Sightseeing car battery status monitoring method, device, equipment and storage medium

Abstract

The invention relates to the technical field of battery state monitoring, and discloses a method, device, equipment and storage medium for monitoring the battery state of a sightseeing vehicle. The method comprises: collecting voltage, current and temperature data of a battery of the sightseeing vehicle, and performing preprocessing to obtain a battery operation parameter data set; constructing a battery equivalent circuit model, and performing global and local two-stage optimization to obtain an optimal parameter set of the battery equivalent circuit model; constructing a state equation and an observation equation, initializing a covariance matrix of a triple forgetting factor adaptive extended Kalman filter, and obtaining a battery state observation model; obtaining a priori state estimation through time update operation and a posteriori state estimation through measurement update operation, performing hierarchical estimation on the battery charge state and polarization voltage, and obtaining a battery state characteristic quantity; executing three-layer hierarchical control, and outputting a battery state control parameter. The invention improves the reliability and safety of a battery management system for sightseeing vehicles.

Description

Technical Field

[0001] The present invention relates to the technical field of battery status monitoring, and in particular to a method, device, equipment and storage medium for monitoring the battery status of a sightseeing vehicle. Background Art

[0002] As the key power source of sightseeing vehicles, the accuracy and reliability of battery status monitoring are directly related to the safe operation and service life of sightseeing vehicles. Since sightseeing vehicles usually operate under complex road conditions in scenic spots, they often face conditions such as climbing slopes and frequent starts and stops, which puts higher requirements on the battery management system.

[0003] There are three main problems with the current battery status monitoring technology for sightseeing vehicles. First, the operating environment of sightseeing vehicles is complex and the vibration interference is large, which leads to a high noise level in the acquisition of battery status parameters, affecting the accuracy of traditional monitoring algorithms. Second, the existing battery equivalent circuit model parameter identification method has a large amount of calculation in practical applications and is difficult to meet the needs of real-time monitoring of the battery status of sightseeing vehicles. Due to the variable load conditions of sightseeing vehicles, a single state estimation algorithm cannot take into account both estimation accuracy and convergence speed at the same time. Summary of the invention

[0004] The present invention provides a method, device, equipment and storage medium for monitoring the battery status of a sightseeing vehicle, and the present invention improves the reliability and safety of the battery management system of the sightseeing vehicle.

[0005] In a first aspect, the present invention provides a method for monitoring the battery status of a sightseeing vehicle, the method comprising:

[0006] Collect the voltage, current and temperature data of the sightseeing car battery and pre-process it to obtain the battery operation parameter data set;

[0007] Building a battery equivalent circuit model based on the battery operating parameter data set, and performing global and local two-stage optimization to obtain an optimal parameter set for the battery equivalent circuit model;

[0008] Based on the optimal parameter set, a state equation and an observation equation are constructed, and a covariance matrix of a triple forgetting factor adaptive extended Kalman filter is initialized to obtain a battery state observation model;

[0009] The battery operating parameter data set is input into the battery state observation model, a priori state estimation is obtained through time update operation and a posteriori state estimation is obtained through measurement update operation, and the battery state of charge and polarization voltage are estimated hierarchically to obtain battery state characteristic quantities;

[0010] A three-level hierarchical control is performed based on the battery state feature quantity, and a battery state control parameter is output.

[0011] In a second aspect, the present invention provides a sightseeing vehicle battery status monitoring device, the sightseeing vehicle battery status monitoring device comprising:

[0012] The acquisition module is used to collect the voltage, current and temperature data of the sightseeing car battery and perform preprocessing to obtain the battery operation parameter data set;

[0013] A construction module, used to construct a battery equivalent circuit model based on the battery operation parameter data set, and perform global and local two-stage optimization to obtain an optimal parameter set of the battery equivalent circuit model;

[0014] A setting module, used to construct a state equation and an observation equation based on the optimal parameter set, initialize the covariance matrix of the triple forgetting factor adaptive extended Kalman filter, and obtain a battery state observation model;

[0015] An estimation module, used to input the battery operation parameter data set into the battery state observation model, obtain a priori state estimation through time update operation and a posteriori state estimation through measurement update operation, perform hierarchical estimation on the battery state of charge and polarization voltage, and obtain battery state characteristic quantities;

[0016] The output module is used to perform three-level hierarchical control based on the battery state characteristic quantity and output battery state control parameters.

[0017] The third aspect of the present invention provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned sightseeing vehicle battery status monitoring method.

[0018] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned sightseeing vehicle battery status monitoring method.

[0019] In the technical solution provided by the present invention, by introducing a triple forgetting factor adaptive extended Kalman filter, accurate estimation of the battery state of a sightseeing vehicle in a high noise environment is achieved, and a series PSO-SA algorithm is used to perform global and local two-stage optimization of the battery equivalent circuit model parameters, which effectively reduces the computational complexity of parameter identification while ensuring the accuracy of the model parameters; a three-layer hierarchical control strategy is designed to achieve all-round control from parameter monitoring to state evaluation to energy management, thereby improving the reliability and safety of the sightseeing vehicle battery management system; fuzzy control is combined with dynamic programming to construct an energy recovery optimization control strategy, thereby improving the recovery rate of the braking energy of the sightseeing vehicle and extending the battery life; a layered battery state monitoring architecture is adopted to achieve coordinated monitoring of multiple parameters such as battery voltage, current, and temperature, thereby improving the fault tolerance and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0021] Figure 1 A schematic diagram of the steps of a sightseeing vehicle battery status monitoring method according to an embodiment of the present invention;

[0022] Figure 2 It is a structural schematic diagram of a battery status monitoring device for a sightseeing vehicle in an embodiment of the present invention;

[0023] Figure 3 It is a schematic block diagram of the structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] Embodiments of the present invention provide a method, device, equipment and storage medium for monitoring the battery status of a sightseeing vehicle. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the sightseeing vehicle battery status monitoring method in the embodiment of the present invention includes:

[0026] Step S1, collecting voltage, current and temperature data of the sightseeing car battery, and preprocessing to obtain a battery operation parameter data set;

[0027] It is understandable that the execution subject of the present invention may be a sightseeing vehicle battery status monitoring device, or a terminal or a server, which is not specifically limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0028] Specifically, the type information, rated capacity and rated voltage parameters of the sightseeing car battery are collected non-invasively in multiple stages. The basic parameter data of the battery is obtained through multiple measurements, recordings and verifications. Considering the differences in battery models and specifications, the relevant parameters are collected in stages to ensure the comprehensiveness and accuracy of the data. The basic parameter data of the battery and the operating parameter range are classified, modeled and stored to form a benchmark data set. The terminal voltage, current and temperature parameters of the sightseeing car battery are sampled in real time to obtain the original sampled data. The original sampled data is digitally filtered. By designing a filter, high-frequency noise and unnecessary signal components are effectively removed to obtain the filtered data. The filtered data is subjected to signal-to-noise ratio analysis and outlier detection. The signal-to-noise ratio analysis can effectively evaluate the quality of the data and identify which parts of the data are affected by noise. Through the signal-to-noise ratio analysis, abnormal fluctuations and errors in the data are detected. On this basis, outlier detection can identify data points that are significantly deviated from the normal range. These outliers are caused by factors such as sensor failure, sampling errors or external interference. After marking these outliers, they are repaired and compensated. The data repair and compensation technology adjusts the abnormal data according to the algorithm to restore it to a reasonable range. The repaired data is reconstructed in the battery state parameter vector format. The battery state data at each moment will be organized in a fixed format to facilitate subsequent processing and analysis. The battery state parameter vector includes multiple key parameters, such as battery voltage, current, temperature, and estimated state of charge. Organizing these data in vector form facilitates state estimation and analysis through computer models. The reconstructed data is time-aligned and data normalized. The collected parameters are matched in chronological order to ensure that the values ​​of each parameter can accurately correspond at the same time point. Data normalization is to map the values ​​of each parameter to a unified numerical range to eliminate the impact of differences in different dimensions and numerical ranges on subsequent analysis and modeling. Finally, a battery operation parameter data set is obtained.

