundaries are more accurate and intelligent than the fixed thresholds of the second level.

[0092] Then, the level performs predictive protection triggering. The final high-precision state of charge estimate, the real-time current and temperature, and the dynamically generated safety running boundary in the previous step are continuously compared. If any real-time parameter is monitored to have a trend of crossing its corresponding safety boundary (for example, the system predicts that according to the current charging rate, the dynamic charge cut-off SOC threshold will be reached in a short time), the system will trigger the corresponding protection logic in advance. Finally, after triggering the protection logic, the system generates a clear safety control instruction. The instruction is directly sent to the actuator for controlling the charge and discharge circuit, such as by disconnecting the relay or adjusting the power of the charge and discharge module, to forcibly make the battery return to the safety running range, thereby achieving accurate and proactive protection of the battery.

[0093] Through the cooperative work of the above-mentioned three-level protection strategy, the method not only realizes comprehensive protection of abnormal conditions such as overcharge, overdischarge, overcurrent, short circuit and temperature, but also greatly improves the timeliness, accuracy and intelligent level of protection through the combination of software and hardware and dynamic prediction.

[0094] By constructing a complete adaptive algorithm framework, the battery monitoring and protection performance is comprehensively improved, the self-optimizing closed-loop system can adapt to battery aging and temperature changes, realize high-precision estimation in the whole life cycle, effectively solve the problems of insufficient precision and error accumulation of traditional schemes, at the same time, the dynamic safety boundary based on neural network improves the protection strategy from passive threshold triggering to active risk prediction, significantly enhances the safety, combined with the online adaptive ability of software and hardware multi-level protection and algorithm, ensures the robustness of the system under complex working conditions, provides a high-precision and high-reliability management scheme for series battery pack.

[0095] Embodiment two:

[0096] In order to better realize the battery pack power supply monitoring of the 8 lithium ion battery series connected in the outdoor monitoring station, the series battery high-precision power monitoring system based on adaptive algorithm provided by the application is introduced, and the specific system structure is as shown in Fig. 3 .

[0097] The initial state of charge (SOC) of the battery pack of the monitoring station is monitored, and the outdoor environment temperature at this time is-15°C. The monitoring station is in a low-power data acquisition mode.

[0098] The system starts to monitor the terminal voltage, charging current and temperature of each single battery in the battery pack in real time. After the charging starts, the battery temperature slowly rises from-15°C. When the temperature reaches-12°C, the temperature-sensitive parameter adaptation layer captures this change, and uses the temperature as an index to retrieve and update the circuit parameters of the complete electrothermal coupling model from the internal parameter library through table lookup and interpolation calculation, providing accurate internal resistance, capacitance and OCV-SOC curve matched with the current extremely low temperature environment for subsequent calculation.

[0099] The solar charging management module outputs the charging current. The electrical state calculation layer receives the current value and the circuit parameters at-12°C provided by the previous layer, and performs forward calculation through the state space equation to output a model predicted terminal voltage and obtain a prior SOC estimation value.

[0100] The prior SOC and the predicted terminal voltage are passed to the main state estimator. The estimator compares the model predicted terminal voltage with the hardware actually sampled terminal voltage, and finds a deviation caused by the low-temperature model error. The Kalman filtering algorithm calculates the optimal Kalman gain according to the deviation (measurement residual) and system uncertainty, and corrects the prior SOC, and finally outputs a more reliable baseline state of charge value.

[0101] Further, a probabilistic error prediction model is started, its state feedback and working condition fusion perception layer receives the current real-time working condition data (voltage, charging current, temperature), and at the same time obtains the diagonal elements of the Kalman gain and error covariance matrix P output by the main state estimator at the last time, and fuses these data into a high-dimensional feature vector.

[0102] The feature vector is sent to the deep coupling error inference layer, and the neural network model inside the deep coupling error inference layer performs calculation to predict the possible estimation error of the current main state estimator. For example, the error variance is a medium-sized value, indicating that the prediction result has a certain uncertainty.

[0103] The confidence distribution generation layer encapsulates the above result into an error probability distribution, and passes it to the double-channel adaptive collaborative correction process.

[0104] The output end gate correction channel: according to the medium-sized error variance, a medium-sized confidence fusion weight is calculated through an inverse function, for example, 0.6. Then, the weight is multiplied by the error mean to obtain a weighted error correction amount. Finally, the correction amount is compensated to the baseline state of charge value. At this time, the system has obtained the final output of the high-fidelity state of charge estimation value at the current time.

[0105] The model end online adaptive channel: at the same time, according to the same medium-sized error variance, a corresponding size of process noise adjustment value is mapped through a monotone increasing function. The adjustment value is superimposed on the process noise covariance matrix Q of the Kalman filter, so that the filter appropriately reduces the trust degree of the current low-temperature model in the next period, and instead relies more on the hardware measurement value.