[0029] Step S2: constructing a battery equivalent circuit model based on the battery operation parameter data set, and performing global and local two-stage optimization to obtain an optimal parameter set of the battery equivalent circuit model;

[0030] Specifically, the correlation analysis is performed on the open circuit voltage source, the internal series resistor and the RC parallel network. Through the statistical analysis of key parameters such as the battery terminal voltage, current and temperature, the relationship between the various components inside the battery is revealed, which helps to determine the topological structure of the battery equivalent circuit model, that is, how to connect and configure these physical elements (such as the open circuit voltage source, the series resistor and the RC network) in the circuit diagram to form a circuit model that can accurately reflect the battery performance. The parameters of the battery equivalent circuit model are optimized. The battery operation parameter data set and the topological structure are input into the particle swarm algorithm for parameter initialization. In the particle swarm algorithm, each particle represents a possible parameter combination, and the entire particle swarm explores the optimal solution in the search space. The particle swarm algorithm guides the particles to find the optimal solution in the parameter space by setting parameters such as the particle swarm size, the maximum number of iterations, the learning factor and the inertia weight. By initializing the parameter matrix of the particle swarm, a preliminary reference framework is provided for the subsequent optimization process. In the particle swarm algorithm, the position of each particle represents a parameter set, and the position change of the particle is guided by the fitness function. The fitness function is used to evaluate whether the current particle position can provide a better battery equivalent circuit model parameter. When the particle swarm algorithm is running, the fitness of each particle is calculated, and the global optimal particle position and the local optimal particle position are selected according to the extreme difference method to obtain the global search parameter set. The global optimal particle position represents the best solution in the parameter space of the battery equivalent circuit model, while the local optimal particle position reflects the best solution of the particle in the local search space. The optimal parameter range of the model is roughly determined by global search, and these parameters are refined by local search to ensure that the final parameter set can effectively reflect the true performance of the battery. The parameter set obtained by the global search is input into the simulated annealing algorithm for processing. The simulated annealing algorithm searches for the optimal solution in the space by simulating the particle motion behavior during the physical annealing process. The core idea of ​​the annealing algorithm is to allow large fluctuations in the initial stage, so as to jump out of the local optimal solution, and then gradually converge to the global optimal solution as the temperature gradually decreases. By using the Metropolis criterion to calculate the state transition probability and local search of the annealing initial solution, the parameter space is effectively explored to find the local optimal solution of the battery equivalent circuit model. The internal resistance parameters and polarization parameters of the battery equivalent circuit model are jointly solved based on the local optimization solution. The working state and performance characteristics of the battery are reflected by solving the resistance and polarization characteristics inside the battery. The preliminary values ​​of the model parameters are obtained by solving these parameters using the local optimization solution. The initial values ​​are input into the verification module for cross-validation operations to evaluate the accuracy and reliability of the model. During the cross-validation process, the performance of the model is evaluated by calculating the root mean square error between the predicted value and the actual measured value of the battery terminal voltage. By comparing the error between the predicted value and the measured value, the model parameters are adjusted in reverse to ensure that the model can more accurately predict the state of the battery.In the cross-validation process, the parameter combination with the smallest error is screened out to obtain the optimal parameter group. The optimal parameter group represents the optimal configuration of each component in the battery equivalent circuit model, including the series resistance value, the parameters of the RC network, and the open circuit voltage characteristic curve. These optimal parameters are used to numerically calibrate the battery equivalent circuit model to improve the accuracy and adaptability of the model. By accurately calibrating these parameters, the optimal parameter set of the battery equivalent circuit model is finally obtained.

[0031] Step S3, constructing a state equation and an observation equation based on the optimal parameter set, initializing the covariance matrix of the triple forgetting factor adaptive extended Kalman filter, and obtaining a battery state observation model;

[0032] Specifically, the open circuit voltage equation is linearized according to the optimal parameter set of the battery equivalent circuit model. The battery state equation is constructed using the battery voltage characteristics and polarization phenomenon to obtain the description of the two key parameters of the battery state of charge and polarization voltage. When constructing the state equation, the dynamic characteristics of the battery are considered and converted into a state space model suitable for extended Kalman filter processing, so that the real-time state of the battery is estimated and tracked through the filtering algorithm. Taylor expansion operation is performed on the state space model to construct an observation equation containing the battery terminal voltage to obtain a measurement space model. The measurement space model quantifies the relationship between the output voltage of the battery and the internal state of the battery (such as the state of charge and polarization voltage), so that the internal state of the battery can be indirectly estimated through the battery terminal voltage. Through the construction of these two models, a complete battery state estimation framework is obtained. The state space model and the measurement space model are input into the triple forgetting factor adaptive extended Kalman filter for initialization configuration. The covariance matrix of the Kalman filter is initialized. The covariance matrix reflects the uncertainty of the system state in the Kalman filter and is configured through a suitable initialization method. The initialization process involves recursive calculation of the covariance matrix and starting the filter operation by introducing the initial value of the first forgetting factor. The first forgetting factor is used to control the memory weight of historical data and determines the degree of dependence of the system on past data when processing new data. The initial value of the second forgetting factor is obtained by recursive calculation of the process noise covariance matrix. The measurement noise covariance matrix is ​​dynamically updated based on the initial values ​​of the first forgetting factor and the second forgetting factor to obtain the initial value of the third forgetting factor. The third forgetting factor is used to adjust the response speed to the measurement noise to ensure that the filter can adaptively adjust the weight when facing different noise levels. By dynamically updating the initial values ​​of these three forgetting factors, the Kalman filter is more flexible and accurate when processing measurement data with different dynamic and noise characteristics. The covariance matrix is ​​updated based on the initial values ​​of the first forgetting factor, the second forgetting factor, and the third forgetting factor. Through matrix multiplication, the influence of different forgetting factors on the covariance matrix is ​​combined to obtain the updated initial value of the covariance matrix. The initial value of the covariance matrix is ​​subjected to singular value decomposition to ensure the stability and numerical accuracy of the matrix. The covariance matrix is ​​decomposed into a set of eigenvalues ​​and eigenvectors, thereby simplifying subsequent matrix operations and making the filter more numerically stable. Based on the decomposed covariance matrix and state transfer matrix, the Jacobian matrix of the state variables is calculated using the least squares method. The Jacobian matrix is ​​a key parameter that describes the change of the system state. It is used in the Kalman filter to estimate the rate of change of the system state and provide an accurate prediction of the battery state. By combining the initial values ​​of the Jacobian matrix and the covariance matrix, the performance of the Kalman filter is optimized and the battery state observation model is obtained.

[0033] According to the relative importance of the first forgetting factor and the second forgetting factor, a weighted average is taken to obtain a comprehensive forgetting factor combination value. The function of the forgetting factor is to control the degree of memory of the filter for past information. By adjusting the value of the forgetting factor, the filter's dependence on historical data and new data can be dynamically adjusted. Based on the forgetting factor combination value, a diagonal weight matrix is ​​constructed. The weight matrix is ​​a key factor used in the Kalman filter to adjust different measurement noises. The weight matrix and the measurement noise covariance matrix are recursively operated. The estimated value of the covariance matrix is ​​dynamically adjusted according to the current noise level. By combining with the measurement noise covariance matrix, the estimated uncertainty of the system state is updated to obtain the initial value of the third forgetting factor. The initial value of the third forgetting factor determines the filter's adaptability to measurement noise. It is carefully adjusted during the recursive calculation process to ensure that the system can adaptively optimize the response of the measurement noise according to environmental changes. The first forgetting factor, the second forgetting factor, and the third forgetting factor are normalized to eliminate the magnitude differences between different forgetting factors and ensure that their updates to the covariance matrix can be performed on the same scale. Through normalization, a normalized forgetting factor matrix is ​​obtained. The normalized forgetting factor matrix is ​​covariance updated. The estimated error of the system state is updated by introducing new forgetting factor information. The covariance update reflects the impact of new data on the uncertainty of the system state, thereby optimizing the filter's ability to predict future states. The updated covariance matrix is ​​partitioned into several small blocks for more efficient subsequent calculations. The covariance matrix is ​​split into multiple parts by the matrix partitioning method, which simplifies the calculation process, reduces the amount of calculation, and improves the calculation accuracy. The condition number of the block covariance matrix is ​​optimized. The condition number is an important property of the matrix, which measures the degree of numerical stability of the matrix. A larger condition number means that the matrix is ​​numerically unstable, which will lead to the amplification of the calculation error. During the optimization process, the eigenvalues ​​of the matrix are adjusted by the regularization method to ensure that the condition number of the matrix is ​​within an appropriate range, thereby improving the numerical stability and calculation accuracy of the filter. The regularization method includes scaling or penalizing the eigenvalues ​​to suppress larger eigenvalues ​​and prevent numerical instability. Through the above optimization process, the optimized covariance matrix is ​​obtained. Based on the optimized covariance matrix, matrix multiplication operations are performed to perform covariance propagation calculations. Covariance propagation refers to propagating the covariance matrix from the current state to the next time step to predict the uncertainty change of the system state. In the covariance propagation process, the state transfer matrix is ​​combined with the covariance matrix through matrix multiplication to obtain the covariance prediction matrix. The covariance prediction matrix is ​​symmetry corrected and positive definiteness verified to obtain the initial value of the covariance matrix. The symmetry correction is to ensure that the covariance matrix is ​​numerically symmetric, because the essence of the covariance matrix should be symmetric; and the positive definiteness verification is to ensure that the covariance matrix is ​​positive definite, that is, the eigenvalues ​​of the matrix are all positive, thereby ensuring the stability and convergence of the filter.

[0034] Step S4, inputting the battery operation parameter data set into the battery state observation model, obtaining a priori state estimation through time update operation and a posteriori state estimation through measurement update operation, performing hierarchical estimation on the battery state of charge and polarization voltage, and obtaining battery state characteristic quantities;

[0035] Specifically, the battery operation parameter data set is input into the battery state observation model. The battery state space model is used to initialize the prediction of the battery state of charge, and a preliminary estimate of the current state of the battery is generated according to the input battery operation data set, and a state prediction matrix is ​​obtained, which includes the estimated value and estimation uncertainty of the current state of the battery. Based on the triple forgetting factor in the battery state observation model, a time update operation is performed on the state prediction matrix to obtain a priori state estimation value. The triple forgetting factor controls the degree of dependence of the model on historical data. Through appropriate forgetting factor adjustment, the model can automatically adjust the memory of historical information according to new measurement data. This time update process gradually updates the information in the state prediction matrix over time and obtains a more accurate priori state estimation value. The priori state estimation value is input into the measurement update unit of the battery state observation model. The measurement update unit combines the measurement space model to calculate the measurement prediction value through observation prediction. The measurement space model establishes a relationship between the battery terminal voltage and the internal state (such as the state of charge and polarization voltage), and through this relationship, the battery state is predicted from the observed data. For the difference between the measured predicted value and the actual observed value, the difference calculation is performed, and the state variable is corrected by the gain matrix of the battery state observation model. The difference calculation reflects the error between the model prediction and the actual observation. Through the application of the gain matrix, the error is effectively corrected to obtain the posterior state estimate. The gain matrix is ​​dynamically adjusted according to the current measurement error and prediction error, and the system state is adjusted to make the posterior state estimate as close to the actual value as possible. The posterior state estimate is input into the covariance update unit of the battery state observation model. The error covariance matrix is ​​dynamically adjusted based on the triple forgetting factor to obtain the updated covariance matrix. The covariance matrix reflects the uncertainty of the state estimation. As time goes by, the covariance matrix will be updated according to the newly obtained measurement information, making the estimation of the battery state more accurate and reliable. By introducing the triple forgetting factor, the covariance matrix is ​​dynamically adjusted according to the different contributions of historical data and current data to improve the accuracy of state estimation. In order to ensure the numerical stability of the covariance matrix, the updated covariance matrix is ​​optimized for numerical stability. The covariance matrix is ​​corrected using the Joseph stabilization algorithm to obtain the optimized state quantity. The Joseph stabilization algorithm is a method for optimizing the covariance matrix. By processing the eigenvalues ​​of the covariance matrix, the numerical stability of the matrix is ​​ensured, thereby avoiding numerical instability problems caused by calculation errors. The optimized state quantity is input into the feature extraction unit of the battery state observation model. The feature extraction unit extracts the two key features of the battery state of charge and polarization voltage from the optimized state quantity to obtain a state feature sequence. Based on the extracted state feature sequence, a multi-dimensional analysis is performed, and feature fusion is performed in combination with the evaluation criteria of the battery state observation model to obtain a comprehensive battery state feature quantity.

[0036] Step S5: Execute three-level hierarchical control based on the battery state characteristic quantity and output the battery state control parameter.

[0037] Specifically, the first-level safety threshold monitoring operation is performed on the voltage, current and temperature parameters in the battery state characteristic quantity. The key parameters of the battery (such as voltage, current and temperature) are input into the monitoring system, and the parameter monitoring matrix is ​​obtained through monitoring operation. The matrix reflects whether the various parameters of the battery are within the safe working range, and then the multi-threshold segmented judgment is performed on these parameters to ensure that the battery is in normal working state. Through multi-threshold segmented judgment, it is determined whether the battery exceeds the safety range, and the first-level parameter monitoring results are obtained according to different threshold intervals. The first-level parameter monitoring results are evaluated by fuzzy rules. Through the combination conditions of the battery state parameters, a fuzzy membership function is constructed to quantify the membership degree of each battery state parameter, reflecting the fuzziness of the battery state. In the fuzzy rule evaluation, the overall health status and working state of the battery are evaluated according to the combination of state variables such as the voltage, current, and temperature of the battery. The constructed state evaluation function matrix obtains the overall evaluation of the battery state according to different fuzzy membership function evaluation results. The fuzzy evaluation results are converted into specific values ​​through defuzzification operations to obtain the state evaluation values ​​of the second layer. The state evaluation values ​​of the second layer are graded and classified. Different state evaluation levels are established according to the health status and operating conditions of the battery. These levels reflect the current health level of the battery and divide the working state of the battery in combination with the battery's use environment and working conditions (such as temperature, load, etc.). Through the hierarchical evaluation, the battery state is clearly divided into multiple levels for subsequent management and optimization. The hierarchical evaluation results are optimized for energy management. By combining the battery state of charge threshold, a multi-objective optimization function is constructed to optimize the battery charging and discharging process, so that the battery can achieve the best energy management and utilization efficiency under different working conditions. The optimization function considers multiple objectives, such as maximizing battery usage time, reducing energy consumption, and extending battery life, and obtains multi-dimensional optimization results through the calculation of the objective function matrix. On this basis, the optimal control sequence is calculated through dynamic programming operations, and the state transfer equation is used to predict the future state of the battery to obtain the power optimization parameters. After obtaining the power optimization parameters, the energy recovery control stage is entered. Combined with the current state of the battery, the regenerative braking adjustment coefficient is determined. The regenerative braking adjustment coefficient is used to recover energy through regenerative braking during the battery charging and discharging process and effectively utilize the remaining energy of the battery. This process can not only improve energy utilization efficiency, but also reduce battery load and extend battery life. By controlling the regenerative braking process, the energy recovery process of the battery is optimized, so that the battery can maximize the recovery and use of electrical energy under different working conditions. The braking energy recovery control amount and power optimization parameters are coordinated and optimized. According to the set constraints, the battery's charge and discharge power is limited to ensure that the battery operates within a safe power range and obtain a safe power control amount.In this process, the charging and discharging power of the battery is strictly limited to avoid over-discharging or over-charging, thereby ensuring the safety and stability of the battery. Through collaborative optimization, the charging and discharging process of the battery is precisely regulated to ensure that the battery works in the optimal state and avoid damage to the battery. The safe power control quantity is input into the hierarchical controller, and the hierarchical hierarchical control operation is performed to obtain the battery state control parameters. Hierarchical hierarchical control is to achieve precise control of the battery state through the collaborative work of multiple control levels. In this control process, according to the current state of the battery and the optimization target, the control parameters are adjusted layer by layer to ensure that the operating state of the battery remains within the optimal range. Through the hierarchical control strategy, the system can respond to changes in the battery state in real time, adjust the control parameters, and ensure that the battery can achieve the best performance in different working environments.

[0038] In the embodiment of the present invention, by introducing a triple forgetting factor adaptive extended Kalman filter, accurate estimation of the battery state of the sightseeing car in a high noise environment is achieved, and a series PSO-SA algorithm is used to perform global and local two-stage optimization of the battery equivalent circuit model parameters, which effectively reduces the computational complexity of parameter identification while ensuring the accuracy of the model parameters; a three-layer hierarchical control strategy is designed to achieve all-round control from parameter monitoring to state evaluation to energy management, thereby improving the reliability and safety of the sightseeing car battery management system; fuzzy control is combined with dynamic programming to construct an energy recovery optimization control strategy, thereby improving the recovery rate of the braking energy of the sightseeing car and extending the battery life; a layered battery state monitoring architecture is adopted to achieve coordinated monitoring of multiple parameters such as battery voltage, current, and temperature, thereby improving the fault tolerance and reliability of the system.

[0039] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0040] The type information, rated capacity and rated voltage parameters of the sightseeing car battery are collected non-invasively in multiple stages to obtain the basic parameter data of the battery. The basic parameter data of the battery and the operating parameter range are classified, modeled and stored to obtain a benchmark data set.

[0041] The terminal voltage, current and temperature parameters of the sightseeing car battery are sampled in real time to obtain the original sampling data, and the original sampling data is digitally filtered to obtain the filtered data;

[0042] Perform signal-to-noise ratio analysis and outlier detection on the filtered data to obtain data labeling results, and perform data repair and compensation processing on the filtered data based on the data labeling results to obtain repaired data;

[0043] The repaired data is reconstructed according to the battery state parameter vector format to obtain reconstructed data, and the reconstructed data is time-aligned and data normalized to obtain a battery operation parameter data set.