[0106] In addition, for multiple protection of the battery, for example, due to line aging or animal gnawing, a transient short circuit occurs at the output end of the battery, and the hardware protection layer hardware circuit monitors a huge short circuit current far beyond the normal range. In the microsecond level time, its hardware comparator immediately acts to directly trigger and disconnect the MOSFET switch of the main loop, cutting off the power supply to prevent the battery from catching fire or exploding due to short circuit. The whole process does not need software intervention, and the response speed is extremely fast.

[0107] For the second level protection of the state-aware software protection layer, for example, after a long time of sunshine, the battery power is charged to a high level, and one battery reaches the preset software charging cutoff voltage first due to faster aging, the state-aware software protection layer monitors that the voltage of the battery exceeds the limit, and immediately triggers the overcharge protection (OVP) to generate an instruction to stop charging and start the battery balancing function to discharge the battery to reduce the inconsistency in the battery pack. For another example, during the charging process, due to low temperature or battery aging, etc., the voltage of any battery may not reach the absolute charging cutoff voltage, but the high-precision state of charge estimation value output by the method of the present application has accurately reached 100%. At this time, the state-aware software protection layer will determine that the battery is fully charged based on the high-precision SOC data rather than only relying on the voltage, and also trigger the overcharge protection (OVP) to generate an instruction to stop charging to prevent overcharging due to model deviation or inaccurate voltage measurement, thereby improving the accuracy of protection.

[0108] For the third level protection of the dynamic safety boundary prediction protection layer, for example, the ambient temperature of the monitoring station is as high as 40°C, and the station is exposed to the sun for a long time, causing the temperature of the battery to rapidly rise to 58°C during the charging process.

[0109] The neural network model of the dynamic safety boundary prediction protection layer receives the current high-precision SOC (for example, 85%), high temperature (58°C), and battery health status (SOH, for example, 88%), etc. information, and immediately recalculates and outputs a new dynamic safety boundary. The boundary indicates that, in the current state, to prevent thermal runaway, the maximum allowable operating temperature should be lowered from the conventional 60°C to 55°C. The system compares the dynamic boundary with the real-time monitored battery temperature of 58°C, and finds that the real-time temperature has exceeded the dynamic safety boundary. The system determines that the battery has a risk of overheating, and triggers the protection logic in advance to generate a safety control instruction, which on the one hand sends a high-temperature warning to the background management center through the wireless module, and on the other hand directly controls the charging module to reduce the charging current or even suspend the charging until the battery temperature falls within the new dynamic safety boundary.

[0110] By constructing a self-optimizing closed-loop estimation system, high-precision state awareness of the battery throughout its life cycle is achieved, and the protection strategy is improved from passive response to active prediction, thereby significantly enhancing the reliability and safety of the system under complex working conditions.

[0111] Although embodiments of the present application have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and alterations can be made thereto without departing from the principles and spirit of the present application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A high-precision power monitoring method for series-connected batteries based on an adaptive algorithm, characterized in that: include: Acquiring multi-dimensional parameter data of the series-connected battery pack, the multi-dimensional parameter data including the terminal voltage, current, and temperature of each single battery; Running a main state estimator based on an electrothermal coupling equivalent circuit model, wherein the main state estimator uses a Kalman filter algorithm to obtain a baseline state of charge value based on the multi-dimensional parameter data online; Using a probabilistic error prediction model, predict an estimation error of the baseline state of charge value based on the multidimensional parameter data, and output an error probability distribution including an error mean and a variance used to characterize the prediction confidence; The error probability distribution is used to generate a final high-fidelity state of charge estimate through a dual-channel adaptive collaborative correction process; The process of generating a final high-fidelity state-of-charge estimate using the error probability distribution through a dual-channel adaptive collaborative correction process includes: performing calculations on two correction channels based on the error probability distribution, wherein the error probability distribution includes an error variance and an error mean; in an output-end gated correction channel, converting the error variance into a confidence fusion weight using a preset inverse function, multiplying the fusion weight by the error mean to obtain a weighted error correction, and compensating the weighted error correction to a baseline state-of-charge value to generate a final high-fidelity state-of-charge estimate; in a model-end online adaptive channel, using the error variance as a dynamic indicator of the degree of mismatch of the electrothermal coupling equivalent circuit model, mapping it into a process noise adjustment value using a preset monotonically increasing function, and superimposing the noise adjustment value on the noise covariance matrix of the Kalman filter algorithm; The final high-fidelity state of charge estimation value and multi-dimensional parameter data are used to construct and drive a battery protection model, and output safety control instructions, which are used to control and protect the series battery pack.