[0044] Specifically, the basic information of the battery is collected, including the type information of the battery, the rated capacity of the battery, and the rated voltage parameters of the battery. A multi-stage non-invasive collection method is adopted to minimize interference with the normal operation of the battery and obtain accurate basic data at the same time. The basic parameters of the battery are read in real time and preliminarily processed through sensors or external devices. For example, the type information of the battery is obtained by scanning the battery label or the battery management system interface, and the rated capacity and rated voltage of the battery are obtained from the technical specification or production data of the battery. The basic parameter data of the battery and the operating parameter range of the battery are classified, modeled and stored to build a benchmark data set. The benchmark data set includes various working states and performance parameter ranges of the battery. During the operation of the battery, the terminal voltage, current, temperature and other parameters of the battery are collected in real time. These real-time sampling data can reflect the actual operating state of the battery during operation, including information such as the charging and discharging conditions of the battery and temperature fluctuations. The selection of the sampling frequency is determined according to the requirements of the battery management system and the specific application scenario. For example, the sampling frequency of the battery current is set to 10 times per second, while the sampling frequency of the temperature and terminal voltage is set to 1 time per second. The original sampling data is digitally filtered to remove noise and interference to obtain filtered data. The filtering algorithm includes a low-pass filter, a high-pass filter or a Kalman filter. For example, the low-pass filter formula is expressed as:

[0045] ;

[0046] in, is the filtered signal, is the original signal, It is a filtering factor that determines the weighted ratio of current data and historical data. Through this filtering method, the fluctuations in the original data are smoothed and the noise caused by external interference is reduced. The signal-to-noise ratio analysis and outlier detection are performed on the filtered data. The signal-to-noise ratio (SNR) is used to measure the strength of the signal relative to the noise. A high SNR indicates a good signal quality, while a low SNR means a high noise. In the battery monitoring system, SNR analysis is performed by calculating the ratio between the signal strength and the background noise at each sampling point. The SNR calculation formula is:

[0047] ;

[0048] in, represents the mean value of the signal, Represents the standard deviation of noise. Through SNR analysis, it is determined which data sampling values ​​have large noise interference, and then outliers are identified. Outlier detection is to identify data points that deviate from the normal range by setting a suitable threshold range, such as voltage values ​​higher than the set safety threshold or current changes beyond the expected range. Outlier detection uses statistical methods, such as calculating the standard deviation of the data to determine whether a sampling value deviates from the normal range. Repair and compensate the filtered data based on the data labeling results. The repair process includes removing outliers and filling missing values. Compensation processing is to adjust the battery parameters according to the model or rules. For example, if the temperature sensor fails, a reasonable temperature value is calculated based on the working state of the battery and the ambient temperature. The repaired data can better represent the actual working state of the battery and ensure the accuracy of subsequent analysis results. The repaired data is reconstructed in the battery state parameter vector format. The reconstructed data includes battery state parameters such as voltage, current, temperature, and related performance parameters calculated by mathematical modeling. Organize the original data into a structured data set for subsequent processing. The battery state parameter vector is represented as a column vector:

[0049] ;

[0050] in, is the battery voltage, is the battery current, is the battery temperature. Vectorization makes data analysis and modeling more convenient. The reconstructed data is time-aligned and normalized. Time alignment refers to synchronizing data from different sources along the time axis to ensure that the data at each time point corresponds to the same observation moment. Due to the different sampling frequencies of different parameters, the data is aligned through interpolation or time synchronization algorithms. Data normalization is to unify data of different dimensions into the same range, using the minimum-maximum normalization method:

[0051] ;

[0052] in, is the original data, and are the minimum and maximum values ​​of the data respectively. The normalized data is convenient for unified analysis in the subsequent processing stage. Through the above steps, the battery operation parameter data set is obtained.

[0053] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0054] Based on the battery operating parameter data set, the correlation analysis of the open circuit voltage source, internal series resistance and RC parallel network is carried out to obtain the topological structure of the battery equivalent circuit model;

[0055] The topological structure of the battery equivalent circuit model and the battery operation parameter data set are input into the particle swarm algorithm for parameter initialization to obtain the particle swarm parameter matrix, which includes the particle swarm scale, the maximum number of iterations, the learning factor and the inertia weight;

[0056] Each particle in the particle swarm parameter matrix is ​​evaluated according to the fitness function, and the global optimal particle position and the local optimal particle position are selected according to the extreme difference method to obtain the global search parameter set;

[0057] The global search parameter set is input into the simulated annealing algorithm for processing to obtain the initial annealing solution, and the state transition probability and local search are performed on the initial annealing solution through the Metropolis criterion to obtain the local optimal solution;

[0058] Based on the local optimization solution, the internal resistance parameters and polarization parameters of the battery equivalent circuit model are jointly solved to obtain the initial values ​​of the model parameters. The initial values ​​of the model parameters are input into the verification module, and a cross-validation operation is performed to calculate the root mean square error between the terminal voltage prediction value and the measured value, and the parameter combination with the smallest error is screened to obtain the optimal parameter group.

[0059] The series resistance value, RC network parameters and open circuit voltage characteristic curve in the optimal parameter group are numerically calibrated to obtain the optimal parameter set of the battery equivalent circuit model.

[0060] Specifically, during the operation of the battery, by real-time acquisition of parameters such as voltage, current and temperature, combined with basic information of the battery (such as rated voltage and capacity), mathematical modeling methods such as the least squares method are used to analyze the correlation between the open circuit voltage source, series resistance and RC parallel network of the battery. Assume that the open circuit voltage of the battery is , the series resistance is The resistance of the RC parallel network is and the capacitance is , then the equivalent circuit model of the battery is expressed as:

[0061] ;

[0062] in, is the battery terminal voltage, is the battery current. By analyzing these battery parameter data, the relationship between the topological structure and parameters of these circuit elements is revealed. The topological structure of the battery equivalent circuit model and the operating parameter data set of the battery are input into the particle swarm optimization algorithm to initialize the parameters of the battery equivalent circuit model. The particle swarm algorithm is a global optimization algorithm that finds the global optimal solution through information sharing and updating between particles. When initializing the particle swarm algorithm, some key parameters are set, including the particle swarm size, the maximum number of iterations, the learning factor, and the inertia weight. These parameters will affect the efficiency and accuracy of the particle swarm search. Assume that the particle swarm size is , the maximum number of iterations is , the learning factor is and , the inertia weight is , then the update formula of the particle swarm algorithm is:

[0063] ;

[0064] ;

[0065] in, It is The speed of a particle, It is The position of a particle, It is The best historical position of a particle, is the global optimal position of all particles, and is a random number, is the inertia weight, and is the learning factor. By initializing the particle swarm parameter matrix, good initial conditions are provided for the subsequent optimization process. After the particle swarm algorithm is initialized, the algorithm searches for the optimal solution. In order to evaluate the quality of each particle, a fitness function is defined, which measures the error between the predicted value of the battery equivalent circuit model and the actual measured value. The fitness function is defined by the root mean square error:

[0066] ;

[0067] in, is the battery terminal voltage predicted based on the battery equivalent circuit model, is the actual measured battery terminal voltage, is the number of samples. By evaluating the fitness of each particle, the particle swarm algorithm can select the global optimal particle position and the local optimal particle position to obtain the global search parameter set. The global search parameter set is input into the simulated annealing algorithm for processing. The simulated annealing algorithm is a local search optimization method that searches for the optimal solution by gradually reducing the temperature. The core of the simulated annealing algorithm is the Metropolis criterion, which determines whether to accept a new solution by calculating the probability of state transition. Assume that the energy of the current solution is , the energy of the new solution is , then the probability of state transition is:

[0068] ;

[0069] in, is the temperature, and as the iteration proceeds, the temperature gradually decreases. Through the Metropolis criterion, the simulated annealing algorithm further optimizes the solution based on the global search and obtains the local optimal solution. Based on the local optimal solution, the internal resistance parameters and polarization parameters in the battery equivalent circuit model are jointly solved. The internal resistance and polarization parameters are closely related to the temperature, current and discharge state of the battery, and these factors need to be considered when solving. Assume that the internal resistance is , the polarization parameter is and , the initial values ​​of the model parameters are obtained by joint solution. In order to verify the accuracy of the obtained model parameters, cross validation is performed. Cross validation is a model evaluation method that divides the data set into multiple subsets and uses different subsets as training sets and test sets in turn to avoid overfitting problems. In the cross validation process, the root mean square error between the predicted terminal voltage and the measured value is calculated, and the optimal parameter group is obtained by screening the parameter combination with the smallest error. The series resistance value, RC network parameters and open circuit voltage characteristic curve in the optimal parameter group are numerically calibrated. The parameters are adjusted through experiments or model calculations to ensure that the model can better fit the actual working state of the battery. For example, the series resistance value is obtained through battery discharge experiments, and the parameters of the RC network are inferred from the voltage changes during the battery charging and discharging process. Through this calibration process, the optimal parameter set of the battery equivalent circuit model is obtained.

[0070] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0071] The open circuit voltage equation is linearized based on the optimal parameter set of the battery equivalent circuit model, and a state equation including the battery state of charge and polarization voltage is constructed to obtain a state space model;

[0072] Perform Taylor expansion on the state space model, construct an observation equation including the battery terminal voltage, and obtain a measurement space model;

[0073] The state space model and the measurement space model are input into the triple forgetting factor adaptive extended Kalman filter to obtain the initial value of the first forgetting factor, and the process noise covariance matrix of the triple forgetting factor adaptive extended Kalman filter is recursively calculated to obtain the initial value of the second forgetting factor;

[0074] Dynamically updating the measurement noise covariance matrix based on the first forgetting factor initial value and the second forgetting factor initial value to obtain a third forgetting factor initial value, and updating the covariance matrix based on the first forgetting factor initial value, the second forgetting factor initial value and the third forgetting factor initial value, performing a matrix multiplication operation to obtain an initial value of the covariance matrix;

[0075] Perform singular value decomposition on the initial value of the covariance matrix, and calculate the Jacobian matrix of the state variables by the least squares method to obtain the state transfer matrix;

[0076] The triple forgetting factor adaptive extended Kalman filter is initialized and configured based on the initial values ​​of the state transfer matrix and the covariance matrix to obtain the battery state observation model.

[0077] Specifically, based on the optimal parameter set of the battery equivalent circuit model, the open circuit voltage equation is obtained. The open circuit voltage of the battery ( It refers to the voltage of the battery under no-load conditions, which is affected by factors such as the battery's state of charge and temperature. In order to introduce the open circuit voltage equation into the state space model, it is linearized. Assume that the relationship between the battery's open circuit voltage and state of charge (SOC) is expressed as:

[0078] ;

[0079] in, is the battery temperature, is the state of charge of the battery. In order to facilitate the subsequent filtering operation, the equation is linearized. Taylor expansion is performed near the operating point to obtain the approximate linear form of the open circuit voltage equation:

[0080] ;

[0081] in, and is the initial value of the operating point, and are the partial derivatives of the open circuit voltage equation with respect to the state of charge and temperature, respectively. The state equation of the battery is constructed by the influence of the current and internal resistance in the battery equivalent circuit model. The state equation of the battery contains the state of charge (SOC) and the polarization voltage ( ). The polarization voltage represents the voltage drop caused by the mismatch of reaction rates inside the battery. Through the construction of the state space model, the state equation of the battery is expressed as:

[0082] ;

[0083] ;

[0084] in, is the capacity of the battery, is the polarization resistance of the battery, is the polarization capacitance of the battery, is the battery current. The state equation describes the dynamic characteristics of the battery's state of charge change and polarization voltage. The state equation is transformed into a form more suitable for Kalman filter processing by Taylor expansion. In order to construct the observation equation, the terminal voltage ( ) is used for modeling. Assuming that the terminal voltage of the battery is determined by the open circuit voltage, series resistance, polarization voltage and current, it can be written as:

[0085] ;

[0086] By performing Taylor expansion on the state space model, the observation equation for the battery terminal voltage is obtained:

[0087] ;

[0088] The form of this observation equation is suitable for inputting the extended Kalman filter for state estimation. Combined with the triple forgetting factor adaptive extended Kalman filter (TEKF), the covariance matrix is ​​dynamically updated. TEKF introduces three forgetting factors, which are used to dynamically adjust the process noise covariance matrix and the measurement noise covariance matrix, thereby improving the filtering accuracy and convergence speed. Given the initial covariance matrix , according to the first forgetting factor ( ) and the second forgetting factor ( ) dynamically updates the process noise and measurement noise. Assume that the process noise covariance matrix is , the measurement noise covariance matrix is , then the initial value of the first forgetting factor is calculated by the following formula:

[0089] ;

[0090] Similarly, the second forgetting factor is used to update the measurement noise covariance matrix :

[0091] ;

[0092] On this basis, the covariance matrix is ​​updated according to the dynamically updated noise covariance matrix to obtain the initial value of the covariance matrix In order to optimize the calculation of the covariance matrix, the singular value decomposition method is used to decompose the initial covariance matrix. Assume that the initial covariance matrix is , then its singular value decomposition is expressed as:

[0093] ;

[0094] in, and are the left singular matrix and the right singular matrix respectively, is a diagonal matrix containing the singular values ​​of the covariance matrix. The Jacobian matrix of the state variables is calculated by the least squares method to obtain the state transfer matrix The state transition matrix describes the system state from time To time The evolution relationship between the state transfer matrix and the initial value of the covariance matrix Combined with the triple forgetting factor adaptive extended Kalman filter, the configuration is initialized. Through these steps, a stable and accurate battery state observation model is obtained, which can effectively estimate the battery state of charge (SOC) and polarization voltage ( ).

[0095] In a specific embodiment, the execution step dynamically updates the measurement noise covariance matrix based on the first forgetting factor initial value and the second forgetting factor initial value to obtain the third forgetting factor initial value, and updates the covariance matrix based on the first forgetting factor initial value, the second forgetting factor initial value and the third forgetting factor initial value, and performs a matrix multiplication operation to obtain the initial value of the covariance matrix. The process may specifically include the following steps:

[0096] Performing a weighted combination operation on the first forgetting factor initial value and the second forgetting factor initial value to obtain a forgetting factor combination value, and constructing a diagonal weight matrix based on the forgetting factor combination value to obtain a weight matrix;

[0097] Recursively operate the weight matrix and the measurement noise covariance matrix to obtain the initial value of the third forgetting factor, and normalize the initial value of the first forgetting factor, the initial value of the second forgetting factor, and the initial value of the third forgetting factor to obtain a normalized forgetting factor matrix;

[0098] The covariance of the normalized forgetting factor matrix is ​​updated, and the state covariance matrix is ​​constructed by the matrix block method to obtain the block covariance matrix. The condition number of the block covariance matrix is ​​optimized, and the matrix eigenvalues ​​are adjusted by the regularization method to obtain the optimized covariance matrix.

[0099] The optimized covariance matrix is ​​subjected to matrix multiplication operation, and the covariance propagation calculation is performed to obtain the covariance prediction matrix. The covariance prediction matrix is ​​symmetry corrected and positive definiteness verified to obtain the initial value of the covariance matrix.

[0100] Specifically, for the initial value of the first forgetting factor ( ) and the initial value of the second forgetting factor ( ) are weighted combined to obtain a comprehensive forgetting factor. The weighted combination formula is expressed as:

[0101] ;

[0102] in, and are the weights of the first and second forgetting factors, respectively, satisfying , where the weight parameters are determined by experiments or optimization methods. By weighting, the comprehensive forgetting factor value is obtained . Based on the combined values ​​of the forgetting factors, a diagonal weight matrix is ​​constructed The diagonal elements of the matrix are the values ​​of the forgetting factor combination, and the other positions are zero. The constructed weight matrix is ​​expressed as:

[0103] ;

[0104] Each diagonal element of the weight matrix is ​​assigned the same value , used to adjust the contribution of measurement noise. Using the weight matrix and the measurement noise covariance matrix ( ) is recursively calculated. The recursive process involves combining the weight matrix and the measurement noise covariance matrix, and adjusting the update of the measurement noise covariance matrix through the formula. The recursive process is expressed as:

[0105] ;

[0106] in, is the measurement noise covariance matrix, and is a matrix Through this recursive process, the measurement noise covariance matrix is ​​updated to adapt to the new forgetting factor combination. On this basis, the initial value of the first forgetting factor ( )、The initial value of the second forgetting factor( ) and the initial value of the third forgetting factor ( ) to ensure that the value of the forgetting factor does not deviate too much from the normal range, thereby avoiding excessive parameter fluctuations during the filtering process. The normalization formula is as follows:

[0107] ;

[0108] ;

[0109] ;

[0110] The normalization operation scales the three forgetting factors according to their proportions so that their sum is 1. The state covariance matrix is ​​constructed by the matrix block method. The large matrix is ​​divided into multiple small blocks, which are calculated and updated separately. For the state covariance matrix, it is divided into several sub-matrices, each of which represents a state variable of the system or a different noise source. Assume that the state covariance matrix It is divided into the following forms:

[0111] ;

[0112] in, and is the main diagonal part of the covariance matrix, representing the covariance between the system states, and and It represents the covariance between the system state and the noise source. On this basis, the condition number of the submatrix is ​​optimized, and the stability of the matrix is ​​ensured by adjusting the eigenvalues ​​of the matrix. The goal of condition number optimization is to make the eigenvalue distribution of the matrix more uniform, thereby improving the convergence and computational stability of the filter. The optimized covariance matrix is ​​adjusted by the regularization method. The regularization method adjusts the diagonal elements of the matrix so that the range of the eigenvalue is within a reasonable range, avoiding abnormal values ​​or excessive values, thereby ensuring the stability of the filtering process. Assume that the regularization factor is , the covariance matrix is ​​regularized as follows:

[0113] ;

[0114] in, is the identity matrix, is a regularization factor used to adjust the eigenvalues ​​of the covariance matrix to ensure that it meets the stability requirements. The optimized covariance matrix is ​​subjected to matrix multiplication and covariance propagation calculation to obtain the covariance prediction matrix. The covariance propagation process is a key step in Kalman filtering, which is used to predict the state and covariance at the next moment based on the current estimated state and covariance. The formula for covariance propagation is expressed as:

[0115] ;

[0116] in, is the state transition matrix, is the covariance matrix at the current moment, is the process noise covariance matrix, is the covariance prediction matrix of the next moment. Symmetry correction and positive definiteness verification are performed on the covariance prediction matrix. Symmetry correction is used to ensure the symmetry of the matrix, while positive definiteness verification is used to ensure that the eigenvalues ​​of the matrix are all positive numbers, thereby ensuring the physical meaning of the covariance matrix. Through these processes, the final initial value of the covariance matrix is ​​obtained.

[0117] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0118] The battery operation parameter data set is input into the battery state observation model, and the battery state of charge is initialized and predicted using the state space model to obtain the state prediction matrix;

[0119] Based on the triple forgetting factor in the battery state observation model, the state prediction matrix is ​​updated in time to obtain the prior state estimation value;

[0120] The prior state estimation value is input into the measurement update unit of the battery state observation model, and the observation prediction is performed in combination with the measurement space model to obtain the measurement prediction value;

[0121] The difference between the measured predicted value and the actual observed value is calculated, and the state variable is corrected through the gain matrix of the battery state observation model to obtain the posterior state estimate;

[0122] The posterior state estimation value is input into the covariance update unit of the battery state observation model, and the error covariance matrix is ​​dynamically adjusted based on the triple forgetting factor to obtain the updated covariance matrix;

[0123] The updated covariance matrix is ​​numerically optimized for stability and corrected by the Joseph stabilization algorithm of the battery state observation model to obtain the optimized state quantity;

[0124] The optimized state quantity is input into the feature extraction unit of the battery state observation model, and the battery state of charge and polarization voltage characteristics are extracted through state space reconstruction to obtain a state feature sequence;

[0125] The state feature sequence is analyzed in multiple dimensions, and the feature fusion is performed in combination with the evaluation criteria of the battery state observation model to obtain the battery state feature quantity.

[0126] Specifically, the battery operating parameter data set is input into the battery state observation model, and the battery state of charge is initialized and predicted through the state space model. The initial state of charge of the battery is estimated through the operating parameters such as battery voltage, current, and temperature. Assume that the battery state equation is:

[0127] ;

[0128] in, is the state vector of the battery (such as state of charge SOC and polarization voltage), is the state transition matrix, is the control input (e.g. current), is the input matrix, is the process noise. Through this equation, the battery state is initialized and predicted according to the battery operating parameters to obtain the preliminary state prediction matrix The state prediction matrix is ​​updated in time based on the triple forgetting factor in the battery state observation model to obtain the prior state estimate. The triple forgetting factor is an adaptive mechanism for dynamically adjusting the filter process noise and measurement noise to enhance the robustness of the filter. In this process, the state prediction matrix is ​​updated by the time update formula, which is:

[0129] ;

[0130] in, is the prior state estimate, indicating that The battery state estimate is obtained based on the state prediction at the previous moment. Through time update, the battery state is adaptively predicted to handle the situation where the battery state changes over time. The prior state estimate is input into the measurement update unit of the battery state observation model, and the observation prediction is performed in combination with the measurement space model to obtain the measurement prediction value. The measurement space model is similar to the state space model and is used to describe how to generate actual observations from the state of the system. Assume that the observation equation of the battery is:

[0131] ;

[0132] in, is the observation vector (e.g. battery terminal voltage, etc.), is the measurement matrix, is the measurement noise. Through this equation, according to the prior state estimate and measurement matrix Perform measurement prediction and obtain the measurement prediction value . The difference between the measured predicted value and the actual observed value is calculated to obtain the innovation value , which is the difference between the measured predicted value and the actual observed value. The innovation amount is calculated by the following formula:

[0133] ;

[0134] Using the gain matrix of the battery state observation model Correct the state variables to obtain the posterior state estimate. The gain matrix is ​​calculated based on the prior covariance matrix and the measurement covariance matrix. , determines the influence of measurement update on state estimation. The calculation formula of the gain matrix is:

[0135] ;

[0136] in, is the prior covariance matrix, is the measurement noise covariance matrix. Through the gain matrix, the prior state estimate is corrected to obtain the posterior state estimate:

[0137] ;

[0138] Posterior state estimate It is the battery state estimate updated through measurement, which is more accurate than the prior state estimate. The a posteriori state estimate is input into the covariance update unit of the battery state observation model, and the error covariance matrix is ​​dynamically adjusted based on the triple forgetting factor to obtain the updated covariance matrix. The covariance matrix is ​​used to measure the uncertainty of the state estimate, which changes with each update. The covariance update formula is:

[0139] ;

[0140] in, is the updated covariance matrix, which represents the error range of the state estimate. As time goes by, the covariance matrix will gradually converge, indicating that the system's estimation accuracy of the state is constantly improving. The updated covariance matrix is ​​optimized for numerical stability and the covariance matrix is ​​corrected using the Joseph stabilization algorithm to ensure its numerical stability. The Joseph stabilization algorithm avoids numerical instability by calculating the positive definiteness and symmetry of the covariance matrix, ensuring that the covariance matrix does not cause error expansion in subsequent calculations. The stabilization algorithm is implemented as follows:

[0141] ;

[0142] Through this algorithm, the symmetry of the covariance matrix is ​​guaranteed, and its eigenvalues ​​tend to be balanced, ensuring numerical stability. The optimized state quantity is input into the feature extraction unit of the battery state observation model, and the battery state of charge and polarization voltage characteristics are extracted through state space reconstruction to obtain the state feature sequence. The feature extraction unit extracts key information of the battery, such as state of charge and polarization voltage, from the optimized state quantity. Extracting meaningful low-dimensional features from the high-dimensional state space can effectively describe the operating status of the battery. Assume that the state feature sequence is , obtained by reducing the dimension of the state quantity or other forms of feature transformation:

[0143] ;

[0144] in, Represents the mapping function of feature extraction. Perform multi-dimensional analysis on the state feature sequence, combine the evaluation criteria of the battery state observation model to fuse the features, and obtain the battery state feature quantity. By combining the information of multiple feature sources, a more comprehensive and accurate assessment of the battery state is performed. For example, by combining the battery's temperature, terminal voltage, state of charge and other information, feature fusion is performed through weighted averaging, principal component analysis and other methods to obtain the comprehensive feature quantity of the battery state. The fused battery state feature quantity is used for battery management and optimization control. The final battery state feature quantity It is expressed as:

[0145] ;

[0146] in, It is a mapping function for feature fusion, which combines multiple input features into a final state evaluation.

[0147] In a specific embodiment, the process of executing step S5 may specifically include the following steps:

[0148] Performing the first-level safety threshold monitoring operation on the voltage, current and temperature parameters in the battery status characteristic quantity to obtain a parameter monitoring matrix, and performing multi-threshold segmented judgment on the parameters in the parameter monitoring matrix to obtain the first-level parameter monitoring results;

[0149] Perform fuzzy rule evaluation on the first-layer parameter monitoring results, construct a fuzzy membership function through the combination conditions of the battery state parameters, obtain a state evaluation function matrix, and perform a defuzzification operation on the state evaluation function matrix to obtain a second-layer state evaluation value;

[0150] The second-layer status assessment values ​​are graded and classified, and a status assessment grade is established based on the battery health status and usage conditions to obtain a graded assessment result;

[0151] Energy management optimization is performed on the hierarchical evaluation results. A multi-objective optimization function is constructed in combination with the battery state of charge threshold to obtain the objective function matrix. Dynamic programming operations are performed on the objective function matrix. The optimal control sequence is calculated through the state transfer equation to obtain the power optimization parameters.