2. The method for high-precision power monitoring of series-connected batteries based on an adaptive algorithm according to claim 1, characterized in that: The electrothermal coupling equivalent circuit model includes: The temperature-sensitive parameter adaptation layer contains a parameter library that stores preset multi-dimensional mapping relationships between circuit parameters and temperature. It receives real-time temperature as a query input, retrieves and outputs the complete circuit parameters corresponding to the current temperature from the parameter library; The electrical state calculation layer receives circuit parameters and updates state variables through integration operations based on the real-time current input, and calculates the predicted terminal voltage value according to Kirchhoff's voltage law; The coupled heat generation calculation layer is used to calculate the total heat power generated by electrical activities; it receives the real-time current and the internal resistance value and entropy heat coefficient corresponding to the current temperature, calculates the irreversible Joule heat and reversible entropy heat respectively, and sums the two to output the total heat generation power value.

3. The method for high-precision power monitoring of series-connected batteries based on an adaptive algorithm according to claim 1, characterized in that: The main state estimator adopts a Kalman filter algorithm, and the process of online estimating the baseline state of charge value based on the multi-dimensional parameter data includes: obtaining the state prior prediction value output by the electrothermal coupling equivalent circuit model after forward operation, the state prior prediction value contains the prior state of charge and the terminal voltage value predicted by the model; comparing the terminal voltage predicted by the electrothermal coupling equivalent circuit model with the actual battery terminal voltage synchronously sampled by the hardware and calculating the measurement residual, calculating the optimal Kalman gain based on the measurement residual and the error covariance matrix, and using the optimal Kalman gain to perform weighted correction on the state prior prediction value to generate the baseline state of charge value.

4. The method for high-precision power monitoring of series-connected batteries based on an adaptive algorithm according to claim 1, characterized in that: The probability error prediction model includes: The state feedback and working condition fusion perception layer is used to receive real-time multi-dimensional parameter data synchronized with the main state estimator, and at the same time receive the diagonal elements of the optimal Kalman gain and error covariance matrix output by the main state estimator at the previous moment, and fuse them into a high-dimensional feature vector; the deep coupling error inference layer is used to model and predict the estimation error of the main state estimator at the current moment based on the high-dimensional feature vector; the confidence distribution generation layer is used to parse and encapsulate the prediction results of the deep coupling error inference layer, and output an error probability distribution containing the error mean and the variance used to characterize the prediction confidence.

5. The method for high-precision power monitoring of series-connected batteries based on an adaptive algorithm according to claim 1, characterized in that: The process of using the final high-fidelity state of charge estimation value and the multi-dimensional parameter data to construct and drive the battery protection model and output safety control instructions includes: The final high-fidelity state of charge estimate and the real-time temperature, terminal voltage, and current in the multidimensional parameter data are input as a combined feature vector into a neural network model for inferring safety boundaries to generate a safe operating boundary for the battery in the current state. The safe operating boundary includes a charge cutoff SOC threshold, a discharge cutoff SOC threshold, a maximum charge and discharge current allowed under the current operating conditions, and a maximum / minimum operating temperature that are dynamically adjusted with the state. The final high-fidelity state of charge estimate and the real-time current and temperature are compared with the safe operating boundary. If it is detected that the real-time parameters have a tendency to cross the corresponding safety boundary, the corresponding protection logic is triggered, and a safety control instruction is generated to directly control the charge and discharge circuit.

6. High-precision power monitoring system for series-connected batteries based on adaptive algorithm, characterized by: include: A data acquisition module, which acquires multi-dimensional parameter data of the series-connected battery pack, wherein the multi-dimensional parameter data includes the terminal voltage, current and temperature of each single battery; a state estimation module, running a main state estimator based on an electrothermal coupling equivalent circuit model, wherein the main state estimator uses a Kalman filter algorithm to online estimate a baseline state of charge value based on the multi-dimensional parameter data; an error correction module, which uses a probabilistic error prediction model to predict an estimation error of the baseline state of charge value based on the multi-dimensional parameter data, and outputs an error probability distribution including an error mean and a variance used to characterize the prediction confidence; The error probability distribution is used to generate a final high-fidelity state of charge estimate through a dual-channel adaptive collaborative correction process; The process of generating a final high-fidelity state-of-charge estimate using the error probability distribution through a dual-channel adaptive collaborative correction process includes: performing calculations on two correction channels based on the error probability distribution, wherein the error probability distribution includes an error variance and an error mean; in an output-end gated correction channel, converting the error variance into a confidence fusion weight using a preset inverse function, multiplying the fusion weight by the error mean to obtain a weighted error correction, and compensating the weighted error correction to a baseline state-of-charge value to generate a final high-fidelity state-of-charge estimate; in a model-end online adaptive channel, using the error variance as a dynamic indicator of the degree of mismatch of the electrothermal coupling equivalent circuit model, mapping it into a process noise adjustment value using a preset monotonically increasing function, and superimposing the noise adjustment value on the noise covariance matrix of the Kalman filter algorithm; The safety control module uses the final high-fidelity state of charge estimation value and multi-dimensional parameter data to build and drive a battery protection model and output safety control instructions, wherein the safety control instructions are used to control and protect the series battery pack.

Images