[0152] Perform energy recovery control on the power optimization parameters, determine the regenerative braking adjustment coefficient based on the current state of the battery, and obtain the braking energy recovery control amount;

[0153] The braking energy recovery control amount and power optimization parameters are collaboratively optimized, and the charging and discharging power is limited by constraints to obtain the safe power control amount. The safe power control amount is input into the hierarchical controller, and the hierarchical hierarchical control operation is performed to obtain the battery state control parameters.

[0154] Specifically, the first-level safety threshold monitoring operation is performed on the parameters such as voltage, current and temperature in the battery state characteristic quantity to obtain the parameter monitoring matrix. Batteries face different working conditions under different working conditions, and a series of safety thresholds are set. These thresholds are set according to the technical specifications and operating requirements of the battery. Assume that the battery voltage , current ,temperature The safety thresholds are The following conditional judgments are performed on these parameters to ensure that they do not exceed the safe range. The judgment rules are as follows:

[0155] ;

[0156] ;

[0157] ;

[0158] The judgment results of these judgment rules form a parameter monitoring matrix , where the monitored value of each parameter is ,and A value of 1 indicates that the parameter is within the safe range, and a value of 0 indicates that the parameter is beyond the safe range. These judgment results are used as the first-level parameter monitoring results. Fuzzy rule evaluation is performed on the first-level parameter monitoring results. Uncertain and fuzzy information is converted into more actionable control signals. The fuzzy membership function is constructed through the combination conditions of the battery status parameters. For example, the combination of voltage, current, and temperature forms a three-dimensional fuzzy space in which the battery status is evaluated as "normal", "warning", or "fault". Fuzzy membership function The likelihood of each state is quantified as follows:

[0159] ;

[0160] The current is constructed by similar fuzzy membership function and temperature The membership function of the state evaluation function matrix is ​​then evaluated based on these membership functions. The state evaluation function matrix is ​​input into the defuzzification operation (for example, the central law or weighted average method) to obtain the second-level state evaluation value The defuzzification process is achieved through the following formula:

[0161] ;

[0162] in is the weight of the membership function, is the membership degree of each state. The second-level state evaluation value is graded and classified. According to the health status and usage conditions of the battery, multiple battery state evaluation levels are established, such as "good", "normal", "warning" and "fault". In this grading and classification process, according to the current evaluation value of the battery The battery status is determined by comparing it with the preset evaluation level threshold. If the battery exceeds a certain threshold, the battery is judged to be in a "warning" state, otherwise it is in a "normal" state. This process is implemented by the following formula: , the battery is judged to be in "warning" state. After the state evaluation results are generated, energy management optimization is performed. The battery state of charge (SOC) is the core of energy management. Based on the battery state of charge threshold, a multi-objective optimization function is constructed to optimize the battery's operating performance under different working conditions. Assume that the battery state of charge range is , construct an objective function To maximize energy recovery efficiency and minimize battery loss, while ensuring that the battery operates within a safe operating range:

[0163] ;

[0164] in, and is the weight factor, is the battery state of charge, is the charging power, is the discharge power. Through this multi-objective optimization function, the optimal balance is found among multiple objectives, and the power output of the battery is optimized. Based on the optimization results, the power of the battery is adjusted to enter the energy recovery control stage. The energy recovery of the battery is optimized through the regenerative braking adjustment coefficient. Regenerative braking adjustment coefficient Determined by the current state of the battery, it indicates how much energy is recovered into the battery through the regenerative braking of the electric motor. The regenerative braking control amount is calculated by the following formula:

[0165] ;

[0166] in It is a regenerative braking adjustment function based on the battery state of charge. When the battery state of charge is low, the regenerative braking amount is increased to ensure rapid battery energy recovery. The braking energy recovery control amount and power optimization parameters are coordinated and optimized. In order to ensure the safety and stability of the battery system, the charging and discharging power is limited. By defining the power limit constraint and , dynamically adjust the charging and discharging power:

[0167] ;

[0168] Through these constraints, we can ensure that the battery is always within the safe power range during operation and obtain the safe power control value. The obtained safe power control quantity is input into the hierarchical controller for hierarchical hierarchical control operation. The hierarchical controller transmits the control decision of the battery from global to local level by level through multi-level feedback control strategy, thereby realizing accurate control of the battery state. The output of hierarchical control is the final state control parameter of the battery.

[0169] The above describes the method for monitoring the battery status of a sightseeing vehicle in an embodiment of the present invention. The following describes the device for monitoring the battery status of a sightseeing vehicle in an embodiment of the present invention. Figure 2 In one embodiment of the present invention, a battery status monitoring device for a sightseeing vehicle includes:

[0170] The acquisition module is used to collect the voltage, current and temperature data of the sightseeing car battery and perform preprocessing to obtain the battery operation parameter data set;

[0171] A construction module is used to construct a battery equivalent circuit model based on a battery operating parameter data set, and perform global and local two-stage optimization to obtain an optimal parameter set for the battery equivalent circuit model;

[0172] A setting module is used to construct a state equation and an observation equation based on an optimal parameter set, initialize the covariance matrix of a triple forgetting factor adaptive extended Kalman filter, and obtain a battery state observation model;

[0173] The estimation module is used to input the battery operation parameter data set into the battery state observation model, obtain the prior state estimation through time update operation and the posterior state estimation through measurement update operation, perform hierarchical estimation on the battery state of charge and polarization voltage, and obtain the battery state characteristic quantity;

[0174] The output module is used to perform three-level hierarchical control based on the battery state characteristic quantity and output the battery state control parameters.

[0175] Through the coordinated cooperation of the above components and the introduction of a triple forgetting factor adaptive extended Kalman filter, the accurate estimation of the battery status of the sightseeing vehicle in a high noise environment is achieved. The series PSO-SA algorithm is used to perform global and local two-stage optimization of the battery equivalent circuit model parameters, which effectively reduces the computational complexity of parameter identification while ensuring the accuracy of the model parameters. A three-layer hierarchical control strategy is designed to achieve all-round control from parameter monitoring to status evaluation to energy management, thereby improving the reliability and safety of the sightseeing vehicle battery management system. Fuzzy control is combined with dynamic programming to construct an energy recovery optimization control strategy, which improves the recovery rate of the braking energy of the sightseeing vehicle and extends the battery life. A layered battery status monitoring architecture is adopted to achieve coordinated monitoring of multiple parameters such as battery voltage, current, and temperature, thereby improving the fault tolerance and reliability of the system.

[0176] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0177] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0178] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0179] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0180] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0181] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0182] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for monitoring the battery status of a sightseeing vehicle, characterized in that: The method comprises: Collect the voltage, current and temperature data of the sightseeing car battery and pre-process it to obtain the battery operation parameter data set; Building a battery equivalent circuit model based on the battery operating parameter data set, and performing global and local two-stage optimization to obtain an optimal parameter set for the battery equivalent circuit model; The state equation and the observation equation are constructed based on the optimal parameter set, and the covariance matrix of the triple forgetting factor adaptive extended Kalman filter is initialized to obtain a battery state observation model; specifically including: linearizing the open circuit voltage equation based on the optimal parameter set of the battery equivalent circuit model, constructing a state equation including the battery state of charge and polarization voltage, and obtaining a state space model; performing Taylor expansion operation on the state space model, constructing an observation equation including the battery terminal voltage, and obtaining a measurement space model; inputting the state space model and the measurement space model into the triple forgetting factor adaptive extended Kalman filter to obtain the first forgetting factor initial value, and performing the process of the triple forgetting factor adaptive extended Kalman filter. The noise covariance matrix is ​​recursively calculated to obtain a second initial value of the forgetting factor; the measurement noise covariance matrix is ​​dynamically updated based on the first initial value of the forgetting factor and the second initial value of the forgetting factor to obtain a third initial value of the forgetting factor, and the covariance matrix is ​​updated based on the first initial value of the forgetting factor, the second initial value of the forgetting factor and the third initial value of the forgetting factor, and a matrix multiplication operation is performed to obtain an initial value of the covariance matrix; the initial value of the covariance matrix is ​​singularly decomposed, and the Jacobian matrix of the state variable is calculated by the least squares method to obtain a state transfer matrix; the triple forgetting factor adaptive extended Kalman filter is initialized and configured based on the state transfer matrix and the initial value of the covariance matrix to obtain a battery state observation model; The battery operating parameter data set is input into the battery state observation model, a priori state estimation is obtained through time update operation and a posteriori state estimation is obtained through measurement update operation, and the battery state of charge and polarization voltage are estimated hierarchically to obtain battery state characteristic quantities; A three-level hierarchical control is performed based on the battery state feature quantity, and a battery state control parameter is output.

2. The sightseeing vehicle battery status monitoring method according to claim 1, characterized in that: The voltage, current and temperature data of the sightseeing car battery are collected and preprocessed to obtain a battery operation parameter data set, including: Perform multi-stage non-invasive collection of the type information, battery rated capacity, and rated voltage parameters of the sightseeing vehicle battery to obtain basic battery parameter data, and classify, model, and store the basic battery parameter data and operating parameter range to obtain a benchmark data set; Performing real-time sampling of terminal voltage, current, and temperature parameters of the sightseeing vehicle battery to obtain original sampling data, and performing digital filtering on the original sampling data to obtain filtered data; Performing signal-to-noise ratio analysis and outlier detection on the filtered data to obtain a data labeling result, and performing data repair and compensation processing on the filtered data based on the data labeling result to obtain repaired data; The repaired data is reconstructed according to a battery state parameter vector format to obtain reconstructed data, and the reconstructed data is time-series aligned and data normalized to obtain a battery operation parameter data set.

3. The method for monitoring the battery status of a sightseeing vehicle according to claim 2, characterized in that: The battery equivalent circuit model is constructed based on the battery operation parameter data set, and a global and local two-stage optimization is performed to obtain an optimal parameter set of the battery equivalent circuit model, including: Based on the battery operation parameter data set, a correlation analysis is performed on the open circuit voltage source, the internal series resistance and the RC parallel network to obtain a topological structure of the battery equivalent circuit model; Inputting the topological structure of the battery equivalent circuit model and the battery operation parameter data set into the particle swarm algorithm for parameter initialization to obtain a particle swarm parameter matrix, wherein the particle swarm parameter matrix includes a particle swarm scale, a maximum number of iterations, a learning factor, and an inertia weight; Each particle in the particle swarm parameter matrix is ​​evaluated according to the fitness function, and a global optimal particle position and a local optimal particle position are selected according to a range method to obtain a global search parameter set; Inputting the global search parameter set into a simulated annealing algorithm for processing to obtain an initial annealing solution, and performing state transition probability calculation and local search on the initial annealing solution through the Metropolis criterion to obtain a local optimization solution; Based on the local optimization solution, the internal resistance parameter and polarization parameter of the battery equivalent circuit model are jointly solved to obtain the initial value of the model parameter, and the initial value of the model parameter is input into the verification module, a cross-validation operation is performed and the root mean square error between the terminal voltage prediction value and the measured value is calculated, and the parameter combination with the smallest error is screened to obtain the optimal parameter group; The series resistance value, RC network parameters and open circuit voltage characteristic curve in the optimal parameter group are numerically calibrated to obtain an optimal parameter set of the battery equivalent circuit model.

4. The method for monitoring the battery status of a sightseeing vehicle according to claim 1, characterized in that: The method of dynamically updating the measurement noise covariance matrix based on the first forgetting factor initial value and the second forgetting factor initial value to obtain a third forgetting factor initial value, and updating the covariance matrix based on the first forgetting factor initial value, the second forgetting factor initial value and the third forgetting factor initial value, performing a matrix multiplication operation to obtain an initial value of the covariance matrix includes: Performing a weighted combination operation on the first forgetting factor initial value and the second forgetting factor initial value to obtain a forgetting factor combination value, and constructing a diagonal weight matrix based on the forgetting factor combination value to obtain a weight matrix; Recursively operate the weight matrix and the measurement noise covariance matrix to obtain a third forgetting factor initial value, and normalize the first forgetting factor initial value, the second forgetting factor initial value, and the third forgetting factor initial value to obtain a normalized forgetting factor matrix; Performing a covariance update on the normalized forgetting factor matrix, and constructing a state covariance matrix by a matrix block method to obtain a block covariance matrix, and performing condition number optimization on the block covariance matrix, and adjusting the matrix eigenvalues ​​by a regularization method to obtain an optimized covariance matrix; A matrix multiplication operation is performed on the optimized covariance matrix, a covariance propagation calculation is performed to obtain a covariance prediction matrix, and the covariance prediction matrix is ​​symmetry corrected and positive definiteness verified to obtain an initial value of the covariance matrix.

5. The method for monitoring the battery status of a sightseeing vehicle according to claim 4, characterized in that: The battery operating parameter data set is input into the battery state observation model, a priori state estimation is obtained through time update operation and a posteriori state estimation is obtained through measurement update operation, and the battery state of charge and polarization voltage are estimated hierarchically to obtain battery state characteristic quantities, including: Inputting the battery operation parameter data set into the battery state observation model, using the state space model to perform initialization prediction on the battery state of charge, and obtaining a state prediction matrix; Performing a time update operation on the state prediction matrix based on the triple forgetting factor in the battery state observation model to obtain a priori state estimation value; Inputting the prior state estimation value into the measurement update unit of the battery state observation model, performing observation prediction in combination with the measurement space model, and obtaining a measurement prediction value; Calculating the difference between the measured predicted value and the actual observed value, correcting the state variable through the gain matrix of the battery state observation model, and obtaining a posterior state estimation value; Inputting the posterior state estimation value into the covariance update unit of the battery state observation model, dynamically adjusting the error covariance matrix based on the triple forgetting factor, and obtaining an updated covariance matrix; Performing numerical stability optimization on the updated covariance matrix, and correcting it through the Joseph stabilization algorithm of the battery state observation model to obtain an optimized state quantity; Inputting the optimized state quantity into the feature extraction unit of the battery state observation model, extracting the battery state of charge and polarization voltage characteristics through state space reconstruction, and obtaining a state feature sequence; The state feature sequence is subjected to multi-dimensional analysis, and feature fusion is performed in combination with the evaluation criteria of the battery state observation model to obtain the battery state feature quantity.

6. The method for monitoring the battery status of a sightseeing vehicle according to claim 5, characterized in that: The performing three-level hierarchical control based on the battery state characteristic quantity and outputting the battery state control parameter comprises: Performing a first-layer safety threshold monitoring operation on the voltage, current and temperature parameters in the battery state characteristic quantity to obtain a parameter monitoring matrix, and performing multi-threshold segmented judgment on the parameters in the parameter monitoring matrix to obtain a first-layer parameter monitoring result; Performing fuzzy rule evaluation on the first-layer parameter monitoring results, constructing a fuzzy membership function through the combination conditions of the battery state parameters to obtain a state evaluation function matrix, and performing a defuzzification operation on the state evaluation function matrix to obtain a second-layer state evaluation value; The second-layer status evaluation value is graded and classified, and a status evaluation grade is established based on the battery health status and the usage condition to obtain a graded evaluation result; Performing energy management optimization on the hierarchical evaluation results, building a multi-objective optimization function in combination with the battery state of charge threshold, obtaining an objective function matrix, performing dynamic programming operations on the objective function matrix, calculating an optimal control sequence through a state transfer equation, and obtaining power optimization parameters; Performing energy recovery control on the power optimization parameters, determining a regenerative braking adjustment coefficient in combination with a current state of the battery, and obtaining a braking energy recovery control amount; The braking energy recovery control amount and the power optimization parameter are collaboratively optimized, the charging and discharging power is limited by constraint conditions to obtain a safe power control amount, and the safe power control amount is input into a hierarchical controller to perform a hierarchical hierarchical control operation to obtain a battery state control parameter.

7. A sightseeing car battery status monitoring device, characterized in that: Used to execute the sightseeing vehicle battery status monitoring method according to any one of claims 1 to 6, the sightseeing vehicle battery status monitoring device comprises: The acquisition module is used to collect the voltage, current and temperature data of the sightseeing car battery and perform preprocessing to obtain the battery operation parameter data set; A construction module, used to construct a battery equivalent circuit model based on the battery operation parameter data set, and perform global and local two-stage optimization to obtain an optimal parameter set of the battery equivalent circuit model; A setting module, used to construct a state equation and an observation equation based on the optimal parameter set, initialize the covariance matrix of the triple forgetting factor adaptive extended Kalman filter, and obtain a battery state observation model; An estimation module, used to input the battery operation parameter data set into the battery state observation model, obtain a priori state estimation through time update operation and a posteriori state estimation through measurement update operation, perform hierarchical estimation on the battery state of charge and polarization voltage, and obtain battery state characteristic quantities; The output module is used to perform three-level hierarchical control based on the battery state characteristic quantity and output battery state control parameters.

8. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and the processor implements the method for monitoring the battery status of a sightseeing vehicle according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the processor is enabled to execute the sightseeing vehicle battery status monitoring method as claimed in any one of claims 1 to 6.

Images